Vagrant Gautam's blog

The Platypus (à la Poe)

2020-05-05T00:00:00Z

Then this egg-laying mammalian moving my sad fancy into smiling, By the grave and stern decorum of the brown fedora it wore, "Though thy temple be capped and hidden, thou," I said, "art sure no craven, Sage, mysterious Platypus wandering from the Tasmanian shore— Tell me what thy lordly name is on the Night's Plutonian shore!" Per the platypus "Agent P."

pandas things that make me sigh

2021-02-03T00:00:00Z

>>> import pandas as pd

I have some poor unsuspecting raw data that's about to be panda'd. It looks like this:

common_name,scientific_name,rating
bogue,Boops bops,7.5
Tibetan blackbird,Turdus maximus,1234567891
large flying fox,Pteropus vampyrus,131719
western lowland gorilla,Gorilla gorilla gorilla,2
tiny sky-tyrant,,

>>> df = pd.read_csv('poor_unsuspecting_raw_data.csv')
>>> df.dropna(subset=['scientific_name'])
               common_name          scientific_name        rating
0                    bogue               Boops bops  7.500000e+00
1        Tibetan blackbird           Turdus maximus  1.234568e+09
2         large flying fox        Pteropus vampyrus  1.317190e+05
3  western lowland gorilla  Gorilla gorilla gorilla  2.000000e+00

Okay, seems reasonable enough. And yet, disappointingly, if I wanted to isolate for the specific data that's being dropped due to NaN values in that column:

>>> df.isna(subset=['scientific_name'])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: isna() got an unexpected keyword argument 'subset'

I end up having to do this instead.

>>> df[df.scientific_name.isna()]
       common_name scientific_name  rating
4  tiny sky-tyrant             NaN     NaN

I was somewhat relieved that at least isna() and notna() seem to have the same signature, but I'm disappointed that they don't implement (almost) everything that dropna() does, because it seems like most of it would be useful. Clearly I can accomplish what I want to do with the different syntax, but there's so much value in consistency. Why make me type a whole new command from scratch when I could just change drop > is and rerun my code? Sigh.

Also while adding the links above, I realized there isn't an obvious way to link to a specific version of the docs. I'm running pandas v1.2.1, so my apologies if you're a reader from the Future™ and everything I'm yelling about has been fixed. From here and now: sigh.

It bothers me that pandas is about the only tool I use where I have to constantly go back to the docs to remind myself of the subtle differences between methods. What's even more frustrating about this is that I'm not even using methods that are rare or poorly documented. All of these are very common and basic use cases for pandas, and pretty much all of them appear in beginner guides. Here's another one. This time my data looks slightly different.

crane,270
crow,93
coot,65
curlew,76
crossbill,28

>>> df = pd.read_csv('cnames.csv', header=None, names=['bird', 'wingspann'])
>>> df
        bird  wingspann
0      crane        270
1       crow         93
2       coot         65
3     curlew         76
4  crossbill         28

Right, okay, so far so good, except for the fact that my dastardly butterfly keyboard has added a superfluous n to my second column. Nevermind, I can rename that with the handy pandas rename method.

>>> df.rename(names={'wingspann': 'wingspan'})
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/vagrant/venv/lib/python3.7/site-packages/pandas/util/_decorators.py", line 312, in wrapper
    return func(*args, **kwargs)
TypeError: rename() got an unexpected keyword argument 'names'

Sigh.

In this case what works is actually columns, not names.

>>> df.rename(columns={'wingspann': 'wingspan'})
        bird  wingspan
0      crane       270
1       crow        93
2       coot        65
3     curlew        76
4  crossbill        28

You see my point about how this is a pain in the ass I shouldn't have to deal with as a developer? I can't even imagine dealing with this if I was someone with less computational experience who was just trying to wrangle their gosh darn data. The inconsistency in the API practically guarantees that the special cases of these commands are never going to become muscle memory.

This was my trouble with find for a long time, actually. grep and sed both require you to specify a path after the pattern; find starts to froth at the mouth if you try that. Eventually I just used find often enough that it got its own little corner of my brain and I could remember the syntax without trying. What helps in this case is that it's an entirely separate binary and project that does something different (or, well, different enough) from the other utilities I mentioned. You know what doesn't have that excuse? pandas.

Okay one last one and I'm calling it a day. I'm just cooking up my own, unoriginal data for this example (birds? what birds?)

>>> df = pd.DataFrame([[1, 2, '3'], [4, None, '6'], [7, 8, '9']], columns=['a', 'b', 'c'])
>>> df
   a    b  c
0  1  2.0  3
1  4  NaN  6
2  7  8.0  9

Since you saw both the raw data and this DataFrame-ified representation of it, we can pretty much assume that the first column's going to have an integer dtype, the third is (hopefully) a string (or rather an "object"), and the second one looks like it was parsed as floats even though I didn't intend for it to be. This was almost certainly because of my None in row 2.

Aside: I should probably sit down someday and puzzle out the differences in behaviour between Python's None, numpy's nan, and pandas' NA, but you know what? Today is not that day. I'm sighing just thinking about it.

This time I'm not even going to give you a story. I'm just going to show you some commands and sigh loudly.

>>> df.dtypes
a      int64
b    float64
c     object
dtype: object
>>> df.c.str.isnumeric() # yay! these are numeric strings
0    True
1    True
2    True
Name: c, dtype: bool

>>> df.index.dtype
dtype('int64')
>>> df.index.isnumeric()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError: 'RangeIndex' object has no attribute 'isnumeric'
>>> df.index.is_numeric() # see that? it's an underscore. I know.
True

>>> # How about trying this with a different thing that's an integer dtype
>>> df.a.is_numeric() # nope
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/vagrant/venv/lib/python3.7/site-packages/pandas/core/generic.py", line 5462, in __getattr__
    return object.__getattribute__(self, name)
AttributeError: 'Series' object has no attribute 'is_numeric'
>>> df.a.isnumeric() # still nope
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/vagrant/venv/lib/python3.7/site-packages/pandas/core/generic.py", line 5462, in __getattr__
    return object.__getattribute__(self, name)
AttributeError: 'Series' object has no attribute 'isnumeric'

Why provide convenience methods like is_numeric() and is_integer() for some but not all objects that pattern a certain way? And no, using dtype directly isn't equivalent because you aren't guaranteed to get int64 in every environment (sys.maxint in Python is what decides what you'll get, as I recall).

>>> df.isna()
       a      b      c
0  False  False  False
1  False   True  False
2  False  False  False
>>> df.isnull()
       a      b      c
0  False  False  False
1  False   True  False
2  False  False  False
>>> df.isempty() # of course this doesn't work
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/vagrant/venv/lib/python3.7/site-packages/pandas/core/generic.py", line 5462, in __getattr__
    return object.__getattribute__(self, name)
AttributeError: 'DataFrame' object has no attribute 'isempty'
>>> df.is_empty()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/vagrant/venv/lib/python3.7/site-packages/pandas/core/generic.py", line 5462, in __getattr__
    return object.__getattribute__(self, name)
AttributeError: 'DataFrame' object has no attribute 'is_empty'
>>> df.empty() # how could I possibly expect any semblance of consistency
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: 'bool' object is not callable
>>> df.empty
False

Siiiiiiiiiiighhhhhhh.

>>> df.b.isna()
0    False
1     True
2    False
Name: b, dtype: bool
>>> df.b.isunique()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/vagrant/venv/lib/python3.7/site-packages/pandas/core/generic.py", line 5462, in __getattr__
    return object.__getattribute__(self, name)
AttributeError: 'Series' object has no attribute 'isunique'
>>> df.b.is_unique()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: 'bool' object is not callable
>>> df.b.is_unique
True
>>> # At this point, I'm all out of sighs to give

Bonus fact: I didn't show you in this last snippet but unique() ALSO exists. 🙃

I wrote a little script that's my assumption of how they designed their API.

>>> from random import choice
>>> prefixes = ['is', 'is_', '']
>>> suffixes = ['', '()']
>>> content = ['numeric', 'unique', 'empty']
>>> for c in content:
...     print(choice(prefixes) + c + choice(suffixes))
...
is_numeric()
unique()
empty

Postscript

Despite appearances I do actually quite like pandas and their conceptual model makes a lot of sense to me (way more than R, anyway). And I know that these are problems that sometimes plague large, open-source projects with many contributors, but I sort of hope they do something about it, yanno? I'm sure this isn't the first post ever written griping about the inconsistencies in their API and workflows that were broken because pandas developers don't follow semantic versioning (I actually didn't even talk about that today). And, yes, I also know that it's possible my time would be better spent actually contributing to the project rather than complaining loudly into the void, so you really don't need to remind me :D Anyway, this whole thing was an attempt at exorcising my pandas demons and hopefully the act of writing it down means that some of it will stick!

A Doll's Story

2021-02-14T00:00:00Z

Drifting in the breeze, it is unmoored at sea, Light-headed, light, for the first time it's... free It was tossed around and cast aside per its owners' whims Untethered now in uncharted depths – let's see if it swims? An empty socket where an arm once was, twisted out in rage Scratches, strange marks taint its body; it's showing signs of age Its woolen hair is getting wet, its blue dress slowly darkens Its plastic body gets waterlogged as the abyss gently beckons No longer can it blow away, it's too heavy now to float Its lashes flutter up as it's pulled down, water up to its throat Submerged, terrified and alone, it can't move, it's drowning, fast! A numbness descends as it is smothered by a flood of memories past Its glassy eyes stare into the depths and schools of fish swim blithely by "Why didn't I leave?" she thinks to herself, "Why didn't I even try?" Anger roils inside her, joined by profound sadness, then fear It's a maelstrom of emotions and her vision starts to blur Tears fall as night falls and the moonlight makes the water gleam Thoughts rattle loudly in her head; she just wants to scream Currents tug at her dress, whisper comforting words to heal all: "How could you have left, or fought, or tried? You were only just a doll..." The castaway listens thoughtfully and the intensity starts to dissipate Held afloat by the kindly currents, she feels less of the weight Days, weeks, months go by, bringing the return of her customary joy She's healing herself and the strength she now has, no one can destroy She is her own person now, and her scars have a song to sing To remind her to never again be someone else's plaything.

Gender Repeal

2021-07-14T00:00:00Z

This piece originally appeared in Vancouver Pride's 2021 Pride Magazine.

Like many folks, I found an opportunity to confront my relationship with gender during the pandemic and engage in gender fuckery. It took nearly a year, but I finally came out as agender. I came out as pansexual five years ago, so I am already intimately familiar with the ongoing, lifelong process of coming out. Coming out, Part 2371. But who's counting?

"I will be deprecating my she/her pronouns," I said to my co-workers. We work in tech, so I have been using metaphors to explain this to my co-workers. People need to, "upgrade their systems to use xe/xem instead," I said, "or they will receive warnings or error messages when they use she/her pronouns."

I wish I had a way of explaining the same things to my grandmother in Chennai.

Even if I could get my grandmother to xe or they me in English, what options would I offer her for our other shared languages—Malayalam, Tamil, Hindi? As a linguist, I'm uniquely positioned to ask these questions and come up with answers, but I have felt isolated. I don't know how to bring up these questions in the predominately white, monolingual, English-speaking, North American queer spaces I'm a part of. It has felt isolating to know so few Indian queer folks to discuss this with. I don't have access to the kind of comprehensive resources about these things as I do in English. And I have been apprehensive about coming out in Indian spaces, never entirely sure how my queerness would be received, as queerness is sometimes perceived as Western or a white thing.

The pandemic changed this for me. Having in-person connections drastically limited made me value my online connections highly and now I have several Indian queer friends. Twitter is conducive to navel-gazing and an incredible way to find queer community and, especially, Indian-origin queer folks.

The conversations I've had with my Indian queer friends mean the world to me. Discussing options for gender-neutral pronouns and words for family members in Indian languages with people who get it is liberating. Especially when talking about nonbinary identity. There's a persistent and pernicious stereotype about nonbinary bodies being white, skinny, and androgynous. My body type is nothing like that, and a big part of that is my genes! Talking about this breed of dysphoria with folks who have bodies like mine makes me feel seen.

Ember, Moss, Steel. Another trope about nonbinary people is the tendency to adopt noun names. Part of the desire to conform to this stereotype is rooted in a desire to be seen as valid, accepted as legitimate. My name is Vasundhara, a Sanskrit noun that means earth—in one way, I conform to this stereotype. But Sanskrit is a language with grammatical gender and Vasundhara is feminine. Other Sanskrit options limit me in the same way. Yet as much as I want to discard my femininity, my name means so much more than that. Every time I make people pronounce my name properly, it is a gigantic Fuck You to colonialism. Giving up my name for an English noun would be a huge loss—a loss other Indian enbies understand.

The loss of community. Coming out to cishet Indian folks, first as bi/pan, now as agender, has produced a range of reactions from welcome and requests for more information, to mild apprehension, to subtle jabs and even outright disgust. "Your life will be easier if you choose not to be gay, or downplay it," is a common refrain heard by queer immigrants.

We also feel the loss of our queer history. Queer immigrant history is often not deemed worthy of preservation. I think a lot about what queer history means to me—not just in the context of not wanting to repeat the mistakes of our past, but also to understand where I come from, and to feel grounded by a connection to my ancestors and my history as a queer person.

Pride has been a time for me to contemplate my queerness. In 2019, I went on a queer history walk where I learned local queer history and was thrilled that our guide had taken care to pay attention to the lives of trans, Indigenous, and immigrant stories in particular. I think about that and wish there were more accessible opportunities to learn these stories.

Pride has also been a time for me to celebrate being queer! The year I was in the parade, I wore a giant bisexual pride flag like a cape and danced in the sun for three hours straight. A teen wearing a pansexual pride flag and watching the parade hugged me and they almost cried. I was hugged by folks in "Free Mom Hugs" t-shirts and saw older queers, gorgeous drag queens, pups in hoods, and my beautiful queer friends. It was a beautiful, happy day. I was so, so proud.

Coming out as agender has made me feel like something within me has settled. The last time I felt like this, I was desperate for a chance to get outside and tell everyone, "I am a pansexual woman!" Pride fulfilled that need. Now that I have figured out I'm not a woman, I'm agender, I want to shout it from the rooftops. And hey—yelling it from my roof is more pandemic-safe than hugging strangers.

Although Pride in its traditional form is not possible this year, I hope to hug my close friends and celebrate our flaming homosexuality in person, perhaps with a mini-parade! And I'll certainly do a number of virtual things—including the Pride Society's Queer History Panel, lots of Zoom dates with queer friends, and maybe even a gender repeal party in gather.town!

Vagrant is a queer, agender Indian Canadian (by way of Singapore) who is very gay and very tired. Xe gets excited about language, code, and birds. You can find xem on Twitter @DippedRusk or on the web at dippedrusk.com.

Guest Lecture in Pronouns

2021-07-14T00:00:00Z

This piece originally appeared as a guest post on Kirby Conrod's Medium blog, as part of Pronoun Studies, their series of blog posts on pronouns.

Disclaimer 1: This is a complicated thing for me to write because my feelings on this are still in flux and there's something about the act of putting pen to paper that feels like it's crystallizing those thoughts. I reserve the right to change my pronouns and the way in which I approach this conversation with people.

Disclaimer 2: It's also difficult for me to write about this because I've always found gender to be a difficult thing to think about and reason about and feel about. Gender is very hard to disentangle from society. It's not just an internal thing — it's inextricably intertwined with society, how I express myself to the world, how my performance is perceived. My thoughts on gender and pronouns are deeply connected with thoughts about my body, race, family, culture, and many other things that wouldn't fit in the disclaimer (or body) of a reasonably-sized Medium post. The delicate dance that strangers do where they size you up and treat you a certain way and assume things about you and pronoun you a certain way is pretty universal. That's something I could talk to other people about and feel seen and heard. But the way in which I experience gender internally was not always the same as the way I heard other people talk about it, and this part has required a lot of solitary reflection to figure out.

Disclaimer 3: I do not claim to speak for all neopronoun users. This is me putting down some thoughts about my pronouns and my feelings, and I know for a fact that there are people who feel very differently about their own gender and pronouns. All genderfeels are valid and should be respected, so don't weaponize my post against a different neopronouner whose use doesn't align with mine!

So what the heck are your pronouns anyway?

My pronouns are xe/xem/xyr/xyrs (pronounced with a z sound and rhymes with he/them/her/hers)! Here are some example sentences to show you how to use this set of pronouns:

Xe wrote a song about trilobites!
The only birder I know is xem.
Is xyr website accessible?
That tweet about a vampire linguist is xyrs.

Why are those your pronouns?

The short answer is that those are the pronouns that fit me best!

The longer story behind that is that I've been having genderfeels for years, and they peaked in the pandemic. My Twitter feed over the last year chronicles a lot of what I questioned about gender — to what extent I felt it, why it had to look or feel a certain way, and how I could experiment with it in ways that I hadn't before. Part of this involved cycling through a few different sets of pronouns over the last year. I went from she/her to she/they to she/they/xe to she/xe/they to she/xe to xe/she to xe/xem. Yes, I collect pronouns like some people collect shoes!

I spent months with different labels rattling around in my head to suss out what felt best, and after considering options like nonbinary, nonbinary woman, femmeby, genderfluid, etc., I finally settled on agender. Agender is a label that feels like home because it feels not like a gender that I experience that exists outside the binary, but rather the absence of gender altogether. My gender is a no thank you, goodbye, and fuck off.

Xe/xem pronouns felt like they fit well with that. Being xe-d affirmed my experience of gender (or its lack thereof) in a way that being she-d, he-d and they-d did not.

Discovering this was a slow process, of course, because it's rare to be in situations where you hear yourself being referred to in the third person. I used to try to simulate what it would feel like to be pronouned differently by constructing tweets where I could refer to myself in the third person and use different pronoun sets. One time when I did this in January, my friend Emma replied and they continued to talk about me in the third person with xe/xem pronouns, and I found myself reflecting on that interaction every few days for several months after that. About four months later I finally decided that if a set of pronouns could make me feel like that then maybe that was the set I should use!

Why don't you just use they/them?

Because being they-d never made me feel the way being xe-d does, even though it does feel better than being she-d. In fact, I've spent a bunch of time thinking about various other standard pronoun options and my feelings are:

she/her: 50% neutral, 30% bad, 10% good (gay she) — also depends on who's doing the she-ing and what the context of the conversation is
he/him: much less evidence and therefore I cannot percentage this. Depends on who's doing it and the context, but gay / genderbendy he-ing feels good!
they/them: 50% neutral, 30% surprising (who — me?), 20% good

Since they/them doesn't actively feel negative, I now list it as an alternative option, but since I'm doing this on a Medium post, let me unequivocally state for the record that no pronoun set has ever made me feel as good as xe/xem pronouns. So if you've read this, you are now obligated to use those for me!

I'm scared I'll mess up your neopronouns — any advice?

It's most important to me that you try your best to pronoun me correctly, even if you don't get everything 10000% right 10000% of the time. I also have a personal soft spot for second language English speakers and am much more tolerant of misgendering in this context. This is not to say that ESL speakers have a free pass to misgender me but rather that I GET that it's hard!

It's taken me LITERAL YEARS OF GENDERFEELS to get here — it's okay if you need a bit of an adjustment period. Back in September 2020 I confessed to regularly making mistakes with the xe/xem pronoun paradigm!

On a related note, when Demi Lovato recently came out as nonbinary, one of the things they said in conversation with Alok Vaid-Menon was that they sometimes misgender themself! It was very validating to hear them say that, because as a bb gay myself, I have often felt wary of making sweeping statements about even my own gender and pronouns without being able to prove my competence and internal consistency with them. Unlearning old habits and learning new ones takes time and conscious effort, so give yourself grace as you follow Kirby's guide to stop messing up pronouns.

I'm questioning whether neopronouns are right for me — any advice?

How exciting!!! I have four nuggets for you, and an open invitation to reach out to me on Twitter (@DippedRusk) if you want to chat more about neopronouns and genderfeels!

Find ways to hear/see yourself referred to with neopronouns

You could write about yourself in the third person like I did, or you could recruit a group of trusted friends to help you along on your gender journey by saying things about you in the third person with those neopronouns! If you're going to be introduced at an event, send in a bio with neopronouns and ask the organizers to introduce you using them! Find ways to experiment — it may open new doors for you!
Pay close attention to how different pronouns make you feel

For me this was the gamechanger with xe/xem pronouns. Maybe a neopronoun set makes you feel gender euphoria like no other pronouns, maybe it makes your body heave a sigh of relief, maybe it makes a new part of your brain itch in an interesting way. Observing how pronouns make you feel will help you figure out what fits best for you!
Be prepared to do a bunch of negotiation / explaining / fighting

Neopronouns are a new concept to many people and so you will likely have to explain how to use and pronounce your pronouns, and this may lead into discussions of gender and identity more broadly. I have found it useful to have a list of links handy (the Wikipedia page on nonbinary gender and this page on xe/xem pronouns are personal favourites).
Be prepared to do a bunch of negotiating — with yourself!

I've done a lot of internal negotiation to figure out my personal sweet spot for the level of neopronouning I am prepared to ask of people knowing that they will inevitably make mistakes, because the mistakes have emotional consequences for me. This balance can change over time and it may not look the same for you as it does for me. Just because people are going to misgender you does not mean you have to give them permission to. Whatever you choose for yourself is right for you and I for one will defend your pronouns (neo- or otherwise) to the death!

Your approach to dealing with misgendering may be fatalistic or it may be feisty. Your balance may be different for your oldest friends or for your immigrant family or for strangers you don't want to bother engaging with. A friend of mine recently brought up a consideration I hadn't thought of — e pointed out that some folks with aphasia struggle to acquire new words, and making them learn a new pronoun paradigm is a big ask, but that you also don't want to force people to out themselves. With this particular case, neither of us had an answer, and I don't think there is an easy or perfect answer. We just have to muddle through and figure out what works and be okay with making mistakes along the way!

Do you have any favourite resources for researching gender, neopronouns and queer identities?

My #1 favourite resource for learning about and discussing queer things is the other quiltbaggers I know! I was lucky to have been welcomed into a community of queer folks in Vancouver by a queer community organizer friend and I've also met many wonderful people on Twitter (including Kirby and other queer linguists studying queer linguistics). I have particular reverence for my queer elders and you should too!
I like this list of popular English pronouns which includes a good number of neopronouns
The LGTBA Wiki is a huge community-edited resource on all things queer and if you enjoy digging around among very granular labels then that might be a useful site for you
[Kirby adds: xe also sent me the link to this cool pronouns dressing room, which is a nice way to "try on" different names and pronouns!]

Whatever you do, make sure you are honouring yourself in the research you do! Your experience of gender and pronouns doesn't need a label, it definitely doesn't need a label immediately, and it may not align with anyone else's experience in the universe! And that's okay!

First post (of sorts)

2021-09-20T00:00:00Z

I have what I affectionately refer to as a blog graveyard due to my habit of starting blogs when I think of a clever name for them and then abandoning them during periods of feeling like what I'm writing isn't good enough or useful enough to be worth putting out there. Writing on its own has always felt quite therapeutic and I've proven to myself time and again that it's an incredibly effective way for me to consolidate my thoughts about things and to learn about myself. So this is my latest attempt at reforming how I think about blogging.

I've ported over some past posts from dead blogs (and I may add more) but this is the first one I'm officially writing on this blog and I want to use it to talk a bit about why I'm doing this at all and what makes this attempt different from all the previous ones. The history of my blog graveyard is relevant to this.

The one blog that has miraculously continued to escape abandonment for the last 8 or 9 years is tibackme.wordpress.com. Don't bother going to that link, though, because it's a private site. I've used it as an online journal of sorts, chronicling changes of location, relationships, mood, my beliefs and values and the things that generally matter to me.

One of the advantages of having this chronicle of my development is that it lets me go back in time -- which, incidentally, is why the blog is called tibackme - it's a rebus. When I look back I see that there are some things I grew out of easily but there are other questions of identity that I've been wrestling with for years. Increasingly, I feel like there is no end to these questions, that I'm just on a lifelong expedition to chip away at answering them, with little glimpses of light in reflective writing and in therapy sessions and conversations with friends.

I always thought that the secret to my private blog's long life was its privacy. I thought that privacy was what allowed me to take for granted that I was going to make mistakes and that I was going to change and grow. But it turns out that that isn't what it is. What my private blog writing shares with less private but extremely intimate moments of connection with other people is actually vulnerability.

I never felt like more public spaces online had room for vulnerability. This is why I ended up off Facebook and Instagram many years ago. It felt too dissonant and artificial for me. So I was very surprised to find so much vulnerability on Twitter when I initially started participating more. I've shared a lot of stuff on Twitter that I previously would have saved for my private blog and I've been amazed at how much community I have gained as a result of it, in this odd, digital way. And aside from community I have gained a lot in the way of understanding my own identity better by being able to talk to a wider group of people about the things I feel.

Over time my Twitter threads have grown longer, my DMs with people have gotten deeper, and I've had relationships that began on Twitter bloom in contexts off Twitter. It's been wonderful and it has encouraged me to share more, sometimes even in more formal contexts, about things that are deeply personal to me, about me. I've had people read my writing (especially about my queerness) and tell me that they feel seen in it. It's been very interesting to hear that as someone who felt at many points in my life like I was the only person who felt a certain way (about my sexuality, about my gender, about my perception of the world) - completely different from everyone else and importantly, from what I was supposed to feel.

All my previous attempts at reframing the way I thought about blogging were by trying to focus on the benefits of the writing and forget about the fears of it being public. This attempt differs from the previous ones in that the public aspect is absolutely central to it. After all, my Twitter isn't private at all, and I spend a lot of time there making mistakes, changing and growing, and other people (confoundingly) seem to want to be a part of that. What I want this blog to be is really a sort of extension of what Twitter has been for me -- a space to explore the things I'm exploring in longer form and in a medium that is still widely accessible for those who wish to engage with it vulnerably.

I'm excited I finally set it up and I'm looking forward to exorcising some writing demons around all the Feelings I'm having on the cusp of my latest big move (from Canada to Germany for a PhD). I am also still a little scared about blogging publicly and vulnerably, especially in a context so closely associated with my name and identity. I hope the worst that happens is just that no one reads what I have to say. But really I hope that people do read it and connect with it and want to talk to me about it and that together, we will save this blog from ending up in the graveyard with all the others.

50 things I want to do in my PhD

2021-09-22T00:00:00Z

Some of these are conditional on what COVID looks like and when I feel comfortable doing these activities again, but broadly, my intentions are:

Get decent at speaking and understanding German
Travel in Europe
Drive on the Autobahn
See 200 European bird species
Make non-academic local friends
Make trans and non-binary local friends
Build local queer community
Meet some cool nerds to nerd out with about CS and linguistics stuff
Do Advent of Code with some of them
Learn some new things in school
Do some cool projects (of any and all kinds)
Try to teach if possible
Learn how to write a paper
Learn what on earth "doing research" is supposed to look like
Do research
Pay attention to the authors you are reading and the authors you are citing
When you inevitably see problems with these lists, fix them
Learn what the following words mean, enough to theoretically work them into your dissertation if you later decide to - ontology, intertextuality, epistemology, monad
Hone your science communication skills and tell non-CS-PhDs about cool research accessibly (start a local Nerd Nite!?)
Don't die
Don't implode emotionally
Quit if the program feels like it's destroying your soul
Set good boundaries for yourself, for your people, about work and emotions and downtime and everything
Go to more opera, ballet, theatre performances
Find a local indie cinema like the ones you had in Vancouver
Write, please (and I mean writing for pleasure / learning / recordskeeping, not just academic writing)
Take care of your body
Take care of your mind (keep going to therapy)
Stay in touch with the folks who have helped you get here
Host as many non-local friends as you can within reason
Don't forget anyone in your acknowledgements, do not be that person, do not embarrass yourself 5 years from now
Speaking of not embarrassing yourself - look at a map, several times, and get so much better at knowing relative geography in Europe
Read some books to learn about German history
Continue to make people around you consider issues of race, gender, sexuality, ability, neurodivergence, etc.
But be careful you do it sustainably and don't burn out on it
No more DEI work for free for any for-profit institutions (this has been my rule in 2021)
Lift up and mentor your peers and more junior folks, both local and remote - especially folks whom you know to be systemically at a disadvantage in academia / tech / society
Be generous
Be gentle
With yourself too
No, really, please do not die or implode, and do literally whatever the fuck it takes to make that happen
Do more photography, make more music, make more art
Find a dark place somewhere in town where you can look up at the night sky and see stars
Forgive yourself for all the plants you are certainly going to kill
Figure out the shortest and/or cheapest path to the ocean
Have good sex - but not with undergrads or professors or really anyone else with whom there's a power imbalance (this is obvious enough to me that I feel it shouldn't need to be explicitly mentioned but some people are really weird!)
Ask for help - you WILL need it
If you haven't asked for help in a while, ask for help with something anyway
Find an advisor who has priorities similar to the ones on this list, and who will remind you and help you when you're slipping up on yours
Don't take this list too seriously - don't self-flagellate when you slip up

Notes from Germany, part 1

2021-10-02T00:00:00Z

Hallo aus Deutschland! My big move from Canada to Germany has happened, if you count my physical relocation as moving. There are still approximately 379 things I have to figure out before I can consider myself reasonably settled, but things are happening - slowly, steadily.

One of my superpowers is dealing with complex systems of constraints and dependencies and holding them all in my head (*when there are rules and the rules are clear and comprehensive). Thanks to German bureaucracy, this move has really been testing the limits of these abilities. I discovered pretty early on that there are some circular dependencies: banks won't let you open a local account without a local phone and a local address. You can't get even a pre-paid phone card without a local address. Worse still, many rental viewings can only be scheduled by phone and some require you to have a local bank account already (which, like I said, can't be done without a local address!). Circular dependencies make for literal unsolveable systems so I'm glad to have gotten very very lucky.

Before I got here, I had scheduled a viewing for an apartment which turned out to be even nicer in person than on the internet. So I signed a lease on my third day here! Here are some apartment highlights.

My place - the highlights

Adorable, bright little loft apartment that I have all to myself!
I made good on my promise to myself to get a queen-sized bed for the first time in my life
The pitter-patter of rain on my skylights is rather lovely
I can see birds flying over the house through said skylights!
3 buses to the university (about 10 minutes)
Within a 3 block radius
- at least 3 bakeries
- restaurants, cafes
- drugstores, pharmacies, a doctor
- supermarket, greengrocers
- town square (complete with a fountain, local pigeons, markets 2x a week)
- banks, phone companies
Slightly further away
- Pools
- Forest (birds!)
- Mini golf
- Tennis courts
- Train station

This place is at the high end of my budget but clearly it checked a lot of boxes for me and a big deciding factor was that the landlady was very kind and didn't have nearly as many hard requirements as other places did. When I came for the viewing, she even introduced me to half the neighbours and they all seemed like lovely people. She's been one of several instrumental people who've made a huge difference in the smoothness of my move here by being kind to me. Here's a quick section dedicated to them!

Helpful people

Landlady

I am most grateful to her for how much she has demystified about German life for me. She let me go shopping with her one day because I wanted to watch someone in their element navigate stores here, to learn how things worked. When I first moved to Canada I remember quickly learning that you didn't just go up to the coffee shop counter and say what you wanted. There's a script and it goes like this:

Cashier: "Hi, how are you today?" You: "I'm fine, how are you?" Cashier: "Good, thanks! What can I get for you?" You: place your order Cashier: "Is that everything?" You: order more things [or] "Nope, that's all." Cashier: "Your total is SomeNumber. How would you like to pay?" You: "Just on debit!" (this later changed to "just on VISA" when I got a credit card) Cashier: "Would you like your receipt?" You: "No thanks." Cashier: "Have a nice day!" You: "Thanks, you too!"

Deviating from this script would result in weird looks. And it's all very regional as well. I remember my "just on VISA" bit getting a confused eyebrow raise in Seattle even though it was universally accepted just 2 hours north in Vancouver! So this shopping trip with my landlady was very useful and I learned a simple German grocery/supermarket script that has been serving me well. It's less verbal here compared to Canada:

Cashier: "Hallo!" You: "Hallo!" Cashier: scans items and tells you your total You: either understand what they said or peek at the register, then hand them money Cashier: hands you your receipt and your change, telling you how much it is "Tschüss!" You: scramblingly put your change into your too-small pockets "Danke schön, tschüss!"

Painter

My landlady hired this painter to paint the common areas of the building. He was playing this really lovely Ethiopian music as he painted (which I figured out later when Shazam decided to cooperate with me), and I told him I liked it. When he was done painting, out of sheer kindness, he came upstairs to help me build my IKEA bed. This was an interesting challenge because he speaks less English than I do German, but we managed and I even learned some German in the process. The landlady had advised me to tip the IKEA delivery guys for bringing all the stuff up the stairs so I thought it would make sense to tip the painter too (especially since I knew she had hired him because he was out of work). But he refused, saying in German that he had lived here 7 years and I had only been here 4 days. I know this is what he said because among other things, I learned German numbers and days of the week from him while we were building the bed.

Phone shop person

This morning I was in said bed practising my German on Duolingo and my lying sideways pressed the power button on the phone and restarted it. When it came back to life my phone wanted me to enter a PIN for my SIM card - a number which was on a piece of plastic that I had thrown away before leaving the hotel. I couldn't call customer service because A, they would probably speak German, and B, my SIM card was locked! So I panickedly went to the phone store at the street corner and was relieved to find that the person there spoke English. I asked if he could help at all, even to just make the phone call for me. He very generously did so and saved my phone.

He was very excited to have someone to speak English with because it's rare here and so he promptly told me his entire life story. Apparently he lived in England for 6 years and had two romantic failures there after which he "hit rock bottom and did a lot of stupid things," but then he moved back here because he missed his family. He said he was Polish and that despite the history here in Germany, he had never faced as much xenophobia here as an Eastern European compared to what he faced in England. He had some adorable language quirks that I haven't heard in a long time such as referring to one of the two people he dated there as an "English lass," and saying "Sorry, love!" when he forgot to give me back my passport.

I signed up for a wifi connection at the store since I hadn't done that yet, and he walked me through the process and the sign-up. It's going to take 2+ weeks for a technician to come and set things up, but I'll take what I can get. More importantly though, I couldn't sign up without a local bank account (which I don't have yet), and you won't believe this but he actually put his own bank details in! He said I should just come into the store when I have my own bank account so he could update it. This way I could keep my technician slot instead of delaying it further just because I didn't have a local bank account. I'm quite floored by his (well-placed!) trust in me.

Old friends: Simon and Sabrina

Also I can't not mention my German friends Simon and Sabrina. They don't know each other personally but they both live in North America and have remotely been exceptionally sweet, kind angels to me. Here's a non-exhaustive list of stuff they have done for me:

explained stuff about the recent German election and the different political parties
listed and ranked German supermarkets
made me a literal Google doc with information about banking in Germany
tracked my flight
gave me advice and moral support when I was figuring out how to get from the airport to Saarbrücken
proofread German messages to Important People (like potential landlords)
told me "if you're totally lost and on the verge of a nervous breakdown, call me" (at 3 am for him)
sent me German memes without translation to make me do the work of learning
taught me never to translate "birding" literally in German because "vögeln" means "to fuck" (ich liebe vögeln, ha ha)

Birds

Speaking of birds, I haven't had a chance to get out and go birding yet because I only just unpacked my camera and binoculars and I've had my hands full with so many other things. But here's a quick list of new birds I have confidently IDed:

Eurasian moorhen
Grey heron
Eurasian magpie
Great tit
Carrion crow

I'm seeing WAY more birds than this short list (including several flying over the house at dusk) but I cannot reliably ID them yet, especially without a European birding book. That's on the neverending todo list but understandably it's lower priority than getting a bank account and wifi and a student card.

The lowlights

I worry I'm making it all sound very lovely and nice and easy when actually it's been (and continues to be) very hard, actually. And I don't just mean the cognitive load of constantly trying to communicate in a language you don't speak yet. Every time something goes wrong (plumbing issues! shorted my magic wand because of voltage differences! locked myself out of my SIM card! forgot to eat anything after breakfast because stress! misgendering! forgetting to hydrate enough! confusing all-German websites and customer service!), I feel like I inch closer to becoming a sobbing mess on the floor who just wants to go home, but I'm trying oh so hard to remind myself that it isn't me, that this is just what it's like to move to a new place and learn new things.

It's been helpful to remind myself of a previous move to an apartment in Port Moody (BC, Canada). In my first few days I couldn't stop setting off the smoke alarm no matter what I did, and it was incredibly loud and stressful and embarrassing because the landlords lived upstairs and could hear. After my first grocery run, I tried to put my eggs in the fridge and they slipped and I broke every last one. I had bought a new pepper grinder on my second day and when I tried to grind pepper over my stir fry (that I had set off the alarm with yet again, and that didn't have eggs in it because I'd broken them all), it turned out that the lid was loose and it emptied itself of half its peppercorns into my pan. I tried making a baked apple and the recipe asked me to pour water into the dish halfway through. It didn't specify that I should use warm water and I didn't know better, so I poured cool water into the hot glass oven dish and it shattered, sending shards of glass and cinammoney apple all over the kitchen. I felt incompetent and awful and like I couldn't do a single thing right.

And yet, none of that is what comes to mind when I think back to my time in that house. I remember trail running in the forest 6 or 7 days a week. I remember walking down the hill to my favourite cafe on the planet with a new book every Sunday, and spending a few hours there sipping on coffee and reading. I remember seeing my first pileated woodpecker in the woods, having folks over to visit me, taking mirror selfies in the bathroom, napping on the couch. In the end I was sad to leave.

I'm impatient to get to that point in this place too - I just have to keep reminding myself that it is necessarily a slow process. I hope I feel a bit closer by the time I write my next post. Until then, tschüss!

How Dialpad Moved Its Python AI Development from Pip to Poetry

2021-10-23T00:00:00Z

This piece originally appeared in The New Stack.

Vagrant Gautam was a Speech Recognition Engineer at Dialpad who recently left to begin a PhD in computer science at Saarland University, where xe will continue to feed xyr passions for code, language and building equitable speech recognition systems. In xyr free time, xe likes to sing, play piano and go birding.

Kevin James is a staff engineer and tech lead for the AI Engineering team at Dialpad. When not finding exciting new ways to (ab)use kubernetes in solving his team's problems, he enjoys experimenting in the kitchen, at the cocktail counter, or with an instrument.

It was a dark and stormy day indeed when, by a stroke of luck, our intrepid heroes stumbled across an entry in the pip changelog:

Maintainers are preparing to release pip 20.3, with the new resolver on by default.

Confusion and curiosity struck their hearts! They asked themselves: What is this new resolver? Will this affect us? How can I learn more? Did I leave the oven on? For Dialpad's AI teams, Python and its usage is central to the bulk of our development efforts: the wrong change at the wrong time could spell disaster.

After hunting down the arcane tomes and nigh-illegible scribbles describing the new behavior, the answer was clear: the new resolver had a laundry list of incompatibilities with our workflows. It would prevent us from deploying our new machine learning models, would cause our CI/CD platform to burn through hours of credits attempting in vain to reconcile incompatible truths of the universe, and would even prevent us from being able to build our development environments.

Something needed to be done. Some brave souls needed to discover a new way, a new style of Python package management for Dialpad, which could lead them into a brighter future — the future of package management that some feel is Poetry.

This is their story.

Why and How Did We Pick Poetry?

The journey towards filling Dialpad with more poetry begins, as do so many, with an RFC (Request for Comments). The AI teams at Dialpad use RFCs — originally named for the documents which encapsulate the specifications for the internet — to discuss proposals for changes to tools, workflows, and our own internal processes. These RFCs get debated on by any interested engineers until all the issues have been worked out and we've reached a consensus on the best way forward.

In this case, the RFC included a few major themes:

What is the problem we're trying to solve?

What are the particular issues caused by our current workflow with pip and how will they affect us? In our case, the changes were several but most notably:

the new resolver acted at installation time rather than development time (meaning we could not know if our dependencies would conflict until we'd already submitted code changes to be built), and;
It could not recalculate previous dependencies (so anywhere we attempted to add a new package as a later step in our build — which we often do to include our AI models as a blackbox on top of our platform code — might find a conflict where none exists)

Additionally, though not quite as important, there was no way to break from the pip process early: if a conflict was found, we wouldn't be able to find out until after pip had exhausted every potential choice, downloading all possibly compatible options in the matrix of potential dependency versions. Since we were not keen to spend our entire yearly budget on a week of added compute time and network usage, this wasn't a great option for us.

What options do we have?

The Python Packaging Authority maintains a list of all available tools in this space. On top of that, they even have a more trimmed-down set of suggested practices one could adopt. Unfortunately, these suggestions tend to be narrow or specific in scope and are mostly updated by the folks who build new tools — most naturally listing their own new tool as the suggestion. We wanted to determine what would fit best for our particular use cases and adopt a tool that matched us rather than match ourselves to a newly adopted tool. As such, we started from the list of all possible tools, scoured the internet for relevant blog posts or Hacker News articles, and eventually trimmed down our list to what appeared to be the most notable and relevant choices nowadays: pip (with several workarounds for our issues), Pipenv, pip-tools, buildout, flit, and Poetry.

Each had its pros and cons, handled a different subset of the lifecycle of the development of a Python package, and would in some way be able to solve the problems we faced. For each of them, we investigated how others had used them, what their development cycle looked like, and what overall goals they were meant to solve.

How do those options fit in our workflows?

On the artificial intelligence (AI) team, we like automated tooling: if we can avoid needing humans to focus on some repetitive task or worry about some solvable issue, that's a big win in our books. We use tools like Renovate to ensure we don't need to micromanage updates to our dependencies, pre-commit to run an ever-growing litany of linters, and CircleCI to automatically run those linters, tests, and everything involved with the continuous delivery of our code; we build applications and libraries (and scripts which act somewhere in between!), run our code on Knative and Kubernetes and some serverless platforms, write code in Python which can interface with our other code in C++, Elixir, Java, … needless to say, there were a whole bunch of boxes to check on this one.

Our ideal tool would be one that worked seamlessly with everything else in our system, functioned in all possible cases with no issues whatsoever, and folded our laundry while it was at it. A bit more realistically, a tool that would be easy for us to integrate and require little in the way of changes to our other setups or large OSS contributions to bridge the gap would be our best bet.

What's the plan for safely migrating to the new solution?

In this case, how could we avoid breaking absolutely everything or spending quarters porting over every individual system one at a time? Would any of these tools be easier or more difficult than the others to make work? Could the migrations be formulaic enough to script the changes rather than needing to write them all manually?

We're ardent believers in there being a safe path to migrate between any two possible systems, but some cases are certainly much easier than others. Ideally, we'd be able to transition from our previous workflow to our final choice without too much in the way of detours or temporary hacks.

In the end, converging towards Poetry as the final choice here was surprisingly frictionless — it would be a migration, yes, but not one we couldn't easily tackle. It would solve all the problems we faced as well as a few more, such as helping to make our virtual environment management a bit easier. All of the other alternatives came with their own larger set of gotchas and limitations, requiring changes to how we do unrelated things or adding yet more tools and complexities to our system.

Though Poetry attempts to solve multiple problems at once (package management, version constraints, virtual environments, et cetera), it does so in a way that lines up pretty nicely with the system we'd been cobbling together since we wrote our first line of code. We generally try to stick with "the right tool for the job" over "the one tool that does everything", but when the "one tool" also happens to do everything right, well, the decision was easy.

Once the RFC was accepted and a couple of those specifics were worked out, it was on to the actual changes.

Migrating to Poetry

Since the AI teams have a number of codebases spread across several teams with varying levels of complexity and importance to our users, it was important to carefully plan how to make our changes. Using some of our less mission-critical systems (such as internal tooling or hackathon-style research on new feature ideas) as a first pass to ensure we'd worked out any difficulties, we then prioritized production-facing code first and worked our way backward. It was most important to get these issues solved first for our most important systems in case our timeline estimate was wildly off and the pip-ageddon happened before we were ready.

In some cases, we made the decision not to upgrade code because the cost-to-benefit ratio was too low. For instance, on the data science teams we commit code from ad-hoc experiments, which sometimes include Jupyter notebooks and other code that isn't packaged and doesn't have unit tests. Packaging experimental code from ~~eons ago~~ 2018 and also manually testing it to ensure that it still worked just didn't seem worth it, especially when most of our experimental code has either not been used after the experiment or has been converted to production code in other repositories already.

Since we were systematically going through and modifying almost all of our code, we tried to fix other problems along the way; though we want to be careful in adding scope creep to an already complicated project, in simpler cases we think it's better to leave things you notice better than you found them. It ended up not being a 1-to-1 migration for this reason, but poetry made it easy enough to package code that it didn't extend the migration time at all. Some of the issues we fixed included:

Renaming badly named packages
Updating / adding package descriptions
Standardizing our Dockerfile conventions and documenting these standards
Fixing any folder name inconsistencies and clarifying best practices in the docs
Checking that dependencies were actually necessary and trimming those which weren't
Expanding version ranges of dependencies we could and exactly pinning those that did not use semantic versioning or any other useful schema (our legal team regrets to inform us that linking to a few popular packages here would not be a great idea)

We focused on solving problems only once and in a composable way. For example, rather than creating custom CircleCI workflows on each of our repositories, we created a reusable CircleCI orb with jobs to deploy arbitrary packages to a package index, run commands in a poetry environment, and so on. We could then use this orb in all of our repositories without having duplicated code.

Overall, switching our packages from pip to poetry was fairly simple and quick to do, but we wanted to distribute the work so everyone would have the opportunity for hands-on experience working with poetry. Education around this took a lot of time and effort, especially since we work with teammates who have varying backgrounds and levels of comfort with software. The bulk of the time spent on our migration ended up being intentional, to leave room for all our team members to learn the new system as we introduce it rather than absorb the changes over time.

Education, Documentation, Presentations

Our primary methods of education were through presentations and documentation.

The presentations were useful for getting the whole team on the same page about what was going on with pip, why we chose poetry, and what the migration process was going to look like. We had collaborative meetings where we went through a list of all our packages and assigned each of them to one of its code owners to handle its migration. By delegating the work out to the entire team in a balanced way, having completion milestones with quick turnaround, and clearly communicating the priority of this work, we were able to ensure the migration was smooth. In some cases, teams elected to have short development freezes for a particular repository, so they could put their heads down and focus on this one change without needing to worry about other ongoing priorities.

We also had to generate a large amount of documentation in a short time. For most of our internal documentation, we use Sphinx and have all our repositories configured to automatically build and publish our documentation from text files on every new change. That makes it easy to include or update our documentation with every change and to easily cross-reference information from multiple sources across multiple teams. It also makes it easy to review these docs or to help give each AI team an easy point of reference to understand what the others have accomplished; by using text files stored in git, reviewing documentation is exactly the same as reviewing code, which is something we all do regularly.

For the speech recognition team, Vagrant created a document outline on a branch that xe shared with the team immediately even though it was incomplete. The team was to start with upgrading packages before upgrading Dockerfiles and testing Docker images, so xe first filled out the section with step-by-step instructions for package upgrades. While the rest of the team worked on that, xe filled out the remaining sections; this parallel work allowed for the upgrade process not to be blocked by a lack of documentation. Another way in which we unblocked ourselves on updating the documentation was to keep this particular document on a git branch rather than pushing to main every time, as that would require reviews and potentially become a bottleneck.

Of course, documentation is never perfect on the first attempt and so based on requests for clarification that came up in the team Slack channel, Vagrant refined the instructions and merged in questions, suggestions, and thoughts from others. We ended up having to add an FAQ section on the document as well, because some questions and gotchas came up regularly and did not necessarily fit with the instructions.

Frequently (Asked and Answered) Questions

Some of what was in the FAQ section were answers already available in the official poetry docs or in the command line usage. For example, one question that came up often was how to delete a poetry virtual environment and start from scratch (the poetry env remove command). Another one that came up a few times was that people were unable to use the poetry command even though they had just pip-installed it: this was because they had missed the bit of output of the pip install command that said they had to add a folder to their PATH environment variable. When questions were asked more than once in a team channel or a private message, they were added to the FAQ section of the poetry migration guide and answered there, along with a link to the official place with the answers.

We also had some frequent questions from intrepid engineers who wanted to go beyond the minimum level of understanding required to complete the migration. For example, we recommended against using caret syntax when specifying dependencies and many people wanted more detail as to why we made that recommendation.

Let's take the hypothetical example of a package named aardvark. Let's say that I propose this change on your PR and you are confused because according to the official poetry documentation, they both mean the same thing.

-aardvark = "^1.2.0"
+aardvark = ">=1.2.0,<2.0.0"

Right now we are definitely good for versions 1.2.0 to 2.0.0 (not inclusive) because of semantic versioning. Let's say aardvark version 2.0.0 is released and we test it and everything works. Now we'd want to expand the upper end of the range, but there is no good reason to lose the lower end of the range — after all, we're not using any features exclusive to v1.3.0!

You simply can't do this with caret syntax. ^1.2.0 always only means >=1.2.0,<2.0.0. To get it to include 2.0.0 you would necessarily have to lose the 1.x.x versions — you'd have to change it to, say, ^2.0.0.

On the other hand, if you had >=1.2.0,<2.0.0, you could just change this to >=1.2.0,<3.0.0.

It's easier to read and reason about >=1.2.0,<2.0.0 than the caret syntax, it's consistent with all the other code, and therefore our documentation recommended the ">=,<" style in all examples of poetry add commands. Fortunately, this is the sort of thing easy enough to add to our linting rules and never need to worry about again, since Poetry's use of the TOML configuration language is very simple to write rules against.

Other FAQs were about gotchas relating to poetry use.

Poetry Gotchas

There was one common user experience gotcha related to configuring poetry with an authentication token for our private package index. When the configuration command was run, people either saw no output or they saw the following:

ModuleNotFoundError: No module named 'keyring.util.escape'
No suitable keyring backends were found
Using a plaintext file to store and retrieve credentials

Both options confused many people because there wasn't a clear indication of success. So we had to explain that both were actually successful cases — in one case, the credentials are stored in a system keyring and in the other case, where it fails to find one that's properly configured, poetry defaults to storing the credentials in a plaintext file.

Perhaps the biggest gotcha (and the one that required the hackiest fix) was poetry's behavior within Docker images and with virtual environments disabled. If you've installed poetry using pip, poetry's behavior of having poetry install –no-dev uninstall any unmanaged packages means that poetry actually uninstalls itself and all of its dependencies, which are sometimes required by other packages. Adding poetry as an optional dependency to all Docker images that used poetry packages was our workaround, despite the fact that we never used poetry itself after that step.

poetry = {version = "^1.1.4", optional = true}

The Future of Poetry

Fast forward several months and the dust has long settled. Poetry is simply the way of the world now and the old ways have vanished with nary a pip. There have been a couple of hurdles along the way, such as a few incredibly niche issues caused by a half dozen simultaneous edge-cases, but none of which weren't resolved quickly and easily. We've submitted a couple of contributions to fix some issues with Poetry's terminal output and to aid in supporting keyring-style authentication, but for the most part, everything has been fast and functional; in fact, it's been easy to forget this was once a problem which could have impacted us so wildly.

Being able to forget this was once a Big Deal™ is, of course, the goal; all in all, we've been happy with the new system and have been able to spend less time worrying about our packaging and more time working on that whole "big data and AI" thing people seem to be excited about.

On better models of myself

2021-11-25T00:00:00Z

Today was the first day in 2 months that I did not have a bowl of chocolate cereal and milk for at least one meal. Most people don't know I've been doing this, and a few of those who do expressed shock and concern because they didn't know why. So here's a fun fact / piece of information about me: I'm autistic!

I've been considering this for about 2-3 years now as a model that neatly explains a lot of things about me, and although I've tried to have conversations with various people about it at various points, it wasn't until early this year that I really started seeking advice and community in a serious way.

I want to use this post to answer some FAQs and talk about the journey!

Labels as tools

I'd be wrong to suggest that labels haven't played a big role in my life. As a child without the word "bisexual," I went back and forth between IDing as a lesbian and thinking, well, I must not be a lesbian because I'm not exclusively attracted to women. Woman was not a label I felt particular affinity for but for a long time I didn't know anything else was an option, but when I eventually discovered the label "agender," I found it felt like home.

Most of my life I've been pretty aware of my verbs (liking people of all genders, feeling no gender, recovering after social interactions, hurting from sounds, etc.), just not my nouns, and discovering the nouns that fit my verbs opened up new worlds, mostly in terms of being able to explain myself to other people.

A lot of things about me seem to really confound people. I can hear, so why would I struggle to follow movies without subtitles? Surely my closing my ears and running away when I hear certain sounds just means that I'm trying to get attention, not that it physically pains me, even when I say so. If I seemed to be able to grasp things, clearly the only explanation for me dropping them all the time was that I was being careless, not that I was having any issues with my hands. And consistently throughout my life, I have perplexed people with how bad I am at getting non-literal language.

I think my favourite example of this was earlier this year when I missed something like 3 jokes in a row that a friend made. This friend is autistic but I hadn't yet brought up my autism. They looked at me and I just threw my hands in the air and said, "I don't know, I just don't get things sometimes," and they said, "Do you realize you're not getting things more often than an autistic person?" and I had to say, well, about that...

Autism is a noun that explains many of my verbs - verbs that I had mostly accepted as just who I am, but that, for other people, are imperative to point out or disbelieve or even make fun of.

I would love to live in a society where I could just ask for the subtitles to be turned on and they were just turned on, but since we instead live in a society where I'm required to declare a justification for my divergent needs and behaviours, the label of autism is a tool to avoid hurt.

Intersectional identities

Kimberlé Crenshaw's framework of intersectionality is popularly interpreted as there being x, y, z variables (gender, race, class, etc.) and your oppression being some sort of linear combination of the oppression you face in any one of those variables. This interpretation is not just superficial but it's also entirely incorrect, as many people much cleverer than me have repeatedly pointed out.

Her framework is actually about how intersections of identities create unique experiences, privileges and hardships that often don't exist outside that particular intersection. It has been a useful lens with which to consider my experiences since I first heard about it and it's been particularly interesting to think about how autism fits in here.

Autism, age, socialization and self-diagnosis

I was brought up as a girl, which has interesting consequences for what my autism journey has looked like. Fact: Young girls are much less likely to be diagnosed with autism than young boys. A popular explanation offered for this is that girls just "have it less," but this is not a great argument. For starters, the medical profession has a long (ongoing) tradition of privileging the bodies of white men in their research and their symptom descriptions. Secondly, gender socialization starts very early and can strongly incentivize "masking," a survival strategy for autistic people to basically "pass" as neurotypical.

The other consequence of this in a world where autism diagnosis and medicalization is geared so strongly towards children, is that if no one picks up on it when you're a child, then official medical diagnosis is basically out of reach. I looked up the wait times for an adult autism diagnosis in Canada once and it was something like 3 years.

All this means that there is a large community of people who discover that they are autistic in their adulthood and self-diagnose instead of jumping through years of bureaucratic hoops only to potentially be dismissed anyway. After all, not having a piece of paper from a prejudiced system can't possibly invalidate your lived reality of sensory issues and communication difficulties.

Interestingly, literally every autistic person I've spoken to has validated my self-diagnosis (or even clocked me as autistic!) regardless of their own diagnosis status, and it's only allistic people who say things like, "Oh, you can't possibly be autistic," and "Self-diagnosis isn't real." I think it is easier to be uncritical about a medical system that doesn't often fail you, but that's perhaps a discussion I'll save for another post.

Autism and being racialized

The curse of my existence is being constantly beaten with a shapeshifting rod, the nature of which is sometimes ableism, sometimes racism, sometimes misogyny, sometimes erasure, and very often, all of the above. Having brown skin and an unplaceable non-prestige English accent means that people who go in with the mindset that I can't possibly have autism come up with pretty awful explanations for some of the things I experience.

Like I said, people very often notice how much trouble I have with understanding sarcasm and jokes (I actually have a number of tricks to compensate for this with limited success - a post for another day!). Multiple times when I've brought up that I think it might be a sign of autism, I've been dismissed with statements like "It's probably just an L2 English speaker thing," (I'm not an L2 English speaker) or "You probably weren't exposed to as much wit and humour, culturally," which, frankly, is quite racist! And it's especially hurtful when it comes from people who know you much better than to make assumptions like that.

Because of these responses, I haven't felt as comfortable talking about other things I experience like days when any kind of languaging feels like wading through sludge, or days when I feel like speaking is difficult but texting would be fine. Many autistic people experience this regularly, but so many people are so quick to try and attribute my experience of these things to factors based on incorrect assumptions, that I don't know if I can talk about it and be taken at my word.

Autism and gender

There's also a correlation between gender-nonconformance and autism that's interesting to think about in the context of my being agender. I've said plenty about my gender and my pronouns before, so I won't say more other than that all of these puzzle pieces sure seem to fit very well together in my understanding of myself.

Future work, future writing and thank yous

You can probably tell that I could expound at length about my opinions on things like medical gatekeeping (in multiple contexts), medicalization, the harms of the deficit model of understanding autism, the problem with accommodations, etc. It's important to point out here that my opinions on these things have come from listening to many disabled people and engaging with disability justice activism. There's so much fantastic writing (both academic and not) that deals with these subjects and I'd do well to use my soapbox to share other people's perspecives too. So I'll try to make my next post about autism a collection of resources I've found useful.

I owe a lot to the folks creating these resources, telling their stories - the autistic people whose writing I saw myself in, those who have had conversations with me about shared experiences and self-diagnosis and official diagnosis and given me nouns for my verbs (like parallel play and dyspraxia and stimming and sensory overload).

I also want to thank all the disabled people who've played a role in radicalizing me against societal ableism and my internalized ableism. This is certainly an area of future work (on myself), probably forever, and could be something I write about too.

Answered FAQs

Referring to me? Please feel free to say: Vagrant is autistic / xe has autism / they are on the spectrum. Please do not say: Vagrant has Asperger's / xe is high-functioning / "you don't look like you're autistic"
Vaccines? Please get vaccinated
A cure? No thank you. Most autistic people don't want a cure - we're actually quite happy being ourselves. I personally would love for allistic people to understand and accommodate some of my autistic needs better, though.
Shock therapy? No thank you. There's a FDA ban on using it on autistic people in the US but there's a bill in Massachussetts currently to overturn that ban. Read more about #StopTheShock.
Applied Behaviour Analysis? No thank you. I didn't experience it as a child and I don't have to as an adult with security and stability, but autistic people who've experienced it call it abuse and it's worth listening to them.
American Birding Association? Hell yes
Autism Speaks? No thank you. They aren't a community-led organization (they don't have a single autistic person on their board, for example), the autistic community hates them, they run awful campaigns that stigmatize autism and encourage eugenics, etc. Please direct your money to autistic people and to self-advocacy groups instead, if you'd like to donate.
Simon Baron-Cohen? No thank you
Sacha Baron-Cohen? Sure, but I'll be honest - I've never actually watched Borat
Why didn't you eat chocolate cereal today? Great question! It's a samefood which essentially means it brings me a lot of comfort and is another form of structure that helps me feel safe. I think that not eating chocolate cereal today maybe means I'm feeling more settled in Germany, settled enough to loosen up some of my rigidity!

One Hundred Bird Facts

2021-12-31T00:00:00Z

This is an improved and CC BY-SA 4.0 licensed version of a list of bird facts I originally wrote on Twitter

Woodpeckers have ridiculously long tongues that coil up inside their heads and the muscles associated with the tongue protect them against concussions when they peck wood.
Chickadees are named because of their call (chicka-dee-dee...) The number of "dees" in the call at any given point in time indicates the severity of a nearby threat. So 3-4 dees for a human, 7-8 dees for a hawk or an owl, for example.
There is a bird called a fiery-throated metaltail and unlike what you may expect, it isn't a bird of prey but rather a small, cute Peruvian hummingbird.
Birds do not have teeth. They typically break their food on the outside (against rocks, with talons, etc.) and swallow it whole.
Sometimes you get gynandromorph birds where a bird has characteristics of both the male and the female birds of the species. It's easiest to spot in sexually dimorphic species (i.e., where the male and female are visibly different).
The smallest bird in the world is the bee hummingbird which can be just 5 cm big (just shy of 2 inches) and not even 2 grams (0.07 oz) in weight.
The piercing cry that you probably associate with bald eagles because of movies is actually from the red-tailed hawk, which is very commonly used as a voice actor for bald eagles. The actual bald eagle call is extremely wimpy.
Not-quite-adult bald eagle
Sometimes birds that eat berries get drunk on overripe fruit that's started to ferment.
The fastest bird in the world is the peregrine falcon (gorgeous bird). It's been clocked at speeds of 320 kmph (220 mph), making it the fastest animal ever (cheetahs are just shy of 100 kmph / 62 mph).
The loudest recorded bird so far is the white bellbird (up to 125 decibels, the equivalent of a jet engine taking off). The males scream into the faces of the females to prove they are worthy mates. Mmhmm.
Flamingos get their colour from the food they eat - they need carotenes to get their bright pink colour.
The iridescence on hummingbirds comes from structural colour, i.e., the literal molecular structure of those feathers, which is why it looks different from different angles. There's cool physics involved - the physics of light and refraction.
Rufous hummingbird
Blue-coloured birds like blue jays, Steller's jays and bluebirds get their blue colour from structural colour too, but in this case it isn't iridescent / sparkly like hummingbirds.
Steller's Jay
Many birds get their colour from melanins! And just like humans, when there are issues with melanin production, you get albino birds and leucistic birds (birds that have partial loss of their colour and look much whiter).
Leucistic female mallard
There's been research that shows that crows can identify and remember individual faces, and even pass on this information to future generations!
When researchers study crow funerals, they actually use masks and wigs so that the crows don't remember them and think that they killed one of theirs. Absolutely not kidding.
After habitat loss, outdoor cats are the number one cause of bird declines in North America. Not window collisions, not power lines, etc., not by far. Peer-reviewed research says so, so please keep your cats indoors and give them catios that are cat-safe and bird-safe!
You know how there's that cartoon trope about owls being able to rotate their heads 360°? Well, they can't. I believe 270° is the recorded max.
Birds are feathered theropods (hollow-boned dinosaurs) and are the ONLY KNOWN LIVING DINOSAURS. Get out of here with your crocodiles and alligators. Bow to our avian overlords.
Birds have a third eyelid, a "nictitating membrane," that they can actively control. It's a transparent-to-translucent eyelid that can be used to moisten and protect the eyes and clear debris while still being able to see.
As far as I know, the latest record-holder for longest uninterrupted flight is the common swift, which has been recorded staying in the air for ten months straight. Ten months. That's ridiculous.
You might associate parrots with a tropical or subtropical climate but there does in fact exist one species of alpine parrot - the kea. It's a large parrot found in New Zealand and they're sweet and curious and have been documented using tools.
Some birds are brood parasites, i.e., they lay their eggs in other birds' nests and have those birds raise their young. The brown-headed cowbird might be the most famous brood parasitic bird and unfortunately it gets a bad rep for this (undeserved, imo).
Male brown-headed cowbird
Female brown-headed cowbird
Birds can be gay. You might have heard of the Central Park Zoo's famous gay penguin couple, but homosexuality has also been seen in other species of birds - flamingos and the amusingly named cock-of-the-rock come to mind.
Birds are sometimes given "eponymous" names in English, i.e., they're named after people. Unfortunately this often erases Indigenous and local history in favour of colonizers' names. Bird Names For Birds is campaigning to change this!
Language fact: 'seagull' is a colloquial / lay term, the technical term used by birders and ornithologist is simply 'gull'.
Silver gull
The American kestrel is a tiny bird of a prey commonly referred to as a "sparrow hawk," despite being neither a sparrow nor a hawk, but a species of falcon.
Funny / naughty bird names should count as a bird fact. E.g., (off the top of my head) bushtit, booby, dickcissel, shag, woodcock, and as previously mentioned, cock-of-the-rock.
Odd fact I learned recently: when pelicans dive, they turn their bodies slightly to the left to avoid breaking their windpipes with the force of the impact.
Spot-billed pelican
Just like me with my goth platform boots, the secretary bird stomps on its enemies (snakes). No, really - they have very powerful feet (force of up to 5x its own body weight!) and they kill snakes with quick stomps of their feet.
The bird most commonly referred to as just "pigeon" is a feral pigeon (feral version of the domestic pigeon which was domesticated from rock doves). They can actually be quite beautiful. Next time you see one, try to appreciate the glossy iridescence on the back of its neck!
Rock pigeon
Birds can have symbiotic relationships with other animals. For example, cattle egrets eat ticks and flies off cattle's backs and get nutrition from it.
Cattle egret with a brown cow
However, the purported symbiotic relationship between Egyptian plovers and the Nile crocodile (where the plover supposedly sits in the crocodile's mouth and gives it a dental cleaning, thereby also feeding itself) is considered suspect. Scientists: "pics or it didn't happen."
Birds can also have similar mutually beneficial relationships with plants. Frugivorous birds (ones that eat fruit) help plants by dispersing their seeds while also getting nutrition from the fruit.
There is a bird called the "vegetarian finch" which is named because its diet consists mainly of vegetarian things like fruits and leaves (not uncommon in birds). It's one of the many species classified under the label Galapagos finches (or Darwin's finches).
Yes, that Darwin! Those finches were a big part of the theory of evolution and are mentioned a bunch in his book On the Origin of Species. This is a good point to note that it is Bad to use evolutionary biology as an argument for eugenics, so please don't.
One last finch fact. There's another finch named for its diet - the vampire ground finch. Yes, it drinks the blood of other birds by pecking them till it draws blood and then drinking the blood. Isn't that wild?
Let me tell you some bird vision facts. Vision is important for almost all birds, with some exceptions. E.g., kiwis rely more on their senses of sound and smell than sight, so they have underdeveloped sight, and some blind kiwis have even been found thriving in the wild.
Most other nocturnal birds (yes, kiwis are nocturnal) have evolutionary adaptations for improved vision in the dark. They usually have larger eyes and proportionally more rods - one of the two types of light receptors in the eye, the one that helps with low-light vision.
Cones are the other type of photoreceptor in the eye, and they're responsible for colour vision. Humans have cones for red, green and blue (roughly) which we combine into colours. Most birds have cones for red, green, blue and ultraviolet. So yeah, these birds have UV vision!
The coolest bird vision fact imo is that birds use their left and right eyes for different things. It's like handedness in humans but for eyes, and it works because like us, birds have lateralized brain function (left and right brain hemispheres that handle different things).
Oh, another thing I remembered that's related to the physics of light is that you know how when things are underwater, there's refraction and the image of the thing looks like in its a different place than it is? Birds that fish have to correct for this, and they do!
Great blue heron
Unfortunately I don't know as much about bird hearing other than that it's better than human hearing. I do know that owls have faces engineered to funnel noise directly towards their ears, and asymmetrical ears that help them echo-locate better without moving their heads.
Besides hearing well, owls also need to be quiet so that they aren't heard by their prey. Their slow flight and the structure of their wing feathers are adapted so they can fly soundlessly and without having to flap as much.
I once read something about how birds near airports seemed to be experiencing hearing loss because of the noise of jet engines, and that this was apparent from the frequencies in their birdsong. I did a quick search and this seems to be true but understudied.
Birds don't have noses. Most birds have nares though (nostrils) and some of them do seem to use smell to track down food - the most classic example is that vultures use their sense of smell to track down dead and decaying meat, which is what they eat!
Speaking of vultures, one of their characteristic things is that they have bald heads with no feathers. This is to keep themselves clean when they stick their heads into the carcasses they eat, and also very important to regulate their temperature when they fly high.
Every time a country has declared a war on birds, it has lost. Australia tried waging a war against emus and deployed the literal army. The emus just gave the army the finger and continued to exist and reproduce and wreak havoc on the crops. Australia eventually gave up lol.
Emu
China—well, Mao Zedong—also tried to declare a war on sparrows. They thought that sparrows were eating lots of grain and so they started to kill them. But this upset the ecological balance because sparrows also eat locusts, and they eventually had to leave the sparrows alone.
Fox sparrow
The thing about how bird migration works is that we Don't Know™. But there are fun theories including that birds have an internal map, that they use smell to navigate, that they use the Earth's magnetic field to navigate. There's good evidence for these theories too.
There are also theories that birds use the sun and/or the stars and/or their internal circadian rhythm to navigate. They could also use landmarks. Maybe some combination of all of this. We don't know. Birds are just really cool like that.
Now sometimes for various reasons you get vagrant birds - birds that are far outside of their typical geographic range. These are my favourites! Sometimes they have bad navigation systems, sometimes they're blown off course by storms, sometimes they're just out for adventure!
Ok, let's do some more specific bird facts. You know how magpies are thought to like shiny, expensive objects? That's a myth! It turns out magpies are actually scared of shiny things (and also blue things, which were tested in the same study).
But satin bowerbirds absolutely love blue things. The males create and decorate bowers to entice the females, and they use more blue than other colours but no one really knows why. If you want to know more about bowerbirds, you should watch the documentary Dancing With The Birds.
Oh you know what I'd love to watch a documentary about? The New Caledonian crow. They've been documented using twigs to "fish" for grubs inside hard-to-reach places. Sometimes they'll literally fashion twigs into hooks. They're such good birds!!!
Ooh, let's talk bird pee and poop. You may know from experience that birds just go for it, mid-flight or whatever. You may not know that what comes out is actually pee AND poop, because they have one opening for both, and also no bladder.
Bird excrement
This opening is called the cloaca and, uh, fun fact - it's ALSO used for reproduction in most birds. When birds mate, they touch cloacas and this is called the cloacal kiss. I swear!
Male birds of some species have dicks. I will let you look up duck penis on your own time, and I will preemptively warn you that you may be grossed out by it.
Back to bird pee and poop. Most bird nests aren't filled with excrement despite being filled with babies (notorious generators of excrement), so what gives? Well, some baby birds will actually go to the edge of the nest and pee / poo over the edge.
Young red-cheeked bulbul in its nest
Other baby birds have evolved to eject their waste encased in a little white bag (called a fecal sac), that the parent then picks up and gets rid of. It's like the baby poos a diaper along with its poo. It's ridiculous and amazing.
The preen oil in ducks (and other water birds) waterproofs their feathers. Bird preening is a way for birds to clean their feathers and evenly distribute this preen oil all over themselves.
Preening black-headed ibises
I haven't talked about the ossifrage / lammergeier / bearded vulture! If you thought it had a cool name, wait till you hear that it's diet is almost exclusively bones!!! Ossifrage means bone-breaker. It drops bones onto rocks from heights to break them and get at the marrow.
Shrikes (colloquially referred to as butcherbirds) impale their prey on sharp things like thorns or barbed wire. Metal as fuck.
Another fun language fact: shrikes are called shrikes because they shriek, and their family name Laniidae comes from the Latin word for butcher (lanius). (I don't know Latin or most scientific names of birds, I swear, this one just stuck because it's cool)
Baby crows have blue eyes and pink "gapes" - the edges of the mouth. Both darken as they grow older. I'm not sure why they have blue eyes, but having brightly coloured mouths is pretty common for baby birds to get their parents' attention to feed them.
Many bird species form creches to raise their young together. Just like a human creche, care for several young is delegated to a small number of responsible adults, so that more adults can spend time foraging, etc. Penguins and mergansers are some species that do this.
I saw a belted kingfisher fishing at the lake today, which is reminding me to tell you that kingfishers eat more than just fish. They eat insects, frogs, worms, etc. too, especially the ones that live in forests.
White-throated kingfisher
The only bird in the world to have a beak that's longer than its body is the sword-billed hummingbird. It literally looks like Pinocchio, it's incredible.
Let's talk bird sleep. I mentioned earlier in the thread that birds have lateralized brain function like humans do, and that they can use each eye independently. This results in birds being able to sleep with one eye open / with one half of their brain, just like dolphins!
Birds sometimes sleep with one leg tucked up into their feathers. They do this to regulate their temperature! One less exposed leg is one less avenue to lose body heat, and can also give that leg a rest.
Oh and an additional consequence of #69 is that birds can sleep while flying!! They take little power naps!!! With half their brain!!!!
Soaring birds are fun. They use air currents to stay up without ever flapping their wings. They use "thermals" (columns of hot air) to stay aloft, because hot air rises and takes them with it!
Plumage (the feathers) of a bird are moulted often. Sometimes it's twice a year, sometimes once, sometimes only once every few years. It makes birds more challenging to identify (as a birder) because some of them literally change appearance with the season.
Male wood duck
A cool fact I learned recently is that in Māori, there are often different names for the same bird in different seasons, which presumably corresponds to a difference in the plumage. Isn't that cool?
Flocking, apart from being a really interesting modelling problem, is cool because it shows cooperation with hierarchy. Birds that fly in V formations are basically exploiting aerodynamics so that the ones behind can fly faster more easily.
It turns out that J formations are more common for flocking birds than V formations, by the way, even though most people think of it as a V and refer to it like that!
Smaller birds have more complicated aerodynamics than larger ones and so this method of straight lines and Vs and Js tends to work less well for them. They tend to just cluster together in large flocks that somehow miraculously fly in an orderly way despite the chaos.
The prettiest form of flocking has to be murmurations of starlings! Thousands of them will gather into a flock and they make beautiful swirly patterns in the sky, like auroras but composed of birds flying in unison.
Fork-tailed drongos will use their alarm calls when there isn't a predator around to fool other birds and animals into abandoning their prey, and then they steal it and eat it! Clever bastards.
Bird banding is one way that researchers study birds' migratory habits and lives. It involves putting a metal band around a bird's leg with a unique ID on it.
Banded ibis foot
When you think of birds that can mimic human speech, parrots might be what comes to mind. But it turns out lyrebirds, mynahs and starlings are also very good at it!
Did you know that birds yawn? Yawning is one of many "comfort movements," a term used for behaviours that birds exhibit that serve their mental and physical health.
Bonaparte's gull yawning
One of the big pieces of bird news this year was that two California condors were produced asexually - they were genetically related to their egg-laying parent, but they contained no genetic material from any others! This asexual reproduction is called parthenogenesis.
We humans have been using birds for centuries, and falconry (using birds to help you hunt) is one of the oldest ways. Despite the name, falconry doesn't just involve falcons - you can also be a falconer with a hawk or an eagle!
Trained birds of prey are now sometimes employed as pigeon control at airports and in open-air areas where humans like to hang out. The next time your plane takes off and a bird doesn't fly into the engine, thank a hawk.
People think of twigs as traditional nesting material but birds that build nests are very creative and can use a lot more than that, including leaves, cotton, moss, fur, spider silk, etc.
Barn swallow with nesting material
But there is a wide range of nesting habits across birds, from pigeon-style messy, almost incomplete-looking nests, to the elaborately woven multi-generational homes of weaver birds.
The oldest bird we know of is a Laysan albatross called Wisdom. She is over 70 years old and she returns every year to Midway Atoll where she was first tagged. This year, she's even incubating an egg with her partner.
Larger birds often get harassed by smaller birds, regardless of whether they are alone or in a flock. This behaviour is called "mobbing" and smaller birds do it to protect their nests. They can afford to do it because their smaller size makes them more agile.
Male red-winged blackbird mobbing a juvenile bald eagle
When it rains, birds generally try to find shelter so that they can stay dry, and therefore, warm. But if it rains for too long they do eventually have to come back out to eat.
Speaking of staying warm, one of the cutest techniques that birds have to stay warm is fluffing their feathers, which traps air into air pockets and provides insulation. This is why you see birds become fluffier and rounder in the fall and winter.
Juvenile male red-winged blackbird
Bird species are often described as being sexually monomorphic (i.e, the male and female birds look identical) or sexually dimorphic (i.e., the male and females look noticeably different).
Sexual dimorphism can manifest in different ways. Often the male birds have brighter plumage than the female ones, e.g., the brown-headed cowbirds above. In raptors, it tends to be a size difference where the female birds are much larger than the male ones.
Other than keeping your cats inside, one of the best things you can do for birds is to not rake up your leaf litter. Birds eat insects that hang out in leaf litter and they use it for nesting material. So leave the leaves, save the birds!
Birds sometimes hang out in mixed-species flocks which has two advantages - with a wider range of skills they can watch out for predators better, and if they don't have the same food needs then they don't compete with each other for food.
Bird babies of different species range widely on the spectrum of precociality. Some are altricial like human babies - they're born helpless and require several years of care to fend for themselves. Others are precocial and leave the nest soon after hatching.
As a linguist, I am interested in the overlaps between linguistics and birds. One such overlap is that birds have accents in different places just like speech accents in humans - they're called song dialects, and they can sometimes even vary in neighbourhoods.
Another area of overlap is "warblish," a term for using existing words from spoken languages to imitate bird calls, unlike onomatopoeia which creates new words (e.g., "the flicker says wicka-wicka" vs "the barred owl says who cooks for you")
Turkeys are weird but their names are even weirder. In English it's a turkey, in Turkish it's a hindi, in Khmer it's a French chicken and in Portugal it's a peru?! Here's a fun article about multilingual turkeys.
Domesticated turkey
You may associate birders with expensive equipment like fancy cameras, telescope lenses and massive spotting scopes. But all you really need to be a birder is an interest in birds!

And there you have it - 100 bird facts.

If any of this piqued your interest in birding, here are some websites, groups and apps I can recommend (unfortunately fairly specific to North America):

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

New name: Vagrant!

2022-04-30T00:00:00Z

I have a new name, and it's Vagrant! I also have a new handle everywhere, and it's DippedRusk.

Instructions on use

Caveat: Any personal communication with you will override these generic instructions for everyone.

Please refer to me as Vagrant
If you come across other people using my old name, please correct them
I am not keeping my old name a secret, but I would be uncomfortable if people went around digging for that information or giving it to other people, so please don't
If you have easily updateable (by your judgement) materials that use my old name, please update them

Vagrant?

I tried out the name for nearly a year with some groups of friends (on Twitter, on Discord, as a D&D character), and it felt perfect and mine so it's now public. I have three main reasons for picking Vagrant, and unsurprisingly, birds are involved:

A vagrant bird is an individual that is found well outside the typical range for its species. Sometimes it's because it was blown off course by storms, sometimes because it's an intrepid adventurer. Whatever the reason, they're unusual and rare and therefore very exciting to come across, as a birder.
The word is also used to describe a person who wanders from place to place without a permanent home. As of this writing, I've lived in India, Singapore, Canada, and Germany. People close to me live all over the planet, and home to me is wherever I am or they are.
It leaves my initials unchanged, which means I don't have to go chasing down the tiny handful of people who've cited me so far to update their citations.

Dipped Rusk?

I've also changed my handle everywhere to something that isn't linked to my name, just in case any part of it is to change in future. Rusk was also a name I tried out briefly but it didn't work quite as well for me as Vagrant.

The phrase "dipped rusk" evokes a lovely childhood memory of afternoon tea at my grandparents' house when I was a kid. I wasn't allowed tea, of course - no caffeine for kids - so I would drink milk while Mooch drank tea that Ammamma had made. We'd usually have a snack with it, which was sometimes rusk, a hard, dry biscuit. He'd dip it in his chaya to soften it and he'd give me some as a sneaky exception to the no-caffeine rule. There's an art to getting the dipping right - you have to time it perfectly so that the biscuit gets soggy enough to be pleasant to eat, but not so long that it loses its integrity and sinks to the bottom of the cup.

Miscellaneous notes on addressing and referring to me

I prefer being addressed as just Vagrant. If you insist on formality, I prefer no title or the title Mx (pronounced like 'mix'). If you build or have control over systems that do not allow for these possibilities, change them.

Instead of words like guy, woman, man, etc., please refer to me as a person in all languages. In German, I tend to use masculine noun forms because they are unmarked, e.g., ich bin Student.

My English pronouns are xe/xem/xyr(self) and they/them/their(self). In German, my pronouns are xier/xiem/xien/xies(...), but I'll accept any neopronouns or no pronouns too. In other languages, if there are gender-neutral versions of words (pronouns, adjectives, nouns, etc.), please make an effort to use those. When in doubt, you can always ask me.

What we owe each other

2022-05-07T00:00:00Z

Content note: There are mentions but no descriptions of assault in this post multiple times. There is a lot of content about the aftermath of dealing with assault.

Assault

I'm a big fan of punchy openings but I never know where to begin with my stories of assault. It's a funny thing, how I've told these stories so many times to so many people for so many years, and I still don't really know where to begin, much less how to stop once I've gotten started. I'll spare you the details in the interest of saving me from re-living trauma and also because that isn't the point of this post anyway. In the big picture, the incidents of assault themselves were the smallest part of the story thereafter.

Like a lot of people who have experienced assault, I went to therapy to try to heal. I actually couldn't have afforded it if not for the Crime Victim Assistance Program, and the only reason I had access to the counselling benefits from that was because one of the people who assaulted me got arrested and charged with a crime.

The months after that particular assault were the hardest because apart from being traumatized and depressed and unable to function properly, I kept receiving mail about the court proceedings. The person was released the morning after the arrest on the grounds that he wasn't dangerous enough to be worth the money to confine him. One of the conditions of his release was that he was not to contact me or be in my vicinity again, and to enforce this they had to name me in the document, de-anonymizing me to this stranger. And well, with a name like mine, I can't disappear, even in a city as big as Vancouver. I was terrified and hurt and angry and all I could see at that point was a system that could've kept me safe by confining and punishing someone who had hurt me.

But as I received more mail about the case and as I talked to case workers, I started to discover more. I heard he was dealing with various issues, including homelessness and mental illness and various addictions, and well, at some point I started to find it rather ironic that I was the one who was given access to a year of therapy and money to replace my ruined clothes, while he was given absolutely nothing. Not even the basics - food, shelter, medication.

One of the final pieces of mail I got about this whole incident invited me to submit a victim statement that would be shown in court. It had a very open-ended format where you could write a letter or a poem or draw something, and if it was writing, it would be read out for the person. I didn't end up sending anything in because I didn't know what to say. All I had were questions, and none were ones that he could answer. I felt that he had been harmed by society more than I had by him or anyone, and so how would he be in a place to understand the harm he did to me? And the harm he was experiencing was on such a different scale, so did it even matter? What might he have needed to be in a place to respond to this better? What did I need from him?

I've been in therapy for years now and all I have is more questions. Not just pertaining to this person and imagining what a world would look like if he got as much therapy and food and shelter and support as I have, but more generally - about other assaults that looked completely different.

Another story features a university student like me, who had the basics and more already taken care of, who was good-looking and successful and on scholarship and who (to my knowledge) never had to reckon with any of what he did in any way, let alone in a punitive legal system. Yet another story is a combination of assault and abuse by someone who was very close to me, cared for me a lot, and promised to keep me safe at a vulnerable time for me, but could keep that promise only partially, doing a lot of harm to me at the same time.

I haven't wanted to interact with these people again since removing them from my life, but I do hope they grew. I hope they found what they needed to not harm other people in these ways again, even if it was just the knowledge of what's okay and what's not. In some cases, I was able to tell them. In other cases, someone else did. And in yet other cases, I don't think anyone knew or said anything at all, and I'm not sure how one learns from that.

In thinking about these experiences, the question that has kept coming up for me has been: What would a world look like in which there was enough space for all of us to heal from the pain we experience, and grow from the harms we inflict, separately or together, safely? I centred my healing and in most of these cases (especially the ones involving men), the world was with me and I am grateful for that, but I no longer think my healing needs to be to the exclusion of other people's healing or their growth.

Abolition

A few weeks ago I virtually attended this fantastic conference on abolitionist feminism, and a lot of what I learned there seemed to directly address this question and it gave me hope. Three ideas stuck with me:

"Abolition feminism cares about safety for everyone, everywhere, all the time. The only way we survive is to see how all of our fights are connected." - Angélica Cházaro

Something about starting with this premise feels correct because it seems to echo my question. It means that when someone harms me, there is space for both of our lives to matter, for both of our safeties to matter. This means there is space for their needs to be addressed as much as mine are, and they are not necessarily incompatible. Under this premise, it is clear that imprisoning someone for my safety or peace of mind is not a globally optimal solution, even if I want to be far away and safe from them.
"I am less interested in labelling oppressors and victims, and more interested in the question: How do we wrest harm out of our communities?" - Treva Lindsey

That's really the question, isn't it? My therapist tells me she's proud of the way I manage to exist in the grey areas of life, and really, I can't imagine any other reaction to my particular set of life experiences. As soon as we peel back a layer in many of my experiences of harm, it becomes clear that the oppressor-victim dichotomy often needs more nuance. What if a person is an oppressor in one way and a victim in another, like several of the people who harmed me? Richie Reseda said it was a "patriarchical paradigm" to split things into a binary - right and wrong, good and bad. Rather than creating these binaries which leave us stuck and unable to move forward, he says, "Abolitionist feminism is about belief in the power of transformation, about leaning into conflict with curiosity and courage and hope."
"We keep us safe." - Zachary Norris

This is one of the central premises of abolition - that we know best how to keep ourselves safe, our communities safe, and that we can't outsource justice to cops or prisons or "the universe," because it is us who must do the work, communally. Kandace Montgomery's vision for an abolitionist future was one where people would "get the care they need and protect themselves from unsafe partners without having to criminalize that person and bring the state into their familyhood." It is about realizing that most of the harm we do and have done to us is from the people close to us, and the solution is, in Richie Reseda's words, "being even more concerned about our choices and how we act with each other."

At this conference, Leigh Goodmark said something that has been rattling around in my head ever since. She said, as a white woman, her whole life, carceral feminism was just feminism. I'm not white, of course, but I have definitely grown up with the same frame. I was reminded of the responses I saw to sexual violence in India when I was a teenager - with folks polarized between victim-blaming on one side, and on the other side, protesters taking to the streets calling for the death penalty and/or chemical castration for rapists. I suppose this felt like a way of taking back control, and when I was younger this was exactly the sort of way I funneled my rage, but it just doesn't feel compatible with who I am anymore. The black and white thinking here is not about whether an act of harm was violent or not, or violent enough or not. The black and white thinking here that I disagree with is that there are no options between death penalties and victim-blaming.

A lot of what I learned and heard at this conference solidified feelings I already had, but it didn't necessarily give me concrete strategies and actions to carry forward in my life. There have been times in my life when I've felt disempowered, and like I needed to invoke carceral systems to re-empower myself, but the more I've thought about it over the years the less that has made sense to me, and I've wanted to learn about alternate ways of dealing with situations both as an individual and at a societal level.

Anarchism

Enter Isis Agora Lovecruft. Now, I already thought they were massively cool but after I read their blog post about anarchist justice, I just have infinitely more respect for them. I highly recommend reading it but the gist is: they tell the stories of two people who raped other people, both of whom were offered anarchist, rehabilitation-focused opportunities for transformative justice. The similarities end there. One participated, grew and changed, apologized to his victims and begged for forgiveness at one point, and was eventually able to convince both his victims and others involved in this process that he was not going to abuse anyone again. The other person retaliated with threats, verbally expressed that he was trying to ruin his victims' lives, and never apologized.

Despite the difference between these two people, and despite the very important fact that they themself were directly harmed by the second, Lovecruft stands firm in their beliefs - that they can't condone violence and threats against this second person, that prison is not a solution (for various reasons outlined in the post), and that he needs help. All these positions are evidence-based and come with links to data in their post.

There are two things I got out of this post: 1, that I admire their unwavering commitment to their moral system (anarchism and transformative justice, in this case), and 2, that they provide, with the case of the first of the two rapists, a concrete example of what such a transformative justice process looks like in practice.

The safety of the victims is critical and they are centred instead of disempowered: they get the final say. But at the same time, there is this fundamental belief in the person's ability to change and someday convince his victims that he would not harm anyone in that way again. Everyone in the situation (including the victims, importantly!) holds this belief. This story offers me a lot of hope as someone who has been a victim. It also offers me a lot of hope because the person who harmed others isn't just thrown away.

A critical point to remember about this story and this process, though, is that we all have a role to play in what transformative justice looks like. In the first case where they weren't a victim, Lovecruft and others did a lot of work and invested time and effort into this person's rehabilitation program. They did this for the people they were in community with (which included the person and their victims, as I understand), and in this case the community was an anarchist collective. This is conceptually similar to initiatives like the Policing Alternatives & Diversion Initiative that I learned about at the abolition feminism conference. The people who are a part of this initiative define community more broadly, based on geography, and help reduce arrest and incarceration by helping address the immediate needs (hot food, clean clothing) of people experiencing poverty, substance abuse and/or mental health issues.

So I've been asking myself whether I would have it in me to do that for someone I'm in community with, and also what communities I am a part of in that sense. I can and would certainly do it for the people who are close to me, but beyond that I think the honest answer is that I am not brave or principled enough to invest the energy and time.

All of this effort to help people address their needs and prevent them from being harmed or from harming others, all of it is a labour of love. And frankly, while I don't think I have that level of love for a broader group of people than my chosen family, I think it's beautiful that these other folks have these broader definitions of communities. Maybe it's something I want, even.

And this model of the anarchist collective and transformative justice in all its forms provides an alternate way of living life, of being in community with each other, one not bereft of consequences but where there's actually hope and the possibility of change, of moving forward.

But again, critically, it relies on people deciding that we owe each other something for whatever definition of community we use - that when there is a harm, there is an important role that can be played by the folks around the situation who aren't directly harmed.

Appreciation

One of my favourite quotes is this Proust one that goes: "Let us be grateful to the people who make us happy; they are the charming gardeners who make our souls blossom." I think about this quote often because I feel like I have had incredible luck with a lot of the people I choose to have in my life.

One of the greatest gifts I have ever gotten from anyone was when an ex and I slowly started to rebuild a friendship after many months of radio silence after the breakup. We spent years talking about the relationship, the breakup, what we did wrong, how we could be better. I remember a lot of bitterness initially in how we each talked about the other. It took time to find the tenderness we had had for each other before, but fundamentally, I think we were both committed to trying to grow and heal in the aftermath of that relationship. We came together to do that, and I think it's rare to have that happen and moreover for it to be successful, but we really both did a great job of listening to the other, understanding and accepting our mistakes, and trying to do better for future partners. This person remains my second oldest friend.

The cornerstones of my closest relationships with people are radical vulnerability and holding each other accountable, honest, and to higher standards. Whenever shit hits the fan, I am reminded that these qualities exist in my relationships in abundance, and I am so grateful for it. We are comfortable questioning decisions and opinions and words and actions. It means telling your friends you think they can hold themselves to a higher standard when they're hurting people and it means hearing them tell you they think you're needy and insecure and it means each of you telling the other they're being a shit friend when they are, and then trying to do something about it in all these cases. (All of this has actually happened.)

As an aside, what's not omnipresent in my relationships is fancy terminology. Having conversations about what it means to be in a strong friendship and how to help each other grow doesn't really require therapy words or words like abolition feminism or anarchism or any of it. To use another fancy word, what's more important is praxis, or what people actually do in their lives. Terminology is cool and certainly a useful tool for me to get access to more spaces and learn what people cleverer than me have to say about the nebulous thoughts I have, but it's not a prerequisite to my friendship and I would never want it to be.

Accountability

So now we come to the main question this blog post was supposed to address: what do we owe each other? This is only the title of this post in the sense that it's a question that's been roiling in my head for years. It is a question I want to answer for myself, and in true linguist fashion, I have no prescriptivist agenda for you to be exactly like me. Deciding what I owe others will let me commit to giving them that, allow me to be held accountable when I don't give people what I decided I owed them, and it is also a way of communicating to other people how I would like to be treated and held accountable for my harms.

Part of what I owe everyone is my beliefs. I believe without exception that no one is irredeemable and that everyone is capable of change. I firmly believe this about everyone, including all the people who have ever harmed me, however egregiously. This belief stems from having made mistakes myself and having learned things and grown and improved, and if I'm able to do that, well, then everyone is able to do that, and if I believe I am owed the patience and opportunity to learn and grow, then I owe that to others too.

Another important thing I owe everyone is basic respect and valuing their existence and needs as human beings. When confronted with emotions of anger and hurt and betrayal, it is so easy to, as my friend EJ puts it, "accidentally become a carceral bastard." But this is a thing I am actively working against, as are they. It seems to me that in the absence of love, we are more willing to throw people away, but I don't want to live my life treating people as disposable. This means wanting people who have hurt me to have their needs addressed and to have communities that hold them close and hold them accountable, so that they get better. The accountability bit is key; communities that make excuses for bad behaviour or allow harms to continue are not okay. But accountability can definitely co-exist with holding people close, rather than banishing them.

I also believe that people need help figuring out when and why and how they're fucking up, and help with becoming better. I know this also from my own life and the familiar experience of being told that if I didn't magically know what I'd done or said that was wrong, then that was even worse than just doing or saying the bad thing (imagine for a second, how this goes down for an autistic person, especially in childhood).

When people have directly harmed me, I believe I owe them the information - the when, why and how. I do not believe I necessarily owe them my presence after that or my help with getting better. My safety, security and mental peace are not things I am willing to sacrifice in order to help someone who has harmed me. This means nothing changes unless the person decides to change, and/or other people who remain in their life work to help them get better.

The role of these "other people" is also my role in situations where I have not been directly harmed. I see a huge opportunity (rather than an obligation) to affect change by being one of the people who helps someone stop fucking up, in whatever capacity. It is a way of showing up for the entire community, not just the person fucking up, because it is also a form of indirectly helping people who were harmed and people who might otherwise be harmed in the future.

The only remaining question is: how far does my willingness to take this opportunity extend? How much will I commit to, as something I owe other people, and how much will I decide I just can't do?

It feels quite easy for me to commit to helping my chosen family get better, with honesty and patience, in situations where I'm not a harmed party and my own safety is not at risk. I think we already all have a demonstrated history of doing this for each other, often even when we have harmed each other.

The harder thing is extending this commitment, one of work and energy and time - to not-close friends, to community members, to strangers, to worse harms than "fuckups," to people who don't seem to want to change, to people who have harmed me directly and badly, and so on. I don't think I can commit to consistently doing this just yet, but I have great respect for the abolitionists and anarchists I have encountered so far who do. And as I keep reading and thinking about this question, I wonder if perhaps the next post I write about this will be me deciding that that too is something we all owe each other.

Acknowledgements

Beka for the pointer to the Isis Lovecruft post. Everyone quoted in my livetweet of the abolition feminism conference. My therapists, obviously. My small army of friends, but in particular for this post: E, K, L, M, O.

Interview with an autist

2022-05-17T00:00:00Z

A friend of mine recently got an unofficial autism diagnosis from some medical professionals in his life and, knowing that I was autistic, asked if I'd be willing to chat. I pointed him to my previous writing "coming out" as autistic, and he had a bunch of follow-up questions to which I wrote detailed answers. A week later, he conveyed personal thanks from his therapist because I apparently got through to him about things he wasn't listening to his therapist about. So with his permission, I've converted his (unedited) questions and my (edited) answers into this blog post, in the hopes that it'll help more people.

Q. What are all the things that have helped since you've realized you were autistic?

So many things! I would say the two broad categories are: strategies to improve my relationships / interactions with other people, and strategies to help myself be happier when I'm alone.

With the first, it helps to articulate to other people what I need from them and what my communication style is so that there is as little confusion and discomfort for everyone as possible. For example, I ask people to be very direct with me because otherwise I don't always know exactly what they want from me, and it has led to many unpleasant situations where someone was saying something indirectly and I didn't understand so I didn't know to change what I was doing, and that made things uncomfortable. Two examples of this that come to mind are: saying, "You don't need to do that," when the actual request was that I stop helping with the dishes, and saying, "Let it go," when the actual request was to please stop digging further. There are fewer of these situations when people can commit to telling me things directly as an accessibility request. (Note that this doesn't always work. Sometimes people will say they'll be direct and still be indirect because allistic habits die hard.)

Another communication thing that I mentioned in my initial blog post was having days or times when words are hard, or when speaking is hard. When I can understand and frame this as an autistic need, it makes it easier for me to ask friends whether we can do texting hangouts instead of a voice call, or a voice call instead of a video call, for example, because it would require less energy.

In terms of how it helps my own personal understanding of things, most of it has to do with letting myself just exist and allowing myself to do the things I want and need to do to give my body comfort. I have gotten much more comfortable stimming, especially when I'm on my own but now even when I'm in the presence of other people but not "performing" attention. I will play with my fountain pen cap and shake my legs and rock and bounce in my chair and do whatever else my body just wants to do. I will let myself eat my samefoods without judgement or anxiety about how it appears to other people or how it's "weird." I can respect the times that I go into focus-mode about birds and don't want to talk to anyone, without judging it or thinking that I'm being rude. Of course, there's a balance to some of these things especially when they involve other people who want attention and time from you, but here the key has been having friends who are willing to work with me collaboratively on solutions that don't require me to be a different neurotype.

In addition to things that increase my general comfort, it is also very helpful to understand that what I sometimes experience if I go too hard or if shit happens has a name - sensory overload and meltdowns. I've been learning about the things that give me comfort in these situations, giving myself permission to do these things my body needs when I get into this state, and also to try and understand what some of the triggers are so that I can reorganize my life so that I'm not consistently experiencing autistic burnout.

Q. What are your thoughts about the whole spectrum thing?

The way I understand and employ the word spectrum is that there's a wide range of ways in which autism surfaces in different people and there's no one way to be autistic. A quote I like is, "If you've met one autistic person, you've met one autistic person."

Fundamentally, since it's a processing disorder, there are different things we might process differently. This shows up in my own autistic friends, for example. My thing is sound - the thing that's always affected me and is usually the first danger sign that I might be approaching a state of sensory overload. An autistic friend who once visited me was shocked at how much they saw sound affecting me on a day-to-day basis because it doesn't bother them as much. Another friend's main thing they talk about a lot is sensory defensiveness, whereas I don't normally experience that. I'm a mostly speaking autist but I know semi-speaking, unreliably speaking, and non-speaking autists. I have two friends who need very specific textures on their clothes, whereas I don't pay much attention to this. Some of my autistic friends are wildly sarcastic and deadpan all the time, and meanwhile I struggle with all non-literal language and it's quite obvious to people who've spent a bit of time with me. Some of them have stims I find unbearable, and some of them probably have issues with mine!

All this to say basically that even though I feel significantly more comfortable with autistic people as a group than allistic people (with some individual exceptions on this side), there is a lot of diversity in what that actually looks like - it's not all the same.

Q. What are your thoughts on the fact that we use the same word for people who with a particular cognitive style and people who have a very severe handicap?

So one of the frustrating things about this way of framing autism is that it creates a dichotomy of "high-functioning" and "low-functioning" autism. I say dichotomy, but it's not always presented as a binary - sometimes people try to propose different functioning levels. There are a bunch of problems with this.

One of the more practical issues with this idea of classification is that it acts like every autistic person is sort of static in their level all the time, whereas the reality is more like: if you met me on a good day when I wasn't stressed in an environment that wasn't overstimulating for me and I masked for you, I would pass as normative, allistic and neurotypical. Whereas if you saw me in meltdown mode rocking under my covers for 3 hours, wanting to be in a vat of acid to melt my skin off so I didn't have to feel it on fire anymore, feeling like rain on my skylights was like daggers repeatedly stabbing my eardrums, and unable to string a sentence together, you'd put me somewhere completely different on the scale.

People cleverer than me have better critiques of functioning labels than I do which you should read about if you're interested but basically it's counterproductive to focus on "levels" and classification and it's more helpful to focuses on the similarities (different neurotype, different way of processing).

More importantly, though, this sort of separation stems from the ableist (and capitalist!) idea of usefulness to society - like, oh, if you're a maths genius type then you're a Good Autist, and if you can't speak then you're a Bad Autist who can't be a productive worker, and all of this is genuinely just a bunch of ableist, eugenicist drivel that is plain awful, beyond being counterproductive. There is no collective liberation to be found in siloing ourselves like this.

Q. What are your reflections on internalized ableism around autism?

Oh gosh, where do I start with this. Society is ableist about autism and yes I have internalized many of these ideas from when I was very young, and we all have, and it is a constant, ongoing process to fight against it. It is not easy. I spend a lot of time thinking about this.

I know that wearing headphones to the grocery store significantly reduces the amount of energy I expend on a shopping trip, but I have a spent a lot of time agonizing over the fact that the grocery clerks probably think I'm being rude, even though I can hear them just fine. My headphones just dull the background noise - it's not like I'm listening to music when I should be paying the cashier. And in fairness, if the clerks knew I was autistic, they would probably (hopefully?) be nice about it, but they don't and it is almost definitely not the first thing that comes to people's minds when they see me. So this type of thing is a recurring conflict in my head - where I try hard to prioritize my needs in a society that doesn't necessarily know I have those needs and that may also be unnecessarily nasty about my actions (regardless of whether they know).

I've also struggled with internalized negativity about having samefoods that give me comfort because people have been weird about it before. Allistic society expects you and wants you to eat a variety of things, and when other people can see what I'm eating I can get cagey about samefoods or try to pretend I don't have them. I even tried to get on a plan with meal prep kits to force myself to increase the variety of things I was eating, but even there I had like one or two recipes that I would want over and over again and nothing else would make me actually happy. This is basically like doing ABA to myself because it's depriving myself of comfort to meet a mostly arbitrary allistic aesthetic standard. It's not good.

Of course, this means that my mental response has been to try to swing hard in the opposite direction, i.e., tell myself I should just be as unapologetically autistic as my body wants and "just stop masking," etc. But it is important to remember that masking literally happens as a safety response and is not really a conscious decision as far as I've read. So it doesn't actually make sense to penalize myself for masking either. It's complicated and contextual, as most things are.

Mostly I try to just... be. I try to be more aware of the things my body wants to do and likes to do and needs to do, and I try to respect that as much as I can, and I try to talk to people about my autism so that I can minimize the gap between what it looks like when I'm alone and feel safe, and what it looks like when I'm around those people.

Q. What is your experience of knowing when (not) to bring it up?

This is a tough question. I think in general I don't bring it up as much as I could with people in my school/work life because I feel like I have to explain many things since people have this preconceived notion of what autism looks like that is incorrect and based in stereotypes that I don't match. I worry about facing microaggressions or disbelief if I bring it up, because that has happened before, so I don't really bring it up unless I have a baseline level of trust and I'm prepared to talk about it in some detail.

I do talk about it a lot with autistic friends and I like talking about it on the internet and writing about it. Also it's a pretty important part of my therapy sessions with both my therapists and I consider it absolutely critical to communicate about in close personal relationships, particularly relationships where the other person is allistic and I have to explain that directness and confrontation are required accommodations for me because I am not able to read minds or read between the lines like they sometimes want me to.

Q. Do you know when you're masking?

I don't usually notice while I'm doing it but I will sometimes notice after the fact based on how tired a conversation leaves me. If I needed to put on a face and a voice the whole time then I feel tired at the end of it. This is why it takes fewer spoons to hang out with people who know I'm autistic and the least spoons to hang out with people who are autistic themselves.

Sometimes I can tell I'm masking when I'm having a difficult conversation with someone and I feel the presence of several layers of linguistic filters before I say anything. If I'm having a difficult conversation and I feel less filtered and safe being honest and direct, then I'm probably not masking.

Q. Do you experience issues with spontaneity?

So this is like the oldest complaint that my family ever had about me, ever since I was a little kid. When I was a kid I'd properly meltdown if plans changed and it has just been drilled into my head so hard that I am bad at dealing with spontaneity / changes in plans that I am extremely self-conscious about it and overcompensate quite hard in the opposite direction.

My strategy to do this to imagine every possible outcome and make my peace with it happening so that if it does actually happen then I've tried to front-load some of the processing. Some people call this anxiety. I don't (although anxiety is very often comorbid with autism).

For me, this habit of "imagining every possible outcome, especially the bad ones" is not something that interferes with my life as much as it is an autistic coping mechanism in an allistic world filled with randomness. It helps rather than hinders me, so even though it may share features with anxiety on paper, I feel uncomfortable thinking of it as a problem or a disorder.

But yes, I do love routine and fixed schedules and whenever I can point to periods of my life when I had that and it was stable, those are generally very good times.

Q. What is your experience of special interests, and why are they any different than just passions?

I think there are two things I can point to that are classifiable as special interests. When I was a kid it was cars, and for the last few years it's been birds.

My car obsession was never identified as a special interest but it looked very clearly like one, in the most classical textbook sense: I would beg my parents to buy me car magazines and I knew so much about them and when we'd drive around town I would point at cars and identify them, right down to minute details about models. Sometimes I wonder if this would've been picked up as an autism thing if I'd been a boy. I'm not interested in cars anymore, but I think everyone who knows me knows that my "thing" is birds.

I'm not entirely sure how one empirically differentiates a special interest from a passion or a regular interest (and would also ask if that's actually necessary), but I know I don't have other passions that are as intense or as consistent. Like, someone once asked me if I was into space and I actually hesitated and botched my response because in my mind all I could think was, "Well, sure, I like space, but I don't like space nearly as much as I like birds." And mind you, I've had star map apps on my phone and I've reported a fireball sighting and I've invested in space-related Kickstarter projects and I sometimes stay up to look for the International Space Station and I have a space-inspired tattoo and custom art! So for me to hesitate to even say whether I liked it, well, I think that's what lets me classify birds as my special interest.

Also, sometimes I get into this deep focus mode where I want everyone to just leave me alone with my birds - sometimes it looks like playing with my Wingspan cards or my German bird flashcards or doing deep-dives on Wikipedia or making lists of bird facts to share with the world or editing my bird photographs. And I get irritated when anything else bothers me, like I get annoyed even if I have to pee! Or I don't notice I need to pee, or I notice it but then because it annoys me, I ignore it and then forget. This is... not ideal.

Despite all this, I do still sometimes question whether birds are my special interest or just an intense passion because I'm not so passionate that I literally go birding every day at 5 in the morning - I like sleep and I have other things to do. But here I would go back to the question I posed earlier about whether it is necessary to delineate a special interest from an intense passion at all.

Q. What is your experience of sensory stuff and strategies that help?

There are various levels of sensory symptoms that I experience. I find I notice the sensory stuff more the more ambiently stressed I am. The more stressed I am or the more I'm not pre-emptively taking time to recover from any kind of interaction or outing, especially when they involve masking, that's when I start hearing the fridge and the pipes and the lights in an unbearable way.

This level of sensory stuff tends to be somewhat manageable with specific strategies, but if it turns into a full blown meltdown then it's pretty nasty. Meltdown for me feels like every sound is daggers stabbing my eardrums. Sometimes lights feel too bright. I usually lose my ability to speak, and at least once I have even lost my ability to read anything but the simplest, shortest sentences. If it's really really bad, my skin will feel like it's on fire and I will notice everything touching it everywhere. Usually I just rock or curl up under my weighted blanket and wait it out, but if this happens I lose a few hours of my day and feel like crap so I try to avoid getting to this point as much as I can.

In terms of strategies against sensory stuff, since sound is my big thing I chose to live somewhere where sound wouldn't be a regular issue for me - unlike, e.g., the city centre. I also have excellent noise cancelling headphones and I recently bought earplugs as well that don't block out all noise but have a calming effect on me. Also one of my best investments since moving to Germany was to buy a weighted blanket because I find when I'm feeling sensory or just generally stressed out for any reason, it makes me feel noticeably calmer to have my body be gently crushed.

I also have strategies to avoid getting to meltdown. I have been trying to be more intentional about mitigating stress and exposure to sensory stuff, e.g., by always wearing headphones to the grocery store, by spacing out my social commitments, by putting non-speaking days on my calendar for general recovery, and just trying to mitigate the stressors in my life from school and work as much as possible. Another thing is giving myself permission to do the autistic things that help with managing stress, e.g., giving myself permission to eat samefoods when I'm stressed out, allowing myself to stim more in class and on calls and even outside of that, etc.

I've had a lot of internalized baggage about this sort of stuff for a long time and allistic society's messaging about all of this doesn't help either, but this is where listening to autistic people and respecting my autistic body's needs is important and ongoing work.

Conclusion

If you just read this, suspect you're autistic, and want to chat, I'm available via email. I'm also always just happy to hear when things resonate with folks, particularly about autism, so please tell me if this does!

This post is mostly full of practical advice, experience and autistic self-love and self-acceptance, but I have longer thoughts for another day about autism and trauma - past and present, and how our experiences of the world as autists affect not just our autistic-allistic relationships but even autistic-autistic ones. So if you want to read that, tell me (it'll be good writing motivation!) and subscribe to my RSS feed, nerds!

Code of Compliments

2022-05-23T00:00:00Z

This is a list of my own personal thoughts and habits when I compliment people but I figure they're also useful for other people to know the mental frame through which I interpret their compliments.

No trivial "favourite" compliments

Anyone who has ever tried telling me that I'm their favourite Vagrant knows how this conversation goes. I ask them how many Vagrants they know, they say just me, and then I ask them if they realize that that trivially means that I am both their most and least favourite Vagrant and this is not a meaningful compliment. (Yes I was Like This even before I took any classes in logic).

Some friends recently taught me that the pragmatic function of this type of compliment is still as a positive compliment that means they like and appreciate me. One friend said, "It means I want to tell you I like you so much you're my favorite, yet I don't really want to rank people so I'm making the category irrelevant. So I get both the pleasure of saying you're my favorite with none of the ranking implications."

Nevertheless I tend not to give these types of compliments - not just limited to names but to any case where the person is the only one in a group and the compliment relies on their being my favourite in that group, e.g., I wouldn't call you my favourite entomologist if you're the only entomologist I know. But this conversation was useful and maybe that will change my habits going forward. It will certainly help me receive this type of compliment from other people in a less insufferable way in future, which is good.
No "favourite" compliments that rely on ranking

I agree with what my friend said from earlier about not wanting to rank friends and put some down. I do have levels of closeness of friends, of course, and the bounds are sometimes fuzzy, but in general within a level, I do not like ranking. This means I have multiple best friends, even if that pisses pedants off. In fact, I imagine this is a fairly common case of the pragmatic use of superlatives being different from the semantic definition.

So this means I can compliment someone by telling them they're one of my best friends or one of my favourite linguists, and I can even estimate or exactly state the size of those groups, but I refuse to rank within them.

I learned the hard way that the way this is phrased can have unfortunate pragmatic consequences. I thought I was being clever and diplomatic in separately texting three friends that they were one of my top 3 favourite goths on the planet. What I meant was that I know a lot of goths but I have three favourites (unranked). But my phrasing had the unfortunate consequence of making someone assume they were #3 because, well, why would I say top 3 unless they were the worst of the top 3, I'd just say #1 if they were #1.

This is a sensible implication to draw even though it didn't occur to me before. It's sort of like in sports: if someone is ranked 1, 2, 3, 4 or 5, you'd say they're top 5. It's trivially true that they are top 10 also, but you usually only hear "top 10" in reference to someone who's ranked 6, 7, 8, 9 or 10. Grice is probably rolling in his grave because of how long it's taken me to figure this out.
No unexplained compliments

This tendency comes from a preference about receiving compliments: I prefer compliments that come with reasons because otherwise I feel a bit like I've lost the plot. Sometimes it's obvious from context, e.g., if my advisor is talking about a thing I just did or if a text compliment is in response to something I just said. But I have recently noticed a tendency to be surprised and confused by more generic or unprompted compliments and in these cases it's best when they come with an explanation so I know where they're coming from and don't confuse them with flattery.
No flattery

I prefer not to give disingenuous compliments for any reason, including and especially personal gain. Sometimes returning compliments is part of a social script, e.g., when a stranger compliments your shoes at a party, general wisdom says you're supposed to compliment something of theirs back. I will only do this if I can actually quickly think of something I want to compliment them about, otherwise I'll just say thanks and maybe tell them where the shoes came from.
No compliments that stem from or reinforce -isms and -phobias

This means that it is not a compliment to tell someone they're thin, and it is not a compliment to call a disabled person inspiring, and just please never ever ever call me articulate because it is just too loaded a word given where I come from and the colour of my skin.

I have tried to stop telling people they're smart as a compliment (with some success) because of how it reinforces this idea of smart = better, which is associated with plenty of things I don't like including saneism, eugenics and capitalism. I sometimes think about how complimenting people on their appearance reinforces the idea that beauty matters and is linked to a person's worth, but frankly I enjoy calling people cute too much and don't see myself stopping.

So what's left?

My rules are a list of don'ts. The big do is basically: compliment people generously, but genuinely. In terms of content, I tend to compliment qualities I value in people or in their actions or in their work (e.g., thoughtfulness, openness to new information, clarity in writing, etc.). They're also often informed by the things I see in other people that I want to be better at.

My friend L is my gold standard for compliment-giving and generosity. She has taught me among other things that compliments and gratitudes are very closely intertwined, so you'll find messages that blur the boundaries of the two in this curated selection of good and representative compliments from our texts (included with her consent):

"Look at all this personal growth and self-knowledge! Love it for you!"
"Thank you for being my friend"
"Omg this [art] is beautiful how are you so talented"
"You bring so much light, knowledge and care to your relationships (and also puns)"
"It makes you look like a painting and I am mesmerized"
"This made me think of you. I feel like you've always been so dedicated to cultivating access intimacy with me, it was one of the first things that struck me about you. You helped me create/share an experience before I even had the words for it."
"I am so grateful to have you in my life (apropos of nothing in particular)"
"I just got your mail - you're so fantastic and I just want to yell about it into the void"
"This aesthetic suits you so well, it's like I'm watching you bloom before my eyes. Every look you crank out is even more powerful than the one before."
"You are genuinely my own personal gay icon"
"That's just how you operate. Doing immensely courageous stuff while somehow remaining composed and acting like it's no big deal"
"You are so good to me"
"That took so much courage and strength, and I am SO PROUD OF YOU. SERIOUSLY."
"I will never shut up about how awesome you are"
"Thank you for sharing this with me. It's incredible. You're incredible. Please never stop writing."
"You are so weird and funny and I love you"
"That is a good gender"
"How does it feel to be so incredibly attractive and photogenic?? What's that like??"
"You are one of my favourite goths in the whole world"

If you ranked these, hers would definitely be the better ones, but I am trying, I swear! Inspect the HTML if you want to know who said what.

Switch by Siouxsie and the Banshees is a trans anthem

2022-06-21T00:00:00Z

Siouxsie and the Banshees are among the original goth rock bands that kickstarted the genre as an emergent and distinct one from post-punk. Their 1978 debut album, The Scream, originally featured Switch as its final song. Music reviewer Kris Needs writes that Switch is about "different people who swop jobs with terrible results - scientists, GP and Vicar" (Needs 1978). This was confirmed by the band members themselves, with Siouxsie Sioux herself quoted in a Melody Maker interview as saying, "The vicar becomes the scientist and he applies his religion to science and the doctor becomes the religious person and applies his medicine to religion" (Birch 1978).

I argue instead that Switch contains trans themes, is an anti-disestablishmentarian war cry against the scientists, doctors and vicars who police the bounds of transness and gatekeep transition, and that this song should be re-analyzed and revived as a goth trans anthem.

Transness and transition

Switch opens by depicting transness through incongruity and miscategorization, and transition through metaphors of both divergence and fusion. These metaphors have clear parallels to transness but simultaneously, the ambiguousness of the specific terms used allows us to read multiple narratives or stages of transness in the first verses, that support a pluralistic view of trans existences.

Different lives In different places Familiar problems Same old faces

"Different lives in different places" speaks to the sometimes fractured nature of living while trans - having contexts of safety, community and deep understanding, versus... everywhere else. The beauty of this specific wording is that it applies to several ways of being "out" as trans or visibly trans: even the most flagrantly gender-noncompliant of us feel and perform and engage differently in different communities and societal contexts depending on how our gender-noncompliance is read. Extending the idea of "place" beyond physical locations, even those trans folks who are out to no one and have different lives externally versus internally (in the "location" of the bodymind), they are also included in this metaphor.

Shuffle lives Into wrong categories

Here in the words "wrong categories," we see the most explicit reference to misgendering or otherwise being constrained by society's boxes. This of course does not refer simply to gender labels, but rather all the consequences of how Gender as an institution categorizes and confines, including the ways our dressing and communicating and interests are often policed.

Cross the wires And fuse humanities

These two lines represent transition, with "crossing wires" indicating a sort of divergence or mixing up, and "fusing humanities" as a sense of combination - broad ways of describing transition without necessarily limiting it to medical aspects, and in fact providing us with a much more accurate metaphor for what transition actually looks like.

Without specifically referencing anything like dysphoria or the "wrong body" narrative, we do not see Switch replicate these transmedicalist requirements, while still leaving room for bodily dysphoria as a potential feature of people's transness. Similarly, the expansive range of interpretations we can read into abstractions about "crossing wires" and "fusing humanities" means that transition can be conceived of and constructed in a myriad of creative ways, through transformative deletions, additions, woven patchworks, blended cocktails, or however else we define it.

Gatekeepers and their hypocrisy

The next three verses describe the switched professions of scientist, general practitioner, and vicar, as indicated in the band's explanation of what the song was about. All three are characters who hold power in mediating various aspects of transition and trans acceptance and inclusion in society: scientists study transness and trans people, GPs are the gatekeepers of various aspects of transition (medical, first and foremost, but often even aspects of social transition), and vicars uphold and correct perceived deviations from values they uphold as Christian (which queerness and specifically transness have historically been - and in many sects still are - in conflict with).

The descriptions of the characters reveal their underlying biases and ways of viewing the world. I take these characters as representatives of the larger institutions of Science, Medicine and Christianity - institutions that, similarly, represent and replicate cisnormative attitudes towards transness and trans people.

Scientist G.P.'s With patient guinea pigs Curing their headaches With drastic side-effects With drastic side-effects

The scientist in the role of the GP sees patients who are compared to "guinea pigs," an apt metaphor for the way trans science has historically viewed trans people as freaks of nature. It is a cruel irony that transphobes today compare trans surgeries to lobotomies, saying that they will soon be a thing of the past and that we will be horrified we ever did them, when lobotomies actually used to be used as a "treatment" to "cure" transness. Conversion therapy and electroshock therapy, too. The scientist here in Switch reduces trans human beings to a lower level, to animals, and even causes them to have "drastic side-effects."

Doctor rectorates Condescending from on high For all hallucinators See druggist in the sky In the sky

The doctor plays the role of a vicar/rector and is described as "condescending from on high," as doctors do even in matters of reporting on our own bodies - as numerous trans, disabled, chronically ill, Black folks and women can attest. The doctor's view of believers as hallucinators and their god simply as a druggist further emphasizes the inability to take someone else's point of view about something without condescension or disbelief.

Vicar experiments But 'tis blasphemy Dismissing thought of progress As the mark of devilry As the mark of devilry

The vicar in the role of the scientist cannot experiment without considering it "blasphemy." Any thoughts of progress are simply dismissed as "the mark of devilry," which I analyze as many Christian sects' conception that gender deviance is against god. The vicar and indeed the larger institution of Christianity view transness, in its violation of assigned gender roles, as anti-Christian and therefore blasphemous.

In their 1978 interview about The Scream, Steven Severin and Siouxsie Sioux declare that the song Switch is about "the hypocrisy of it all" (Birch 1978). About the characters, the band members say: "These people, whatever the given situation, they will apply their same standards to it and won't change." Here their explanation lines up nicely with the trans angle I am arguing; the overwhelmingly cis gatekeepers of trans care and acceptance apply their cis standards to transness - the scientists who view us as strange guinea pigs, the doctors who condescend to us, and the vicars who conceive of our existences as deviant and unacceptable. They will apply their same cisnormative standards to us and, lacking the same epistemological frame of identity, find us failing. These lyrics point to the flaws and gatekeeping of these individuals (and the institutions they represent), and the fact that they will not change. I read this as an urgent call for trans solidarity and for us to take care of each other when no one else will.

Beyond transmedicalism

The discussion of these institutions reveals their role in the ideological construction of gender and in enforcing the bounds of these constructs in the name of societal order, and examining their existence is important in the context of current trans realities for many reasons: the legitimacy they grant or take away, the transmedicalist definitions and requirements of transness, and the actual resources trans people are prevented from accessing because of this larger system of biopolitics.

When I argue that these verses are a call to action for trans solidarity there is a broad range of actions and viewpoints that this could include. I do not see the mere discussion of institutions in these lyrics as incompatible with the transhumanist, anti-transmedicalist, anarchist vision of transness that I currently prefer. Abigail Thorn sums up my point of view nicely in her essay on why she dislikes the word dysphoria:

I would prefer to centre desire and will when talking about transition. To me, transition is more than meeting a medical need, just like crossing the Rubicon was more than Caesar getting his feet wet. It's taking your life in your hands and shaping it yourself. I think that's beautiful. I didn't transition to "alleviate my dysphoria," I transitioned because I fucking wanted to. Who is the state, or a doctor, to tell me I can't?

Since there is nothing in my interpretation of the lyrics that imply dysphoria as a hard requirement for transness, there is room for transition stories like mine and Thorn's.

Violence, death, desire

The final part of the song highlights the real societal consequences of structural and institutional transphobia and the lack of trans care - violence and death. It ends on a high note by reinforcing our desire as trans people to transition and to do transness.

People walk And even talk People listen Then they halt Something blows up Won't come down

I analyze the "people" in the first line as referring to trans people and the "people" in the third line as referring to cis people. The trans people's "walk" and "talk" is viewed by cis people and it gives them pause. There is an act of perception here which mediates how we can analyze the words walk and talk in this context. The trans people's "talk" could refer to literal communication of their transness using language, but the combination of "walk" and "talk" could also refer to a broader set of visual and linguistic cues (certainly not limited to spoken languages or ambulatory movement). These are read (clocked) by the cis people as being deviant or deficient for a particular gender, making them bad gender subjects.

The next two lines state that something "blows up" and won't come down, which I take as implying violence from the cis people towards the trans people. This violence could be on a spectrum from an inability to accept someone's transness all the way to the very literal explosion of a bomb or a gun. This is not an exaggeration, as violence against trans people happens at alarming rates around the world - overwhelmingly to transfeminine people, sex workers, migrants, trans people of colour.

Scattered muscles twitch Too late to switch They're dying to switch

The idea that it is "too late" to switch leads me to interpret the following line's "dying" to switch as literally implying death. This would line up with many stories from trans folks in the UK who are dying in the face of wait times of several years before gender clinic appointments. Goth culture is often characterized by a fascination with death and mortality. These themes are discussed openly and there is less of a fear of confronting this reality. This is why, within the context of Switch as a song by a goth rock band, I am comfortable reading a literal implication of death into these final lyrics rather than simply an idiomatic implication of excitement and desire.

The song closes with repetitions of the phrases "They're dying to switch," and "Switch," which I analyze as the latter sense - desire and excitement to transition, to embrace and celebrate transness, not as a destination but as a continuous journey of making and doing our genders.

The expansive potential of switching

The word "switch" needn't be analyzed as a binary swapping of gender in the way a lot of transmedicalist writing and systems still do. I analyze it instead as "a shift from one to another" or "a change from the usual," as Merriam-Webster defines it.

Nothing in the lyrics indicate a binary switch - Sioux sings of "different" not "opposite" lives, and of "wrong" categories, not "the wrong" category. Dying for "a brand new switch" includes expansive possibilities of gender and transition that include everything from man to monster, and even dynamically changing genders. Goth culture embraces the monstrous rather than relegating it to the sidelines, making it about the only cultural niche where my personal gendergoal - to be a gender-treacherous trans abomination - fits quite naturally alongside others.

There are additionally no requirements that imply that medical transition is a requirement for transness as I analyze it in these lyrics. Siouxsie and the Banshees discuss "lives" and "humanities" rather than bodies or organs - lives are "different" and "shuffled," and humanities are "fused." This leaves much room for flexible interpretation - everything from social transition and presentation to surgery could come under this, including other forms of transhumanist body modifications.

Conclusion

In sum, Switch is a trans goth anthem that tackles the serious issues of medical and societal gatekeeping and prejudice, directly confronts the serious consequences of this on trans people, all the while allowing for a range of possible meanings and manifestations of transness, and arguing for liberatory, transhumanist, powerful queer futures.

Dedications and disclaimers

This is an elaborate shitpost dedicated to J Calder, my new personal trans goth academic idol.

Here are my disclaimers:

It's entirely possible someone has made this connection before but a cursory search didn't turn anything up and I desperately wanted this to exist so I wrote it.
Goth still isn't a label I use for myself, just one that others call me. This may shift in future, though, particularly as I ponder something amusing J told me: that it was very goth of me not to ID as goth.
I made the choice to have the character of the vicar here represent Christianity as an institution and not simply Religion as a whole, because not all religions have as much of a fraught history with accepting transness as Christianity does, nor are other religions as instrumental even today in promoting moral panics about gender (I am thinking specifically of the Christian right). My wording choices around this are most likely to be where I'm missing nuance since these are not histories I am personally familiar with. This is why I had Maureen Kosse be my sensitivity reader for Christianity, as a queer ex-Catholic who studies gender, language and fascism. I welcome further feedback/criticism about this from people who have my phone number and/or have ever had a meaningful one-on-one meeting with me.

References

Needs, K. (1978) "Siouxsie and the Banshees: The Scream". ZigZag.
Birch, I. (1978) "Scream and Scream Again". Melody Maker.
Sioux, S. (1978) "Switch" [Recorded by Siouxsie and the Banshees]. On The Scream.

In perilous pursuit of pelicans

2022-07-03T00:00:00Z

It all started because I wanted to see pelicans.

The only American white pelican nesting colony in British Columbia is Stum Lake in White Pelican Provincial Park. The lake is closed from March to August to protect the pelicans so I decided to go in the first week of September to see if I could spot any stragglers who hadn't flown east yet.

This was going to be a bit of a road trip because Stum Lake was over 600 kilometres away from where I lived in Vancouver. When I looked up the route on Google Maps it looked like there were some forest service roads on the route which in my experience are often unpaved. So considering that I had a small car and there was going to be some offroading involved in getting there, I decided I would play it by ear and only do what I was comfortable with - even if that meant giving up without seeing pelicans.

I got off a paved road and onto a dirt road at 1512 kilometres on the odometer.

If you've never been in a small car on a dirt road - it's very, very loud. It's like a loud, deep rumble that you feel with your whole body because it's bumpy. You feel like you're being shaken alive. You hear little stones clanking against the sides and the bottom of your car.

I saw a little black bear on the side of the road that hurried into the forest when I zoomed by. A couple of times I saw families of deer crossing the road in front of me. I also stopped a few times to look for birds.

At some point, Google made its first mistake. It made me turn right onto this narrow two-track road in the middle of the forest. When done recreationally, driving on two-track roads is also called greenlaning because in between the two tracks that are made by tires, weeds and plants grow freely because they aren't run over by tires or feet. A "green lane."

I went down this path for 3 kilometres all the while hoping that I wouldn't run into another car going the other way because the path had narrowed quite a bit and there was no real way to turn. It had also rained all morning and so there were several muddy puddles along the path. Every time I went over them I just hoped against hope that they weren't deeper than they looked and that I wasn't going to get stuck.

Then I reached a fence. This would have been exciting if I wasn't still something like 50 kilometres away from the park. There was no choice but to go back because this route was a dead end. Like I said earlier, the path had narrowed too much to be able to turn without hitting trees, so I ended up having to back up for 2 kilometres before I could make a turn. It was stressful. At this point I could've chosen to give up and turn around but I just laughed it off, chalked it up to adventure, and continued on a new route that Google gave me - back on the dirt road I'd turned off of.

The road with a real name turned into a forest service road which later turned into a nameless road. I had 43 kilometres to go and I also noticed that I'd lost cell service.

Really, it would've made a lot of sense to go back at this point, but the pull of the pelicans was too strong so I kept going. The scenery changed. The trees became shorter and the terrain became much more hilly. Almost my whole drive I'd been seeing signs warning me to watch for logging trucks, but now it was a lot visually clearer that I was in logging country - in the distance I could see clearings. It was a Saturday so there were no logging trucks around. In fact, I saw no one at all. No trucks, no cars, no humans, and I'd been driving for quite a while already.

Eventually I got to a fork in the road where Google said I could take either fork and they both had a similar ETA because they joined up again later. Both paths were two-track roads like the one I went on earlier, except these were gravel roads and a lot rockier. I picked the left one and started down it. It was quite beautiful - I was on a mountain covered with short shrubby foliage so I could see down into the valley, and the sky was a gorgeous blue punctuated by fluffy cumulus clouds.

Unfortunately, this was the beginning of the end. As I went on, the path got more and more challenging. It narrowed and got rockier, the green lane in the middle had taller plants, and there started to be big height differentials between the two tracks that my wheels were supposed to go on. The sound of plants tickling the belly of your car is a bizarre sound. I thought I wouldn't be able to hear it, that it would be like feathers on metal. But no, you can actually hear it - it sounds almost like soft kitty claws on metal. When one side of your car is much higher than the other, it really feels like you're going to topple over. Luckily I didn't, but it started to feel sketchy. I wasn't going anywhere near the 60 km/h that Google expected me to (in order to cover 43 kilometres in 44 minutes). I was going somewhere between 5 and 20 km/h - anything faster in that car would've been a terrible idea.

Then the path itself started to incline up and down more violently. I did 2 downhill runs where I found myself literally holding my breath as I held the brake and eased the car down. I'd been trying this experiment for 70 kilometres already and at this point, I decided that enough was enough and I should cut my losses and go back. To turn around, I had to execute a careful 10-point turn on this narrow dirt path so that I didn't fall into the pits that were on either side. When I tried to climb back up those last couple of downhill runs, my tires struggled to find traction and started spinning. They pulled through but after I was on a more level part of the path, I stopped to change the destination on my Google Maps app. When I looked at the map, it showed me way off any marked path, in the middle of nowhere. I'd followed all its directions and I'd seen no other paths on the way that I could turn onto. So this made me panic.

This was when it really hit me that this had been a bad idea. I was out alone in the middle of nowhere with no way to describe where I was to anyone - not that I could even reach anyone because remember, I had no cell service! I theoretically know how to change a car tire but I've never actually had to do it. And what would happen if rocks damaged the underside of the car and all the fuel leaked out and I was stuck there?

I just wanted to go home.

Eventually I talked myself down and set up the maps app and began to go back. The app made another mistake - it had me turn into a field that opened out onto a path, and then wanted me to turn onto a path that was blocked by a horizontal log. I gave up on the app and tried to retrace my steps from memory. I saw a tree branch I recognized that told me I was on the right track and I laughed at the absurdity of that. After that gravel nightmare I was utterly relieved to be back on the wide dirt road. I never thought I'd be so grateful to be on a dirt road but going from a terrible road to a bad road will make you feel like that.

So no, I didn't see any pelicans. But the universe was kind enough to grant me a spruce grouse on my way back.

Code crimes: mama and beer

2022-07-04T00:00:00Z

It all started with Simon saying, in classic sarcastic fashion:

I have another amazing fact: If you change only four letters in the word mama, it spells beer!

Now, of course there are plenty of boring ways to do this in Python, but we wanted what we always want: trickery, fuckery, jiggery-pokery.

Disappointingly Kevin couldn't think of any way to hack the str.replace implementation enough such that 'foo'.replace('o', x) returned a different value of x for each char. Obviously I took this as a challenge.

Solution 1

Here's a solution to that problem.

The secret here is wildly cursed monkeypatching with the Forbidden Fruit Python package.

Here's the code. The trick is to "curse" the in-built Python string replace function with my redefined replace function that picks a random letter every time. Yes, it's evil. But it's also easy.

from forbiddenfruit import curse
import string
import random

def replace(self, a, b):
    new_str = ''
    for i in self:
        if i == a:
            new_str += random.choice(string.ascii_letters)
        else:
            new_str += i
    return new_str

curse(str, 'replace', replace)

'mama'.replace('m', 'x')

Solution 2

This is a solution to the original problem. It's less internally devious but it ends up looking kludgier as a result.

As you might guess from the syntax there, yes, it does use classes!

class S1:
    def __init__(self):
        self.x = ['a', 'mama']
    def __get__(self, i, o):
        return self.x.pop(0)

class S2:
    def __init__(self):
        self.x = ['b', 'beer']
    def __get__(self, i, o):
        return self.x.pop(0)

class Box1:
    s = S1()

class Box2:
    s = S2()

my = Box1()
myother = Box2()

print(my.s, myother.s)
'mama'.replace(my.s, myother.s)

It's not very neat or concise code, and it's even brittle because it relies on elements being popped from an array and therefore you have to call the get methods exactly once before doing the replacement for everything to work perfectly. But it's certainly less crimey than the other solutions.

Solution 3

Of course I couldn't stop there. I wanted to write code that would do the original cursed thing, i.e., replace 'mama' with 'beer' by hacking string replace. So I did that!

"In fact," I said, "I'll do ya one better, I'll prove I'm not monkeypatching with forbidden fruit anymore."

People were perplexed. Some replies:

BUTT HOW?! Wait a sec ... did you monkey patch sys.modules? Or string? This is approximately as bad as time travelling and changing history I dunno Jonas — I think it's worse. I think it's reaching in and monkey patching pi to be exactly 3 True, more like actually changing laws of physics

I told them that what I had actually done was sneaky as hell (a sneaky combination of a few things, in fact) and that they were all going to hate me when I told them.

Some more replies: lol it's just manually typed in an editor PHOTOSHOP python3 is aliased to a small program that just prints out the above text. Speaking Latin while waving a wand?

The secret is two things. I'll let you have a look at the code first before I explain (bash first, then Python):

alias python3='python3 -i -c "_b_='\''\b\b\b\bbeer\b'\''"'
python3

print("mama".replace('a', _b_))

What I'm doing here is that I'm secretly aliasing python3 to first execute a variable assignment before opening up the interpreter. I have a variable _b_ and I assign it a string with the word beer in it surrounded by some sneaky backspace control characters.

The second piece of the trick is that I installed FontForge on my computer and created a new custom font that would display underscore characters as single quotes. Then I made iTerm use this custom font.

That's why the second version of the code is the TRUE version, and the first version is almost photoshop, but not quite. Personally I think it's cleverer than that!

Kevin and I discussed it for an inordinate amount of time before I arrived at this solution. He gets partial credit for reminding me that control characters exist. I had initially wanted to defeat the lexer and make it allow me to assign to a variable called 'b' but I couldn't do it. I forget why (this is from December 2020, i.e., Too Long Ago).

That's all! Happy hacking!

Dr. Ian Malcolm was right

2022-07-31T00:00:00Z

Dr. Ian Malcolm: Gee, the lack of humility that's being displayed here, uh... staggers me.

Donald Gennaro: Well thank you, Dr. Malcolm, but I think things are a little bit different than you and I had feared...

Dr. Malcolm: Yeah, I know. They're a lot worse.

Donald Gennaro: Now, wait a second now, we haven't even seen the model output...

John Hammond: No, no, Donald, Donald, Donald... let him talk. There's no reason... I want to hear every viewpoint, I really do.

Dr. Malcolm: Don't you see the danger, John, inherent in what you're doing here? Machine learning is the most awesome force the planet's ever seen, but you wield it like a kid that's found his dad's gun.

Donald Gennaro: It's hardly appropriate to start hurling generalizations...

Dr. Malcolm: If I may... Um, I'll tell you the problem with the scientific power that you're using here, it didn't require any discipline to attain it. You read what others had done and you took the next step. You didn't earn the knowledge for yourselves, so you don't take any responsibility for it. You stood on the shoulders of geniuses to accomplish something as fast as you could, and before you even knew what you had, you patented it, and packaged it, and slapped it on a plastic lunchbox, and now [bangs on the table] you're selling it, you wanna sell it. Well...

John Hammond: I don't think you're giving us our due credit. Our scientists have done things which nobody's ever done before...

Dr. Malcolm: Yeah, yeah, but your scientists were so preoccupied with whether or not they could that they didn't stop to think if they should.

Gentle intro to category theory

2022-08-08T00:00:00Z

We take a high-level look at why category theory matters before diving into the definition and then constructing two example categories with dogs and birds. We use zero math along the way, except for some mild name-dropping of fancy math terms when I give you context for why it matters.

Why does category theory matter?

This post is a math-free introduction to category theory, but the people who use and do category theory are generally doing a lot of math. They care less about the dog and bird categories we'll see here, and more about mathematical and abstract concepts, including sets, groups, rings, posets, stuff from algebraic geometry, topology, homology, and lots of other words from subfields of math that I don't know or understand. The definition of a category and various operations you can do on categories provide mathematicians with a system that lets them define and reason about all these other kinds of math, very much like how set theory (if you're familiar with it) provides a foundational system to define and reason about other types of math.

Category theory is also relevant to allied nerds, including some types of computer scientists and some types of linguists (not me on either count): it provides a natural way to think about functional programming, type theory, formal semantics, lambda calculus, etc. We'll revisit some of these ideas at the end to make the connections more concrete.

It's hard to convey the power of category theory in this blog post for two main reasons:

I am only showing you the definition of a category and some intuition for it, not any operations you can do with categories - including concepts such as isomorphisms, functors, natural transformations, limits, colimits, duality, exponentials, products, coproducts, the Yoneda lemma, etc. These operations give you interesting properties and ways of relating a category to itself as well as to others. This is the stuff that really allows you to do all the fancy math.
Category theory is extremely abstract and mathematical (no joke: it's actually been described as abstract nonsense by mathematicians!), and my entire goal in writing this blog post has been to try to explain it in as concretely imaginable a way as I could.
Bonus reason: The post got too long and I had to stop.

Even so, I hope that once I show you what a category is, I can draw the first few connections to related concepts so that if this interests you, you can go deeper into it. So without further ado, let's dive into the definition of a category.

Dog categories

Let's try to build a category with dogs! So clearly I want the objects in my category to be dogs - all the dogs in the entire world of various species and ages and with various names and homes.

What about the arrows? Well, let's first try the older-than arrow. That would mean that there is an arrow from Max to Owie (both dogs) if Max is older than Owie. So now we have some candidate arrows, but we have to check that identity and composition are satisfied.

An identity arrow is an arrow from a dog to itself, so does this relation give us identity arrows for free? If an identity arrow on Max existed, it would imply that Max is older than... Max. This doesn't make much sense because a dog cannot be older than itself. So unfortunately that means that these candidate arrows do not work for us.

But all hope is not lost! A dog is definitely the same age as itself, so we could always use a slightly modified relationship - the older-than-or-same-age-as relationship! Here we definitely have identity arrows because Max is the same age as himself. We also have arrows to other dogs Owie, Cypress, etc., if Max is older than those dogs or if Max is the same age as those dogs.

Now we just have to check for composition. Composition means that if there's an arrow from Max to Owie and an arrow from Owie to Cypress we have to already have an arrow from Max to Cypress as well, by the definition of our arrows. So let's say Max is 13 years old and Owie is 11 years old. Since Max is older than Owie, there's definitely an arrow from Max to Owie. Let's say that we have a dog named Cypress who is also 11 years old. There's an arrow from Owie to Cypress because they are both the same age. Do we know that there's an arrow from Max to Cypress? Yes! Because 13 is older than 11, and therefore there must be an arrow from Max to Cypress as well. This means we have composition (and associativity because of our math hack I alluded to before), which means we just built our first category!

A fun observation about this category is that every single dog in the world will have arrows besides the identity arrow, either going from it to other dogs or from other dogs to it. All dogs are connected. This is because every single dog has to be younger than or older than or the same age as other dogs. The youngest newborn puppy just emerging into the world will have an arrow pointing to it from every other dog because they are all older than it (or the same age, if multiple puppies are being born at exactly the same time in different places around the world). The oldest alive Guinness-world-record-holding dog will have arrows going out from it to every other dog in the world because it's older than all of them. No dog is left arrow-less and isolated in this category.

Bird categories

Let's try building another category. Because I'm me, we have to build a category with bird species as our objects. Note that here, unlike in our previous dog category, we don't have individual birds as objects, but just one object per species of bird, so we'd have one California condor object in our category, as opposed to 537 (or so) objects for each of the California condors that exist globally.

As for arrows, I want to try something interesting - a mating arrow! By this I mean that there would be an arrow between bird species A and B if members of A and B have been known to mate successfully. Hybridization (or inter-species mating) happens sometimes among birds, particularly between gulls, terns, hummingbirds and various waterfowl species. So let's do our usual thing of checking these candidate arrows for identity and composition.

If an identity arrow on the species Turdus migratorius existed, it would imply that members of Turdus migratorius can mate with members of Turdus migratorius. This is definitely true because a species has to reproduce among its members so that it continues to exist as a species. So we have identity arrows!

Now let's check about composition. If western gulls and glaucous-winged gulls can interbreed, and glaucous-winged gulls and herring gulls can interbreed, does that automatically mean that western gulls and herring gulls can interbreed? Well, not quite, unfortunately. Western x glaucous-winged gulls and glaucous-winged x herring gulls are popular and often-spotted hybrids, but I don't know of a single western x herring gull, and it is not clear that that is a possible hybridization. So inter-species mating is not compositional and unfortunately this means that we can't build a category with this arrow definition.

So we have to try a different arrow. This time let's try the same-genus arrow. What this means is that there is an arrow from species A to species B if they are both in the same genus (the next higher level of classification). For example, there would be a arrow from the spectacled thrush to the bare-eyed thrush (and vice versa!) because both are in the Turdus genus.

We have identity arrows because a species is definitely in the same genus as itself. Now let's look at composition. If red-tailed hawks and ferruginous hawks are in the same genus, and ferruginous hawks and rough-legged buzzards are in the same genus, does that mean that red-tailed hawks and rough-legged buzzards are in the same genus? Yes (it's Buteo)! So that means we have a bird category.

One ornithological problem that we've ignored in our operationalization of this category here is that species and genera are not defined statically forever. Sometimes we find new species and new genera, sometimes species get reclassified together or varieties get separated out into distinct species, and sometimes a species gets moved into a different genus based on new DNA information. So really the world of classification is a lot messier than we're making it out to be in this toy example, but that's a slight tangent from category theory.

Something to notice about the structure of this category is that unlike our dog category, our bird category has all sorts of different clusters, including the possibility for a bird species to have no arrows to any others (other than itself; the identity arrow). This would imply that that bird species is the only one in its genus. This is true of many bird species, including for example the collared lory (a fun tongue-twister), the only one in its genus Phigys. On the other hand there would also be larger, fully-connected clusters representing genera which contain more species, e.g., the Turdus genus which has over 80 species classified under it.

Connecting back to computer science and linguistics

Now that we know what categories, objects, arrows, and composition are, let's go back to the terms I'd mentioned earlier from computer science and linguistics and make those connections a bit more explicit in case they are useful reference points for you.

Functions are the fundamental unit of functional programming and you do things by composing different functions together. So category theory gives us a natural way to think about computer programs because we can treat functions as arrows that have composition. Category theory is also relevant to type theorists because a category whose objects are types is just a different way of viewing or defining a type theory, which means that we can then do category theory math on type theory stuff.

Formal semantics cares about compositionality in exactly the same way we do here, which is that we view predicates as functions (the arrows of a category) and subjects as arguments to those functions (the objects of a category). Lambda calculus and type theory are tightly connected, and they are both represented by what are called CCCs (Cartesian Closed Categories) in category theory.

If you're not a mathematician, computer scientist or a linguist, and I haven't managed to convince you of the importance and power of category theory because of my hand-waving the math away, that's okay. Regardless, if there's one fundamental idea I can leave you with, it would be this:

Category theory is all about the arrows. The magic is in the arrows. Category theorists don't care what specific objects they're dealing with, they just care about how you can do compositional operations on them. It's unlike set theory in this way because set theory does care about objects. It's ground-breaking and different specifically because it gives you this arrow-centric way of reasoning about math.

Appendix (warning: math!)

All the examples I present in this blog post are binary relations on sets. If a binary relation is composable with another, then you get associativity for free! This of course does not apply to all sorts of other math, including (as I mentioned previously) subtraction.

Related to this, a bonus fact I learned was that you can do something called demonic composition of binary relations, which isn't associative, but this isn't really composition. I find this term amusing.

Resources

The following list is arranged in order of level of mathematical background assumed, mostly from my personal experience of using these resources. If you're interested in one of the books, I encourage you to try your library or a local bookstore rather than the union-busting warehouse-worker-abusing anti-antitrust panopticon company.

Eugenia Cheng's book Cakes, Custard and Category Theory (alternatively known as How to Bake Pi in some countries; I've heard excellent things about this lay introduction to category theory - I haven't read it yet but I've read other books by her and I think she's a fantastic maths communicator)
Bartosz Milewski's Category Theory videos (not all human-captioned; this is best for programmers, especially those already familiar with a bit of Haskell which is what he uses in all his examples - I find his style of explaining concepts accessible because he connects abstract math to actual code, but he talks too slowly for me so I play these at 2x)
Bartosz Milewski's book Category Theory for Programmers (I've read many of the blog posts that were turned into this book and they're great but the same caveat as before applies - they're better if you have some programming background, especially a bit of Haskell)
Linguistics Using Category Theory blog post (this post discusses linguistics and category theory specifically and has a vibrant reply section discussing lexicalist vs anti-lexicalist interpretations, semantics, NLP etc., and it's part of a larger blog called the n-Category Café that discusses category theory often, does it well, and is famous for it)
Chenchen (Julio) Song's Category Theory blog posts (these posts from a categorical linguist start with an exploration of category theory from a linguistics background and go on to discuss their current research - I don't recommend this as a resource to learn content because it assumes familiarity with category theory and formal lingistics concepts, but I recommend it for observing someone's thoughts as they learn category theory and draw connections between it and various other fields)
Steve Awodey's Category Theory Foundations videos (not human-captioned; I found the pace a little too fast for me, also perhaps because more mathematical background is assumed than I have)
The nLab's category theory wiki content (this wiki is a well-known resource for concepts related to category theory, homotopy theory, and type theory, and also draws connections to philosophy and physics)
Steve Awodey's book Category Theory (this book brands itself as being for non-mathematicians - as opposed to the classic Saunders Mac Lane textbook - but I've read this textbook as a non-mathematician and I found it hard)
Saunders Mac Lane's book Categories for the Working Mathematician (this is the original, classical category theory book, a textbook for mathematics grad students, written by one of the founders of category theory - I've read little excerpts of this textbook and can confirm that it's very mathematical and formal and not hand-wavey at all)

What I read in August

2022-08-31T00:00:00Z

For fun and edification

Books

McPherron, P., & Ramanathan, V. (2011). Language, body, and health.
Yergeau, M. (2017). Authoring autism: on rhetoric and neurological queerness.
McRuer, R. (2006). Crip theory: cultural signs of queerness and disability.
Feinberg, L. (1993). Stone butch blues.

Papers

Irvine, J. T., & Gal, S. (2000). Language Ideology and Linguistic Differentiation.
Cuboniks, L. (2015). Xenofeminism: A Politics for Alienation.
Toze, M. (2019). Developing a critical trans gerontology.
Bucholtz, M., & Hall, K. (2016). Embodied sociolinguistics.
Ahearn, L. M. (2012). Language, Thought, and Culture.
Schoenebeck, S., & Blackwell, L. (2021). Reimagining Social Media Governance: Harm, Accountability, and Repair.
Heilker, P., & Yergeau, M. (2011). Autism and Rhetoric.
Konnelly, L. (2021). Nuance and normativity in trans linguistic research.
Andrus, J. (2012). Language ideology, fractal recursivity, and discursive agency in the legal construction of linguistic evidence.
Bauer, H. (2010). 'Race', normativity and the history of sexuality: Magnus Hirschfeld's racism and the early-twentieth-century sexology.

Other

150 pages of readings for the University of Alberta's Palaeontology: Theropod Dinosaurs and the Origin of Birds course

For pay

Papers

Singh, M., Oualil, Y., & Klakow, D. (2017). Approximated and domain-adapted LSTM language models for first-pass decoding in speech recognition.
Deena, S., Hasan, M., Doulaty, M., Saz, O., & Hain, T. (2016). Combining feature and model-based adaptation of RNNLMs for multi-genre broadcast speech recognition.
Huang, Z., Zweig, G., & Dumoulin, B. (2014). Cache based recurrent neural network language model inference for first pass speech recognition.
Klakow, D. (2006). Language model adaptation for tiny adaptation corpora.
Kneser, R., Peters, J., & Klakow, D. (1997). Language model adaptation using dynamic marginals.
Besling, S., & Meier, H.-G. (1995). Language Model Speaker Adaptation.
Kneser, R., & Ney, H. (1993). Improved clustering techniques for class-based statistical language modelling.
Moore, G., & Young, S. (2000). Class-based language model adaptation using mixtures of word-class weights.
Shi, Y., Wiggers, P., & Jonker, C. M. (2012). Towards recurrent neural networks language models with linguistic and contextual features.
Bengio, Y., Ducharme, R., & Vincent, P. (2003). A Neural Probabilistic Language Model.
Klakow, D. (2000). Selecting articles from the language model training corpus.
Adel, H., Kirchhoff, K., Vu, N. T., Telaar, D., & Schultz, T. (2014). Comparing approaches to convert recurrent neural networks into backoff language models for efficient decoding.
Jalalvand, S. (2013). Improving Language Model Adaptation using Automatic Data Selection and Neural Network.
Bacchiani, M., Roark, B., & Saraclar, M. (2004). Language model adaptation with MAP estimation and the perceptron algorithm.
Sanchis-Trilles, G., & Cettolo, M. (2010). Online language model adaptation via n-gram mixtures for statistical machine translation.
Vapnik, V. (1991). Principles of Risk Minimization for Learning Theory.
Mikolov, T., Karafiat, M., Burget, L., Cernocky, J., & Khudanpur, S. (2010). Recurrent Neural Network Based Language Model.
Chen, X., Liu, X., Gales, M. J. F., & Woodland, P. C. (2015). Improving the training and evaluation efficiency of recurrent neural network language models.
Chen, X., Liu, X., Wang, Y., Gales, M. J. F., & Woodland, P. C. (2016). Efficient Training and Evaluation of Recurrent Neural Network Language Models for Automatic Speech Recognition.
Mikolov, T., Yih, W., & Zweig, G. (2013). Linguistic Regularities in Continuous Space Word Representations.
Chen, X., Tan, T., Liu, X., Lanchantin, P., Wan, M., Gales, M. J. F., & Woodland, P. C. (2015). Recurrent neural network language model adaptation for multi-genre broadcast speech recognition.
Gangireddy, S. R., Swietojanski, P., Bell, P., & Renals, S. (2016). Unsupervised Adaptation of Recurrent Neural Network Language Models.
Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. (2014). Dropout: A Simple Way to Prevent Neural Networks from Overfitting.
Li, K., Xu, H., Wang, Y., Povey, D., & Khudanpur, S. (2018). Recurrent Neural Network Language Model Adaptation for Conversational Speech Recognition.
Kuhn, R., & De Mori, R. (1990). A cache-based natural language model for speech recognition.
Grave, E., Joulin, A., & Usunier, N. (2017). Improving neural language models with a continuous cache.
Grave, E., Cisse, M. M., & Joulin, A. (2017). Unbounded cache model for online language modeling with open vocabulary.
Lau, R., Rosenfeld, R., & Roukos, S. (1993). Trigger-based language models: a maximum entropy approach.
Singh-Miller, N., & Collins, M. (2007). Trigger-Based Language Modeling using a Loss-Sensitive Perceptron Algorithm.
Deoras, A., Mikolov, T., Kombrink, S., Karafiat, M., & Khudanpur, S. (2011). Variational approximation of long-span language models for LVCSR.
Chu, C., & Wang, R. (2018). A Survey of Domain Adaptation for Neural Machine Translation.
Mikolov, T., Chen, K., Corrado, G., & Dean, J. (2013). Efficient Estimation of Word Representations in Vector Space.
Mikolov, T., Sutskever, I., Chen, K., Corrado, G. S., & Dean, J. (2013). Distributed Representations of Words and Phrases and their Compositionality.
Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1986). Learning representations by back-propagating errors.
Bojanowski, P., Grave, E., Joulin, A., & Mikolov, T. (2017). Enriching Word Vectors with Subword Information.
Singh, M., Virpioja, S., Smit, P., & Kurimo, M. (2019). Subword RNNLM Approximations for Out-Of-Vocabulary Keyword Search.
Pennington, J., Socher, R., & Manning, C. (2014). GloVe: Global Vectors for Word Representation.

An autistic defense of joke explanations

2022-09-07T00:00:00Z

For a long time now I've been adding joke explanations when I tweet jokes on Twitter, usually in a separate tweet and beginning with "#JokeExplanation" so that people who don't want to see explanations can mute that hashtag or just not read the tweet. For some reason this still bothers a lot of people who insist up and down to me that explaining the joke ruins it. The thesis of this post is: I don't care that it ruins it for them. I care about the people who can participate in my joke because of it.

If you're new here, hello, I'm Vagrant, and I'm autistic. I am in fact the particular flavour of autistic where I often miss big chunks of any given conversation that I'm having with someone. Sometimes people react to things I say in ways that I don't expect, sometimes people say one thing but expect me to do something else (and also to intuit that they want me to do that other thing), and sometimes people make jokes or references that go completely over my head and then they get confused or upset when I don't laugh.

My favourite instance of this - and it's a story I tell over and over - is the time I was hanging out with a wildly funny autistic friend who didn't know I was autistic yet. I missed 3 of their jokes in a row and they just said, "Do you realize you're ''not getting it'' more often than an autistic person?" And this, dear readers, is the story of how they clocked me as being autistic before I even told them.

This is one of the fun stories. There are also times when I'm later unsure if people were making fun of me and I just didn't know. I often analyze conversations afterwards, alone or with friends or with my therapist. Like scientists examining a bug with a magnifying glass, we try to answer questions like: why did the group laugh at that sentence and not the one after, the one that contained the actual joke? What is it exactly about this tweet that is funny? When I asked someone a question and everyone else laughed, did they think I was mocking them? Am I funny? What is humour, anyway? By the way, this is where it's really great to have a lot of autistic friends and a lot of linguist friends (and let me tell you, there's a LOT of overlap there).

I've learned that I'm adept at making accidental puns that make people howl with laughter while I just stare blankly until I do the math on what I just said. I'm open about the fact that when I give talks, I script all my jokes ahead of time. With a couple of autistic friends I've discussed my algorithm for figuring out whether someone just made a joke when I don't know them very well. As you can imagine, all of this has a high cognitive load. I'm not one of the people blessed with the effortlessly-funny gene.

You know what lessens the cognitive load? Tone indicator tags over text (like '/j' for joking, '/s' for sarcasm), laugh tracks, joke explanations.

Laugh tracks are comforting because they point to the joke and say THAT'S A JOKE. A joke is a joke in the social context it's in, so I may not always agree that it's funny or acceptable, particularly in media where making fun of fat people or trans people is considered something your laugh track can laugh at. Nevertheless, having an explicit signal of how the creators expect me to react to something makes me more equipped to understand the social values represented in a piece of media rather than having to guess. If only people came with laugh tracks that revealed all their values.

To me joke explanations combine a form of accessibility with a warm invitation to learn something new. It's my way of saying come here, I did something clever and I want to show it to you and show you how I did it. Given that I often describe how I come up with jokes as a process of "construction" or "engineering," this makes a lot of sense. Sharing the end result and the process is a love language.

Many people I know (even allistic people!) recognize the access and the love in my explanation tweets, with some telling me that my explanations make intent explicit, make them feel fuzzy and loved, help them understand and appreciate joke craftsmanship they otherwise would've completely missed (especially when I make multilingual jokes - often). So if you're one of the people annoyed by jokes being explained, I encourage you to take a more expansive view of who is let in via the explanations, and remember that letting those people in doesn't stop you from understanding and appreciating the joke if you do instantly get it.

If you cannot find it in your heart to enjoy joke explanations, you should also feel free to simply not read them, mute the hashtag, or even unfollow me. To everyone who loves my explainers, don't worry - the haters will pry my joke explanations from my cold, dead hands.

Jokes (click for explanations)

La Croix should enter the German market with an apple flavour and call it Apfelsprudel.

La Croix is a company that makes fruit-flavoured sparkling/carbonated water, Sprudel is the German word for sparkling water, and Apfelsprudel would make for catchy branding because it sounds like the classic pastry Apfelstrudel.

Math transes love me because I'm homo and I'm toppy.

There's a big intersection between math twitter and trans twitter and they all seem to study category theory and homotopy theory - which looks like homo and toppy (but isn't pronounced the same).

Call for submission! Are you a kinky queer? Submit to the Journal of Trans Disciplinary Studies!

The journal of transdisciplinary studies is a real journal. Trans disciplinary studies could be the study of trans people engaging in (BDSM) disciplining. Our journal is seeking *submission*, presumably from subs (ha).

Do British transfems say they're on O?

Alternate (predominantly British) spellings of estrogen and estradiol are oestrogen and oestradiol.

My grandmother (who believes in astrology - I don't) says this is an auspicious time for me. For years I have held myself back from telling her she's *technically* right, but only because birding is a big part of my life.

This joke requires etymology + Latin knowledge. Auspicious comes from Latin auspice / auspex which literally means "one who looks at birds." This is related to the ancient roman practice of augury (using birds to interpret omens).

Ich habe einen Sitz-/Stehschreibtisch von IKEA - ich kann stehen und hüpfen oder sitzen und meine Beine schütteln. Außerdem habe ich Zappelspielzeug in den Schubladen. Das ist mein AutisTisch!

I describe various autist-friendly features of my desk and then call it my AutisTisch which is wordplay because "autistisch" is an adjective meaning autistic and also "Tisch" is table in German.

Himbeeren. Herbeeren. Thembeeren. Xembeeren.

This is a bilingual German + English joke about berries and pronouns. Himbeeren is the German word for raspberries. Him is an object pronoun in English and the remaining words are German berry names I invented with other English object pronouns.

I hope I see a magpie on New Year's Eve - a Silvelster, if you will.

This is a niche German x bird x blend joke! Silvelster is a blend of Silvester and Elster, the German words for New Year's Eve and magpie respectively.

What do you call a restaurant by a German train station that serves homemade Thai food? Bahn Thai.

Bahn is train in German and house/home in Thai.

Sometimes when I'm annoyed my accent shifts. I have Irritable Vowel Syndrome.

It's a play on IBS, Irritable Bowel Syndrome but for vowels instead which can change with different spoken accents. This doesn't actually happen to me but it makes for a decent joke.

My German friend is a philosophy major currently doing an internship. I think that makes him a PraktiKant.

Praktikant is the German word for intern/trainee (masculine) and Kant is a famous philosopher (who also happens to be German).

A uterus that can map humans to new humans. A hystomorphism, if you will.

Morphisms (category theory) map between objects of the same type, e.g., histomorphisms. Hysto sounds like histo- and I'm using it as a uterus-related prefix. This is slightly cheating because "hysto" refers to hysterectomies in trans circles and the technically correct prefix for uterus-related would be hystero-.

What do you call it when a ternary computer stores its most significant bytes at its least significant address? Big-enbian.

Endianness in computers describes how bytes are ordered in the system. Big-endian = most significant bytes at least significant address, little-endian is the opposite. A ternary computer is not binary, like enbies!

What do operant conditioning chambers (from psychology / animal behaviour) have in common with slaughterhouses? They're both skinner boxes.

Animals get skinned in often vaguely box-shaped buildings known as slaughterhouses. And operant conditioning chambers are another name for Skinner boxes à la B. F. Skinner, psychologist prof dude.

Apparently ternary computers use trits, not bits. So would 8 of them still make a byte? Or would that be... Trite?

Trite means cliché and it also works with the rhyming scheme of bit : byte :: trit : ?

BREAKING: CHURCH RELIC NOT SAINT HEART BUT CRANE HEART. SACR cœur.

SACR = ABA 4-letter bird code for sandhill crane, cœur = French for heart, Sacré-Cœur = church and minor basilica in Paris.

Why did Labov hate his sabbatical in Singapore? Because there was no fourth floor.

Labov is a sociolinguist famous for an experiment where he went around to New York City department stores and made salespeople say something was on the "4th floor" to see how they said the R sound in both words. He found out that the way you pronounced your Rs correlated with socioeconomic status (quantified here by how fancy the department store was). Buildings in Singapore often skip the 4th floor because 4 is considered an inauspicious number. So he wouldn't have been able to do his experiment there. Also as far as I know Labov never did a sabbatical there.

I have an inner gender cop. A genderme, if you will.

Genderme sounds like gendarme but with gender in it. A gendarme is a French cop.

What I read in September

2022-09-30T00:00:00Z

For fun and edification

Books

hooks, bell. (2000). All About Love: New Visions.
Wright, K. E. (2022). Black Professionalism: Perception and Metalinguistic Assessment of Black American Speakers' Sociolinguistic Labor.
Graeber, D. (2018). Bullshit Jobs.

Papers

Rogoff, I. (1996). Gossip as testimony: a postmodern signature.
de Sousa, R. (1994). In Praise of Gossip: Indiscretion as a Saintly Virtue.
Collins, L. (1994). A Feminist Defense of Gossip.
Ayim, M. (1994). Knowledge Through the Grapevine: Gossip as Inquiry.
Hammer, J., & Reig, S. (2022). From Individual Rights to Community Obligations: A Jewish Approach to Speech.
Spiel, K. (2022). Transreal tracing: Queer-feminist speculations on disabled technologies.
Vieth, R. (2022). Addressing Serious Harm, Reconsidering Policy and Building Towards Repair:
Warren, D. C. (1928). Inheritance of Earlobe Color in Poultry.
Goldinger, S. D. (1996). Words and Voices: Episodic Traces in Spoken Word Identification and Recognition Memory.
Schmidt, M. (2019). Creating Worlds: Fan Modifications of Civilization 4.

For pay

Papers

Sennrich, R., Haddow, B., & Birch, A. (2016). Neural Machine Translation of Rare Words with Subword Units.
Kudo, T. (2018). Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidates.
Seppnen, J. J. (1982). Recursive Functions for Computation of Natural Secret Languages.
Gaspari, F. (2006). Look who's translating.
Somers, H. (2005). Round-Trip Translation: What Is It Good For?
Tilk, O., & Alumäe, T. (2014). Multi-Domain Recurrent Neural Network Language Model for Medical Speech Recognition.
Shi, Y., Larson, M., & Jonker, C. M. (2015). Recurrent neural network language model adaptation with curriculum learning.
Varjokallio, M., & Klakow, D. (2016). Unsupervised morph segmentation and statistical language models for vocabulary expansion.
Durrani, N., Dalvi, F., Sajjad, H., Belinkov, Y., & Nakov, P. (2019). One Size Does Not Fit All: Comparing NMT Representations of Different Granularities.
Mager, M., Oncevay, A., Mager, E., Kann, K., & Vu, N. T. (2022). BPE vs. Morphological Segmentation: A Case Study on Machine Translation of Four Polysynthetic Languages.
Radford, A., Wu, J., Child, R., Luan, D., Amodei, D., & Sutskever, I. (2019). Language Models are Unsupervised Multitask Learners.
Smit, P., Virpioja, S., Gronroos, S.-A., & Kurimo, M. (2014). Morfessor 2.0: Toolkit for statistical morphological segmentation.
Arisoy, E., Chen, S. F., Ramabhadran, B., & Sethy, A. (2013). Converting Neural Network Language Models into back-off language models for efficient decoding in automatic speech recognition.
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019). BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding.
Creutz, M., & Lagus, K. (2002). Unsupervised discovery of morphemes.
Kohonen, O., Virpioja, S., & Lagus, K. (2010). Semi-Supervised Learning of Concatenative Morphology.
Koehn, P. (2005). Europarl: A Parallel Corpus for Statistical Machine Translation.
Feng, S. Y., Gangal, V., Wei, J., Chandar, S., Vosoughi, S., Mitamura, T., & Hovy, E. (2021). A Survey of Data Augmentation Approaches for NLP (arXiv:2105.03075).
Li, B., Hou, Y., & Che, W. (2022). Data augmentation approaches in natural language processing: A survey.
Liu, P., Wang, X., Xiang, C., & Meng, W. (2020). A Survey of Text Data Augmentation.
Shorten, C., & Khoshgoftaar, T. M. (2019). A survey on Image Data Augmentation for Deep Learning.
Bayer, M., Kaufhold, M.-A., & Reuter, C. (2022). A Survey on Data Augmentation for Text Classification.
Shijie, J., Ping, W., Peiyi, J., & Siping, H. (2017). Research on data augmentation for image classification based on convolution neural networks.
Park, D. S., Chan, W., Zhang, Y., Chiu, C.-C., Zoph, B., Cubuk, E. D., & Le, Q. V. (2019). SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition.
Chia-Hui Chang, Kayed, M., Girgis, M. R., & Shaalan, K. F. (2006). A Survey of Web Information Extraction Systems.
Mollá, D., & Vicedo, J. L. (2007). Question Answering in Restricted Domains: An Overview.
Gupta, P., & Gupta, V. (2012). A Survey of Text Question Answering Techniques.
Grishman, R. (2015). Information Extraction.
Almansor, E. H., & Hussain, F. K. (2020). Survey on Intelligent Chatbots: State-of-the-Art and Future Research Directions.
Chen, H., Liu, X., Yin, D., & Tang, J. (2017). A Survey on Dialogue Systems: Recent Advances and New Frontiers.
Shum, H., He, X., & Li, D. (2018). From Eliza to XiaoIce: challenges and opportunities with social chatbots.
Hussain, S., Ameri Sianaki, O., & Ababneh, N. (2019). A Survey on Conversational Agents/Chatbots Classification and Design Techniques.
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., Agarwal, S., Herbert-Voss, A., Krueger, G., & Henighan, T. (2020). Language Models are Few-Shot Learners.
Wang, T., Zhao, C., Wang, M., Li, L., & Xiong, D. (2021). Autocorrect in the Process of Translation — Multi-task Learning Improves Dialogue Machine Translation.
Wermter, S., Riloff, E., & Scheler, G. (1996). Learning Approaches for Natural Language Processing.
Marcus, G. F. (1998). Rethinking Eliminative Connectionism.
Garoufi, K., & Koller, A. (2011). Combining symbolic and corpus-based approaches for the generation of successful referring expressions.
Marcus, G. (2018). Deep Learning: A Critical Appraisal.
Gonen, H., & Goldberg, Y. (2019). Lipstick on a Pig: Debiasing Methods Cover up Systematic Gender Biases in Word Embeddings But do not Remove Them.
Blodgett, S. L., Barocas, S., Daumé III, H., & Wallach, H. (2020). Language (Technology) is Power: A Critical Survey of "Bias" in NLP.

Pronouning me in English

2022-12-03T00:00:00Z

Usage

xe/xem

Xe is a birder. Xe loves birds.
This mug belongs to xem.
Xyr jacket is on the table.
That bag is xyrs.
Xe wrote this xyrself/xemself.

they/them

They are a birder. They love birds.
This mug belongs to them.
Their jacket is on the table.
That bag is theirs.
They wrote this themself.

Grammar and linguistics

What are pronouns?

Pronouns are used to stand in for nouns or noun phrases, e.g., who, those, you, I, anybody, no one, etc. Personal pronouns are associated with a specific person, e.g., first person (I, we, us, etc.) second person (you, yourself, yours, etc.) and third person (he, itself, their, etc.). When you are alone and conversing with a person, the personal pronouns you use will mostly be first and second person pronouns, of which the standard English ones are not gendered. When you are conversing with someone and referring to someone else (who may or may not be present), you generally use a name or a third person pronoun. The standard English ones you are definitely familiar with using for people are he/him and she/her, although more exist and are in use.

Just like when people tell you their name in an introduction, presentation, email signature, profile or internet handle, so that you can refer to them using that information regardless of whether they are present, listing third person pronouns in these spaces generally indicates that they would like you to refer to them using those pronouns.

What are all the ingredients I need in order to use a set of pronouns?

To properly use a set of pronouns for someone, you need to know its complete "declension" or all the different forms of a pronoun, depending on the grammatic role it is playing in a sentence. In English there are five forms - the subject, the object, the dependent possessive, the independent possessive, and the reflexive. The five forms of he are: he, him, his, his and himself. Similarly, the full declension of xe is: xe, xem, xyr, xyrs and xyrself.

Often rather than giving the entire declension of a particular pronoun set, we tend to use shortened versions - for example, you might see she/her, she/hers or she/her/hers, rather than she/her/her/hers/herself since there is some repetition and some forms can be inferred from other forms. In the case of pronouns that are less familiar to you, you may need more information on the full declension in which case you can use your favourite search engine to find out, or ask the person.

Another piece of information that is important especially with English pronoun sets that may be less familiar to you is whether it takes singular or plural agreement with verbs. This is the difference between saying He is a student and Ey are a student.

Isn't "they" only for multiple people?

Nope. Singular they exists too, and you are probably already familiar with using generic singular they, e.g., in the sentence Someone left their keys in the kitchen. This use of they refers to a single person who is unspecified and generic, sometimes even unknown. You might be less familiar with singular specific they, e.g., Vagrant left their keys in the kitchen, but that can be learned.

Aren't pronouns unchangeable? Aren't they a closed class of words that cannot be added to?

All language is productive and ever-changing, thanks to the creativity of its users. There are pronouns that were once part of standardized English but no longer are (e.g., thou, thee). Conversely, the pronoun you was plural before it was singular. Swedish now has a widely accepted gender-neutral pronoun that was first proposed by a linguist less than 60 years ago. So it is not true that pronouns are a class of words that can never be changed.

Pronouns in English are often described as a closed class, i.e., a class that does not easily acquire new members. Anecdotally, this is clearly not the case in many English language communities I am part of. There is also evidence that English pronouns are shifting to an open class category, like they are in Japanese.

Miscellaneous pronoun links

Eugenics, everywhere, all the time

2022-12-27T00:00:00Z

I've been thinking about how the logics of ableism and eugenics are what I see as the "last frontier" of what I want people to understand and undo in their lives, because this is one of the patterns I consistently see in people across the political spectrum.

It shows up when otherwise progressive (abled, usually white) people will prioritize their environmentalist views as being more important than an accessibility requirement like plastic straws (or than certain cultural eating practices!), or when people make arguments about how disabled people (or trans people!) "take up a lot of resources" and we need to decide whether there's a "critical mass" of them in a community before we put in work to make something accessible for them, or when people insult other people whose opinions they disagree with with words like [slur for intellectually disabled people] (or ugly! or fat!), or when people will justify a higher level of death risk from a disease as long as it's "just the people who are already sick" (or just The Gays™!).

All of these are things I've seen progressive / liberal / leftist people do, and some are even disabled themselves. What unifies all of the main examples is ableism - the idea that there are some ideas or people or things that are more important than disabled people being able to live, or lead good lives or... just be considered as worthy of respect as abled people. The point I'm making with my parenthetical connections to gay people and the AIDS crisis, trans people, fatness, etc., is that I see all of these issues as interconnected and I see all of them using the same underlying logic: the logic of eugenics.

I deeply value consistency in my views, actions and belief systems, and when people can be comfortable using this kind of logic in some contexts or for some people even when they never would in other contexts, that makes me uncomfortable. And I get it, we're brought up on a steady cultural diet of ableism, transphobia, homophobia, fatphobia, capitalism, etc. so none of us is immune to absorbing and reapplying this type of logic - least of all me. But if our unlearning process is oblivious to the fact that we are still replicating and reinforcing harmful ideas regardless of who we do it to, then I think we're just not doing social justice right.

You can't say you value pluralism in human beings when you then slice and dice that pluralism and say some pluralisms are more valuable than others. Caring for humanity means caring for the same basic things for every person who comprises it.

Pluralism under capitalism: an indifferent eugenics

Lots of people are pro-pluralism in theory, but run into problems again when you add capitalism to the mix. Then the argument becomes that they care for the same basic things for every person who comprises humanity, as long as it doesn't take too much effort (money). Capitalism makes people evaluate other people in terms of their cost and benefit to society, and cheaper means better. More productive means better.

My most recent example of this was a horrifying conversation with a friend who said that German statutory insurance covers PrEP (a highly effective HIV prevention drug) thanks to activism from gay men in Germany, but that it's very expensive and perhaps society shouldn't pay for gay men to safely have sex if it's that expensive. To someone like me this is already absurd but it's at least exponentially more absurd to hear a eugenicist argument from someone who has only had sex with gay men, and is trans and disabled. The irony of making the same arguments regularly weaponized against these groups was sadly lost on them.

While we were dissecting this, they found it confusing that I called their argument 'eugenicist'. According to them, Real Eugenics™ is killing and sterilizing people, stopping them and their "bad genes" from reproducing. They hadn't expressed a desire for gay men to die or not reproduce, so how could what they said be eugenicist?

I find their definition of eugenics too narrow. For one, it misses positive eugenics - encouraging (and sometimes even forcing) people with "desirable" traits to reproduce. Their definition is also too focused on a specific set of actions, whereas I prefer to work with a broader definition that can identify eugenicist thought before it turns into actions as dire as death and sterilization. If we are unwilling to label something as eugenicist before it turns into a full-blown Nazi-level coercive action against groups of people, we risk missing opportunities to stop it earlier. It's just like the importance of identifying genocidal rhetoric before it turns into full-blown genocide, e.g., what the Dangerous Speech Project tries to do.

So if we step back and try to formulate a more abstract definition of eugenics, it's the idea that there are desirable traits and undesirable traits, and that classifying people as desirable or undesirable based on this is grounds for how we should actively control or nudge or care about their life, death and reproduction.

I'd like to propose a third kind of eugenics that fits under this umbrella and thrives under capitalism: indifferent eugenics. Indifferent eugenics still classifies people as desirable or undesirable, and remains indifferent to the (quality of) life and death of "undesirable" people.

There are a number of examples of indifferent eugenics beyond my PrEP anecdote - Canada's Medical Assistance in Dying law and the response to the COVID-19 pandemic in most countries come to mind. I strongly recommend the first two links in the further reading section for writing from disabled people about each of these issues, and both blogs discuss disability justice more broadly as well. Their writing has nuance and complexity and addresses important questions that are out of scope for my blog post, e.g., how autonomy around suicide is not incompatible with being against MAiD.

As before, these issues are primarily about disability and ableism, as they are examples of society devaluing disabled life and being indifferent to disabled death. Because of this I think it is important to centre disability and disabled voices in these conversations, but in the spirit of showing the interconnectedness of different struggles, it is also worth noting that people who are oppressed along axes other than ability are also often oppressed by these policies. For instance, Black and brown people have accounted for a disproportionate number of COVID-19 deaths in the US - deaths met mostly with indifference. Similarly there have already been cases of people who are experiencing poverty using MAiD.

Solutions to indifference

I asked my friend if they intended to just throw me away if I had another depressive episode and became "expensive" for society and useless to everyone, including them. They claim they wouldn't, which is inconsistent with the rest of what they said as applied to the generic Other. It's because they cannot be indifferent when it comes to someone they care about, and I do believe they care about me.

The way I see it, this is the only antidote to indifference: care. Radical care for the people around us, the people who are our community and society for multiple definitions of both. Society, in my view, exists to serve the people in it and we are - we have to be - accountable to one another in a bigger, collective sense. Mia Mingus frames this as "interdependence," an idea that "acknowledges that our survival is bound up together, that we are interconnected and what you do impacts others."

We have to hold fast to the idea that every person inherently has value and apply it tirelessly to the people who are consistently and systemically devalued in our societies, not just ourselves and not just our friends. We must value disabled life. We must value Black life. We must value trans life. We must particularly value life that exists at the intersections of many of these groups and we must invest in their lives and their joy, rather than investing in or being indifferent to their deaths.

It is also absolutely critical that we do not fall into the trap of applying our logic selectively. We cannot just care about our friends. We cannot only care about a group of people as long as they are "cheap" to care about or "easy" to care about. We cannot pull a Hans Asperger and divide ourselves up into "good" crips and "bad" crips. We cannot find eugenics horrifying only when it's on the scale of Nazi eugenics, but we should find it horrifying and unacceptable at every stage before that which shares the same logic. We simply cannot afford to be indifferent.

All of this is obviously completely incompatible with neoliberalism, which stresses individualism and austerity, and with capitalism, which I see as one of the driving forces that makes us see human beings as commensurable, fungible and replaceable in the first place, which is why those are systems that need to be dismantled in favour of others, for a society that includes people like me all the time. "Interdependence cannot exist in scarcity, competition, comparison, domination or greed. It flourishes in abundance, appreciating and honoring difference, collective care and collective access," according to Mia Mingus. As I have said throughout this post, I also see multiple kinds of oppression as being inseparably connected and that's why I believe that one violent system (e.g., ableism) cannot be meaningfully dismantled without addressing all the others.

Conclusion

The title of this post is a reference to a film where, incidentally, it is not violence or nihilism that saves the day, but rather kindness and radical care. One of the main characters, Waymond Wang, embodies a subversive masculinity that is vulnerable, communicative and caring, and the analysis of his character in this video essay about Everything Everywhere All At Once is definitely worth watching if you've watched the film.

I don't believe I can convince people who don't already value disabled life to start valuing it now after reading one blog post, nor convince people who aren't already aware of the problems with capitalism that capitalism has problems. But for those who know both things already, yet express inconsistency with your thoughts, words and actions where they come together: I hope you join me and Waymond and choose radical care.

Bureaucratic violence

2023-02-25T00:00:00Z

I now understand why it's called a deadname. It's because when I see it, I want to die.^[1]

What's in a deadname?

The term "deadname" is usually used to describe a trans person's previous name that brings them pain. This definition lacks nuance for my purposes because there are trans people who don't change their names, trans people with previous names that aren't painful, and cis people with painful names they'd like to move past—associated with an abuser, estranged family, or just a different time of their life. In this post, I use the word "deadname" liberally because in pretty much all these cases, using the right name for someone is simply an issue of dignity - regardless of the presence or absence of pain, and regardless of one's relationship to gender.

It is worth pointing out though, that in studies and surveys of trans youth, chosen name use is associated with reduced depressive symptoms and suicidal ideation, changing our names and/or gender markers on legal documents is associated with lower rates of attempting suicide, and all of this is in a group that experiences worse mental health outcomes when compared to cis LGTBQ peers. So when I said that repeatedly seeing my deadname makes me want to die, that is neither an exaggerated nor exceptional experience.

When policy fails you

My least favourite word in the German language is Geduld. It means patience, and I hate it because it appears in nearly every response to me in the ticket I've had open with my university's IT department for an entire year, begging them to change their naming policy for Microsoft Teams.

I've contacted the IT department and the Equal Opportunities Department (Gleichstellungsbüro), and the queer working group of the university's student council has been trying to get this policy to change for years.

It would be one thing if they changed names promptly on request, but they don't do this. Instead their policy is to require legal name changes to change your display name, a process that I cannot access as a foreigner, and that costs between 2 and 3 thousand euros even if I could. I'm not the only student and employee at the university for whom this is a barrier, and I know several other trans people who would like to see this policy change. Some others have contributed testimonials below.

For us, until we magically produce name change documents, we still have to attend university events and research group meetings and message our colleagues under an incorrect name. Every online meeting displays the wrong name front and centre, even if we introduce ourselves differently. We don't actively participate as much because we know we'll be called on with the wrong name if we speak and that is significantly distressing.

When this is what we're dealing with, how on earth are we meant to realize our full potential as students, researchers and employees?

The Sisyphean task of a legal name change

In parallel, I've been trying to get a legal name change in Canada through a loophole I discovered.^[2] The requirements seemed simple enough - a form with a notary's signature and stamp, fingerprints, and some money. But I underestimated how much trouble I would have finding a notary and anyone (city halls, police stations) who would take my fingerprints in Germany. The first notary ghosted me after a series of emails, the second never replied, the third agreed to notarize it in a non-standard way and charged me extra because the form was in English. The city halls said they couldn't take my fingerprints but to try the cops; the cops oscillated between 'We can't do this', 'We can do this if you give us a letter from the consulate' (which of course I produced for them), 'German law doesn't allow us to do this even with a letter from the consulate' and 'We can do this!'. I've been on this merry-go-round for a while and it is neither merry nor have any fingerprinting appointments materialized yet.

For my (still unsuccessful) name change, here's all that I've had to do so far, a lot of which has happened in German, a language I've been learning for, oh, 16 months:

10 emails with notaries
2 notary visits
3 notary phone calls
3 police station visits
4 police station phone calls
2 city hall phone calls
4 police station emails
3 consulate phone calls
6 consulate emails
9 Vital Statistics Agency emails

In fact, I'm about to send the Vital Statistics Agency a 10th email because my special name change process has a time limit on it and I'm about to run out with nothing to show for it - which means, once again, I just can't have a legal name change. It is absolutely soul-destroying.

Bureaucratic violence

A lot of people think that policies are just policies, that they're somehow neutral. But when a policy disproportionately impacts a group of people and makes us do the equivalent of a part-time secretarial job in order to be afforded the dignity of being referred to by our names, then that turns into violence, bureaucratic though it may be.

My mental health is on the whole pretty good but fighting these ridiculous bureaucratic barriers at the university and around my name change have been having detrimental effects on me. I told a friend half-jokingly that if things ended up getting bad and I killed myself, she should politicize my death as being about these transphobic policies. So far I'm still kicking but there are literal lives at stake, and many people are less vocal than me.

What I feel is powerless. When states require that I tell them my name, my gender, my state of transness, I am now powerless to resist, simply because institutions like my university cannot bother to dignify me with an identity outside my state-given one. It is institutional policy to see me only in the ways I have institutional legitimacy, rather than to pose the question: why the fuck does my neural networks professor need to know my government name? I have discovered I am powerless even in the face of ostensibly trans-friendly policies such as PLOS One's naming policy for trans authors, because as it turns out, when no one monitors the mailboxes you send your name change requests to, the policy is as good as nonexistent.^[3]

I'm spending my evening writing this post, making posters so more people know the impact of my university's policies on its trans students and employees, and sending more emails to the Vital Statistics Agency to apologize for wasting their time with a name change request I simply can't seem to produce the required documentation for. It's deeply disappointing that this is the world I live in.

I want to write fun blog posts. I want to write about my deep dive into vampire films and about my incredibly fun compilers class and about the intersectionality literature I've been reading lately. I want to go outside and look at birds and go on long walks and collect more skull graffiti. Cross your fingers that the next post I have time to write is a fun one.

Appendix: Testimonials from other students

I have collected these contributions from current students at Saarland University, some of whom have asked to be quoted anonymously.

Anonymous Contribution:

When first enrolling in classes at Universität des Saarlandes, I had to put my legal name, as seen on my ID card and passport, into the required forms. I did not realize this meant that not only my first, but also my second and third name would be visible to all my fellow students come the COVID online semesters. My names are so long that lecturers often don't see my last name displayed on MS Teams, which means they read out all three of my very gendered names. I've wanted to cry during class on more than one occasion because of this. It is very tiresome to request that my lecturers address me by my preferred name or read out that name instead all three of my deadnames when they take attendance. It means I have to out myself every time I have a new lecturer. So far my lecturers have been understanding and accomodating of my request but what happens if a future lecturer is not?

Since I'm not a binary trans person it is very unlikely that I get a name change approved by the court under the current law, even though I am a German citizen. I also have neither the time, the energy nor the money for this process. It infuriates me that the university could make the process easier for trans students. There's nothing that prohibits them from accepting something called the DGTI Ergänzungsausweis, a form of identification for trans people that can't or don't want to access a legal name change through the courts. They simply choose not to. They don't seem to care that this hurts their trans students and leaves them vulnerable to discrimination.

It pains me to see the wrong name on my university account. It pains me to see the wrong name on my bachelor's degree. I really hope I won't have to see my deadname on my master's degree as well.

Kamila:

The current University's naming policy makes it difficult for me to study. My UdS ID, CMS, exams, etc... use my real name, while LSF, MS Teams (Office), and other services use my legal name. This leads to the examination office notoriously losing my references. I was e-mailed by my professor about the difficulties he had with entering my data to the system. I have to constantly make sure that nothing is lost and I am genuinely scared that the University will sooner or later lose one of my grades and shrug me off. The admissions office also refuses to update the LSF entry claiming that it would be illegal of them to use my real name everywhere. Finally, I have already transitioned socially and nobody whom I interact with in my daily name knows/remembers my legal name. This leads to plenty of confusion and incredibly awkward situations especially when the University coerces me to use it.

In July 2021, I wrote a piece for Pride Magazine about my Indian heritage, gender, and various comings out, where I said it would be a loss to give up my erstwhile name—a gendered Sanskrit name—in favour of one that was more neutral. I spent more time wrestling with this and trying various options out before settling on Vagrant, which feels perfect for me. In April 2022, I started the full social transition and technical upgrade to Vagrant with another blog post announcing my new name where I refer to my "old name," not a deadname. This was a conscious choice because at the time, that's what it was. It was just an old name, not one that prompted pain or really any feeling at all. As of this blog post, written in February 2023, that's changed. It turned from a previous name to a deadname. ↩︎
British Columbia's Vital Statistics Agency says on their website that name changes are only available to residents, but through emailing them I discovered that there's an exception to this: you can apply for a name change from abroad if you were born in BC. ↩︎
I have written to plosone at plos dot org and privacy at plos dot org in May 2022, August 2022 and October 2022, requesting to have my name updated on a publication I have in PLOS One. I've received no response aside from one automated acknowledgement email, and my name has not been updated. ↩︎

Skull-spotting in Saarland

2023-05-19T00:00:00Z

Most people who know me know that I love birding, but only people in Saarland know about my new hobby of collecting a special type of local graffiti: numbered skulls created by the KB63 graffiti crew. I started about a year ago and my official count is 101 skulls (plus another 15 to 20 I've marked to properly visit later), so this felt like a good milestone to ~~brag~~ blog about.

Skull-spotting seems to scratch the same itch as birding because there's a thing to collect and make lists of, and both hobbies make it easier for me to deepen my connection with places. Getting to know a city's regular bird visitors turns it from a place I sometimes walk through into a living, breathing home whose rhythms I know. Just as I now recognize the return of the screaming swifts as a temporal anchor heralding the spring, the skulls have become my geographical anchors as I've learned the layout of Saarbrücken. Telling me street names or store names is futile, but if you tell me your house is near that one skull with the really elongated head I will know exactly where you live. Since I found out that these skulls are spread out all across the (federal) state of Saarland, I've been venturing further out on my skull-hunting trips, to places I would have little reason to visit otherwise (like Luisenthal and Güdingen).

When I started systematically collecting pictures and locations of the skulls, I thought I could collect them all and I hypothesized that a plot of all the coordinates when connected would give me a much larger skull over Saarland. That theory was easily falsified when I had collected some data and made a preliminary plot. More importantly though, the process of data collection revealed several assumptions that turned out to be bad. My first setback was finding that not all skulls were numbered, unlike a typical connect-the-dots puzzle. I also found that several were numbered 18, meaning that numbered skulls weren't even uniquely numbered. And the worst blow to this project by far was when I realized that 3 skulls I'd previously collected had recently been painted over, i.e., I was never going to be able to get all of them because some might disappear before I ever even saw them.

Despite my flawed assumptions and no grand unifying pattern of skull graffiti, I've still learned some interesting things from the data:

Beyond falsifying my theory about the shape the coordinates would form, plotting skull coordinates on a map of Saarbrücken was also useful to confirm some anecdotal observations about their geographic distribution; I've found a lot of skulls along main roads and highways after spotting them on the go. I'm reluctant to say that the artists prefer to place skulls in these locations without actually asking them, because I can't be sure that my observation isn't just a result of selection bias (further away from home, I am more likely to travel on buses and trains than on foot).
Even though there don't seem to be any patterns that unify all the skulls, there are some pockets with progressions of consecutive numbers, e.g., skulls 80 to 92 seem clustered between the neigbourhoods of Sankt Johann and Rotenbühl, and the 18s all seem to be in Alt-Saarbrücken, close to the Betriebshof.
Anecdotally, although some skulls are placed in more visible areas, many others are tucked away and signalled with the word 'tut' spray painted in a more visible location. For instance, skull #260 was on the back of a billboard, making it impossible to see from the road or even the sidewalk, but it appeared close to a 'tut' spray painted onto the side of a shed that I could see from the bus.
Across locations, the artists seem to love using power boxes, cement and brick walls, and corrugated metal surfaces like garage doors and sheds. These observations are probably limited by survivorship bias because I have found a handful of skulls that were so faded from the weather that they were almost impossible to see. I suspect that over time the artists have developed preferences for different surfaces and materials, and that's also something I'd love to ask them about.
Assuming that numbers increase with time, there has been an evolution in the art style: older skulls are usually wider and white or silver, and they tend to have open jaws, noses shaped like inverted Vs, and numbers painted outside the skull. Newer skulls on the other hand are generally yellow and narrow, with a compact arrangement of numbers, closed jaws, noses shaped like inverted hearts, and more decorative squiggles. The newer skulls also seem to show more experimentation with other expressions and accessories - I've seen a worried skull, a skull with sunglasses, a skull eating what appears to be bricks, and a masked skull.

While I have learned a lot just from collecting data and talking to Saarlanders, I still have many unanswered questions that I would love to ask the artists if I should ever meet them:

How did you start with skulls?
Is the skull a specialty of one person in the group or are there multiple artists who create each one?
What does "tut" mean in the context of your skulls?
What's the process? How do you decide on a new location to graffiti? What time of day do you generally do it? How often do you do it?
What is the history of the numbering of the skulls? Why are a few of them numbered #18? Why have you chosen not to number some?
What materials do you use? And are there any reasons for the apparent changes in colour over time? E.g., it seems like you phased silver out in favour of white, and the more recent ones seem to mostly be yellow.
Do you keep track of all of the ones you've done on a map or a list or anything?
I've noticed that some of the skulls have paint splashed over them - why do you think that is? Are there people who dislike your skull graffiti or your group?
I've been honked at while trying to get up close to photograph some of the skulls in slightly more dangerous locations (by highways, train tracks, etc.), and I assume you have gotten into trouble while creating the graffiti in the first place. Do you have any fun stories you can share?
Do you have a favourite skull you've done?
How long have you been doing this? What have you learned along the way?
Would it be okay if I got a tattoo of one of the skulls?

If one of the artists is reading this, I would love to grab some beers - on me. As for everyone else, I hope that wherever you are, you're enjoying your local art and your local birds!

Intersectional subgroup fairness is not intersectionality

2023-09-15T00:00:00Z

In the last decade, intersectionality has become something of a buzzword in diversity and equity circles, and is commonly misinterpreted as being exclusively about how people at the intersections of multiple minoritized identity categories experience "more" oppression.

Intersectionality is in fact a much more complex and nuanced framework for examining society. It posits that systems of power and oppression interlock to produce social inequality, and that this produces qualitatively different outcomes for different groups of people. In Lisa Bowleg's words, Black + Lesbian + Woman ≠ Black Lesbian Woman. From its very conception, intersectionality has had the explicit goal of obtaining political and social justice - through intellectual activism, yes, but also through legal advocacy, movements for prison abolition, participatory action, etc. Its activist roots make it a framework that requires both critical inquiry and praxis (i.e., practical action, not just academic theorizing) to examine and dismantle systems of power at all the levels where they exist - socially, structurally, politically.

Intersectionality thus has great potential for application to any field that studies societal inequalities and seeks to eradicate them. AI fairness is one such example, with a growing body of papers using intersectionality in their titles or motivation. We surveyed a number of these papers in our work, seeking to answer the following questions:

How is intersectionality discussed in the AI fairness literature?
What are the gaps in how it is used in AI fairness, as opposed to in core intersectionality literature?
Can we translate these gaps into recommendations for us to do better AI fairness work going forward?

What "counts" as intersectionality?

The broad scope of intersectionality is sometimes considered one of the reasons it's been so successful as a travelling theory. A more practical consequence of its breadth is that everyone isn't always agreeing with each other; we've now read original texts, critiques and counter-critiques of intersectionality, spanning critical theory, Black feminism, legal studies, cultural studies, political science, sociology, and many more fields.

In the end we settled on grounding our study with Collins and Bilge (2020), a book that synthesizes decades of intersectionality literature and comes up with six key tenets of intersectionality. For each tenet, we came up with 3-4 questions to guide our paper reading and see what patterns came up.

What did we learn?

Our annotations gave us quantitative data for an inductive analysis and our guided paper-reading and weekly discussions gave us rich qualitative observations for a complementary deductive analysis. We squeezed a lot of juicy findings into our 10-page paper so read that for more, but I will highlight my top 3 below:

Overwhelmingly, AI fairness researchers reduce intersectionality to intersectional subgroup fairness, i.e., showing that an algorithm is fair (for some definition of fair) to Black women to the same degree as it is fair to Black men and to white women. Quantitatively, this finding translates to a high level of engagement with the "complexity" tenet of intersectionality.
Power, relationality and social context were the tenets with the lowest engagement, and these results are related. Papers often neglect to situate their work within a specific social context (country, language, group of users) without which it is difficult to engage in relational thinking - to see the relations between different groups in a society, the decisions we make as researchers, the technical artifacts we produce, etc. Relational thinking means acknowledging our power as AI system builders, something most papers didn't do.
Citational praxis is questionable at best. Some papers define the term but cite no core intersectionality literature, continuing an erasure of Black feminist knowledge in the academy. For instance, one paper defined intersectionality and provided Gender Shades as a citation, a paper that studies intersecting subgroups without ever mentioning intersectionality. But even among papers that use appropriate citations, many then dilute their definitions of intersectionality, neutralizing their (and readers') engagement with power structures and inequality.

Re-imagining the future of intersectionality in AI research

Our main hope is for more precision in how AI fairness researchers talk about intersectionality, through intellectual and citational vigilance when translating scholarly work across domains. This means that we expect AI fairness researchers using intersectionality to centre people at the margins, to build relationally (in community with users), to name their power, and to cite the people these ideas came from. And conversely, when researchers do work that is not about intersectionality but about intersectional subgroup fairness, they should be precise in calling it intersectional subgroup fairness, not intersectionality. Precision and rigour in how we treat intersectionality should be expected at every level - from our colleagues, reviewers, in industry and academia.

When it comes to system-building, smaller, scoped and iteratively built solutions in community with users is better. Computer scientists hate when I say this because our discipline encourages us to build universally, neutrally and quantifiably, but social contexts vary widely, all knowledge is situated, and quantitative and legalistic definitions of fairness are limited in their power (and sometimes even harmful). Combinatorial explosion with intersectional subgroups would not be an issue if we replaced solely quantitative measures of fairness with more expansive, people-centred ones - definitions that might look more like gathering people together for designing, iteratively developing and evaluating if, what, where, why and how any tech solution should be used and built.

In Memory of Jon: Reflections on Crip Linguistics

2023-09-23T00:00:00Z

This blog post is about the paper Unsettling Languages, Unruly Bodyminds: A Crip Linguistics Manifesto, by Jon Henner and Octavian Robinson. Linking critical disability theory, history and linguistics, they describe how linguistics fails to include multi-modal languaging and disabled languagers, from curricula and research to the way we discuss language. They draw connections to other subfields with similar observations and call for us to crip linguistics by centering those at the margins and valuing all types of languaging.

This paper and Jon Henner's presence on the internet have profoundly impacted my thoughts on disability justice, linguistics and myself. I continue to share the crip linguistics paper with disabled friends, both academic and not, and in fact I originally recorded an audio version of the pre-print for a disabled friend who finds it easier to listen to papers than to read them.

Jon passed away on August 14th, 2023, and is sorely missed by pretty much everyone he crossed paths with. There is a celebration of life scheduled for October 14th in Greensboro, North Carolina, USA, and contributions to the GoFundMe (originally for his medical expenses, now to help with the family's medical debt) are welcome. I regret that I wasn't able to get this draft out quick enough for him to read it; it's been languishing in my draft graveyard since 2021, but I am resurrecting it now in his honour. May his memory be a revolution.

The narrowness of status quo linguistics

One of the central theses of crip linguistics is that existing linguistics teaching and applied linguistics continues a long-standing tradition of

marginalizing certain types of languaging and languagers (e.g., modalities other than speech and writing, non-prestige dialectal variation, etc.);
constructing an idealized version of language that disorders some types of languaging (especially disabled languaging); and
separating language from the people who produce it, as if it were coherent to do this

I've seen evidence of the patterns they mention in my own undergraduate linguistics career: my training in analyzing or engaging with signed languages was minimal, and as they say in the paper, I did probably learn more about animal acquisition of gestures than of actual signed languages like ASL, the primary signed language where I lived then. Most of the other modalities mentioned in the paper (Augmentative and Alternative Communication and ProTactile, for instance) were never mentioned in my classes at all. This is rather a shame, because like some speaking autistic people, I also use AAC from time to time, but I only learned of its existence from disabled people on the internet and not from my linguistics degree years earlier.

Similarly, most of my peer linguists-in-training wanted to become speech language pathologists, but I never heard any critiques of speech language pathology, about its aims to "correct" and "fix" languaging. It was taken for granted, as a neutral fact, that some disabled people have "disordered" languaging that needs to be corrected. Henner and Robinson posit instead that it is social perceptions of disability that disorder language use, and that language is "not inherently disordered, although impairments may exist."

Disability and race and gender and...

The idea that no kind of languaging is inherently disordered neatly extends to other forms of languaging that are treated as in need of correction, e.g., gender-associated language (like vocal fry), the entire industry of "accent reduction," (beautifully critiqued in JPB Gerald's book Antisocial Language Teaching: English and the Pervasive Pathology of Whiteness), and more.

Raciolinguistics and queer linguistics also make very similar claims about it being the social perceptions of a group of people that disorders their languaging in the eyes of society. One of my favourite and most fucked up examples of this is Kutlu et al.'s 2022 paper, The Impact of Race on Speech Perception and Accentedness Judgements in Racially Diverse and Non-diverse Groups. They found that listeners behave differently when presented with audio paired with differently-racialized faces - they gave lower intelligibility scores and became worse at speech transcription when the same audio of any English variety (American, British or Indian) was paired with a South Asian face. By drawing parallels between disabled languagers and languagers minoritized along these other axes - race, class, gender, sexuality, etc. - Henner and Robinson show us how we as a field keep replicating society's relegation of all these languagers to the margins.

It is interesting that all of the subfields that point out this power dynamic fall under the umbrella of sociolinguistics. Most subfields of linguistics that don't use a sociolinguistic lens don't name or notice these power dynamics at all, because they often operate on the basis that language production and perception can be separated from the embodiment of the languager producing or perceiving said language. Henner and Robinson, the whole field of sociolinguistics, and almost everyone in a marginalized bodymind, know that this isn't possible without erasures and idealization along the same hegemonic lines as always.

So how do we crip linguistics?

The 3 recommendations that I've pulled out from my reading of the paper are:

Value and embrace all types of languaging and languagers

This means no modality chauvinism (privileging speaking and writing over other modalities), no pathologizing certain kinds of language use, and recognizing that there is a rich ecosystem of what counts under languaging, some of which is actually invisible to abled people, e.g., concepts we know from critical disability theory such as crip time, care work, access intimacy, etc.
Critically examine the ableist assumptions in "practices, attitudes and rhetorics around language"

Among other things, this means recognizing that to separate language from languagers is to leave out several dimensions of how communication actually works in a society, and it leaves you unable to critique the assumptions embedded in doing so.
Use disidentification ("identifying with but not as") to develop coalitional theories

Coalitional theories built through communication across subfields of linguistics would be stronger in their ability to account for all manners of difference. I see the idea of disidentification as closely connnected to another one of my favourite papers - Liu and Shange's Toward Thick Solidarity, which they define as "a kind of solidarity that mobilizes empathy in ways that do not gloss over difference, but rather pushes into the specificity, irreducibility, and incommensurability of racialized experiences."

The final recommendation I want to make to any linguistics educators reading this blog post is to put the crip linguistics paper on your syllabus and to work with students to read all linguistics literature through this critical lens. A crip linguistics lens can be one of several tools in your critical toolbox towards doing better linguistics by paying more attention to gaps we may often take for granted. This paper can help us sharpen our skills in dismantling ableist logic, re-centring languagers at the margins, and incorporating disability justice concepts like care work and crip time into our linguistic practice (and lives!). The paper's final sentence says it best: "disabled people do really cool things with language if people would pay attention."

Note: If you're looking for any of the papers I mention in this post and you don't have access, email me for a PDF.

Birds of 2023

2023-12-25T00:00:00Z

It's getting to the end of 2023, another year I've spent thinking a lot about birds. This post is a random assortment of low-quality pictures of birds I saw and bird-related thoughts I've had this year.

Ruddy shelduck, Eurasian jay, white wagtail, greater scaup.

About half of the photos in this post are of birds that were lifers (i.e., a species I identified for the first time in my life), and the other half are birds I've seen before. The latter is often just as exciting—you know, when it's not mallards—because it's like seeing a familiar face after a long time apart. Some other species I saw but could not photograph are: collared kingfishers, a black-naped oriole, a fieldfare, a great spotted woodpecker, Eurasian kestrels, Eurasian bullfinches, mute swans, bar-headed geese, common ravens, a crested tit, and a great egret.

This year I've been focusing on the aspects of birding that bring me joy - going on long, quiet walks, birding with either my binoculars or my camera but not both, pointing birds out to non-birders, IDing blurry photos from friends, and cataloguing my photos and bird lists. I've also spent a bunch of time learning stuff about birds, mostly from bird descriptions on Wikipedia, All About Birds, eBird and my field guides, but also from the Cornell Lab of Ornithology's Handbook of Bird Biology and a course on the evolution of birds (from dinosaurs!).

Focusing on the fun stuff has meant making my peace with the aspects of birding I'm not really interested in or good at, like getting up early. When I'm out birding, I don't like using my phone, which is why my eBird profile is so sparse. I'm also no longer very good at remembering or caring about facts and figures like scientific names or wingspans. This is the sort of thing I was really into before with cars, but my bird knowledge is much less concrete.

Red-whiskered bulbul, yellow-vented bulbul, zebra dove, common flameback, leucistic blue peafowl.

It's hard for me to describe the kind of bird knowledge I have now because it's so abstract, but I can describe what it has helped me with: I am now much faster and more accurate at looking at a bird and being able to narrow down what order / family / genus it is from. This applies even when I'm in new places looking at birds I've never seen before, or when I'm looking at really blurry pictures from places I've never even been to. It's just easier for me to tell when a beak has finch vibes or when an eye has falcon vibes. Fun fact: what I call vibes is referred to by some birders as a bird's jizz (no, this is not a joke).

As it turns out, there is a scientific basis for this, as this article about how birding changes your brain explains. Experienced birders apparently use the same parts of their brains when looking at birds as they do faces of people. The more you practice, the more you develop neural pathways that get better and faster at recognizing and remembering birds you see in new settings.

What some of the birders in that article describe is exactly my experience: being able to, in a few seconds, synthesize a whole bunch of features ("markings, silhouette, shape, movements, and sounds"), ignoring any extraneous information, and my brain "just converges very fast on what it is." It's rather remarkable to stop and think about because when I started out birding, I was the kind of person who couldn't tell apart a blackbird from a crow. For non-birders: this is like not being able to tell apart a wallet from a tote bag in the same colour. I've been birding for about 6 or 7 years now and it's cool to see how sharply my skills have improved.

Javan myna, ashy tailorbird, spotted dove, house crow.

I tried testing my bird ID skills recently at a birding trivia event hosted by the Cornell Lab of Ornithology. Audience members, a pair of birding experts, and the Merlin Bird ID app were all competing to identify birds. I know I didn't do as well as the experts but I couldn't keep track of my score; I was distracted by some really high-quality discussion of machine learning tech at the event, which I wasn't expecting.

The Merlin app uses machine learning to automatically identify birds through sound or pictures, and one of the engineers who works on it, Sam Heinrich, was called on periodically to explain some of the app's answers. For example, one round of trivia involved identifying an extremely clear picture of a Steller's sea eagle seen in Newfoundland. This particular individual happens to be a rather famous vagrant (hello!), with multiple news articles and even a section on Wikipedia about it. I was able to identify it as soon as it showed up on screen, as were the audience and the birding experts. The app, however, came up with no guesses. Heinrich explained that Merlin can't read the news and is just stuck with what it knows, which is that Steller's sea eagles are extremely unlikely in Newfoundland. This is precisely why this individual bird made the news in the first place - because it is so rare. So while the app can be really good at identifying birds with a high probability of being in your location, it may not be so good at identifying even really distinctive-looking vagrants.

On the other hand, when asked to count the distinct species of birds in a chorus of migration calls, all the humans struggled. The bird experts counted 4 species whereas there were in fact 5 distinct species, and the app even correctly identified all of them. Heinrich said that humans have to zone in on one call at a time to count them, whereas Merlin, with its effectively unlimited attention span, has an advantage in this kind of situation.

I've been raving about this event all year because it's now my favourite example of communicating about the limits and possibilities of machine learning technology, with accessible, non-technical language, and both positive and negative examples of the app's abilities, to demonstrate to users when it's useful and when to just ignore it and rely on your own skills and intuitions. In the end, the bird experts won, Merlin came second, and the collective wisdom of the audience came in last, but we all left more knowledgeable about how best to use Merlin. I left hopeful about a future where we have more birders, more actually useful applications of machine learning like Merlin, and more responsible communication about all machine learning tech like I saw at this event.

Pink-necked green pigeon, blue-tailed bee-eater, oriental magpie-robin, blue peafowl.

On this hopeful note, goodbye to 2023 and happy new year, reader. I hope you see lots of good birds in 2024, and maybe even a vagrant, if you're lucky!

Efficiently teaching computers to know what they don't know

2024-01-01T00:00:00Z

I've hit a PhD milestone - my first first-author paper! This is a non-technical post explaining what it's about.

TL;DR We use simple swaps to turn answerable questions into unanswerable ones, and we use this extra data to create better question answering systems. Our method to create unanswerable questions is better than previous computer-based methods in almost every way, despite being simpler, so new methods should be compared to ours. Simpler is better; don't invent sledgehammers to crack nuts.

Answering questions with computers

Have you ever wondered what the capital of Saarland is?

If you wanted to answer that question, I imagine you would go to the Wikipedia page for Saarland and then perhaps skim the first few paragraphs until you found the answer in the following sentence:

Saarbrücken is the state capital and largest city.

A common way to build question answering systems using machine learning is to follow the same 2-step pipeline:

Picking the most relevant document given the question (the Wikipedia page for Saarland, for instance)
Then extracting the answer by "reading" the document

But what if your system picked the wrong first document? The Wikipedia page for Capital (Marxism), for instance, rather than Saarland.

Or what if your system only had access to documents from an ancient copy of Wikipedia, one with the 2002 entry for Saarland, which didn't contain any information about the capital?

Or what if you wanted to know the capital of Lewis Carroll's Wonderland, a fictional country that doesn't have a (known) capital at all?

All of these cases are failure modes for the typical question answering system, where we would expect it to tell us that it can't answer the question but most systems tend to guess at an answer anyway. Sometimes this may be okay, like in a trivia game with no negative points, but in other cases, like if people were taking health advice from such a system, this could have a serious impact on people's lives.

Teaching computers to know what they don't know

The reason that these systems always attempt to answer questions is because they are built and trained with the assumption that every question can be answered. One way to fix this is to train these systems with negative samples as well, i.e., questions that cannot be answered with a given document.

To train and evaluate question answering systems, researchers have curated lots of high-quality datasets of answerable questions, but very few datasets exist that contain both answerable and unanswerable questions. There are a few ways to augment only-answerable datasets with unanswerable questions, each with its own advantages and disadvantages:

Human annotators - high-quality, but slow and expensive
Machine learning models trained to generate unanswerable questions - fast, but low-quality and needs expensive computers (GPUs)
Simple word swaps - medium-quality, fast and cheap!

Examples of methods to create unanswerable questions

Let's see how these methods compare when they're all given the same document about Bermuda and an answerable seed question.

Document:
The only indigenous mammals of Bermuda are five species of bats, all of which are also found in the eastern United States: ...

Answerable seed question:
What are the only native mammals found in Bermuda?

Other than the human annotators, all methods below are computer-based methods to create unanswerable questions. UNANSQ and CRQDA are two machine learning models specifically designed to generate unanswerable questions, and antonym and entity swaps are the lightweight swaps that we recommend using instead. When a method has more "training parameters," that means it requires more expensive computers to run.

Human annotators^[1]: What is one of five indigenous mammals of Bermuda?

UNANSQ^[2]: what is the only native mammals found in bermuda ?
(relevant and partially grammatical, but answerable and has 1 million training parameters)

CRQDA^[3]: What are the only native mammals are found in?What?
(relevant and unanswerable, but ungrammatical and has 593 million training parameters)

Antonym-swapped: What are the only foreign mammals found in Bermuda?
(relevant, grammatical, unanswerable and has 0 training parameters!)

Entity-swapped: What are the only native mammals found in United States?
(relevant, partially grammatical, unanswerable and has 0 training parameters!)

Creating unanswerable questions with simple swaps

As the examples in the previous section show, antonym-swapped questions are created by swapping a word for its antonym, and entity-swapped questions are created by swapping an entity (i.e., a proper noun for a person, place or organization) with another entity of the same type.

Document:
Saarland is a state of Germany in the southwest of the country. It is the smallest and least populous German state apart from the city-states. Saarbrücken is the state capital and largest city...

Answerable seed question:
What is the capital of Saarland?

Entity-swapped unanswerable question:
What is the capital of Germany?

Answerable seed question:
What is the least populous state of Germany?

Antonym-swapped unanswerable question:
What is the most populous state of Germany?

Reminder: "Unanswerability" here only means that a question is unanswerable given a certain document. The swapped questions are unanswerable only because they are paired with the above document about Saarland.

It's very intuitive to see why swaps create questions that make sense. It also helps that getting antonyms and identifying entities are very well-known tasks in the field of natural language processing. Machines are quite good at this in English, which makes the quality of the swapped data better.

How good are simple swaps?

Data quality

Since our method and the computer-based methods that came before us all involve creating unanswerable questions, we want to evaluate the quality of all these questions and compare them to ones written by humans. Our method's unanswerable questions turned out to be high-quality along 3 axes that are important for question answering:

Less data noise: our questions are unanswerable more often than the other computer-based methods, and almost as often as human-written ones
Higher relatedness: our questions are more relevant to the document
Higher readability: our questions are more human-readable and grammatical than other computer-based methods, but not as good as human-written ones

These results are especially impressive when you consider that I can run our method on my 2017 MacBook Air, whereas the other methods require much more powerful GPUs.

The catch(es)

My post so far paints a rosy picture of swaps but they definitely aren't perfect. One of the main problems is that swaps create heuristically unanswerable questions, i.e., sometimes we accidentally get questions that can actually be answered with the document, but we put them in the unanswerable box.

Document:
On April 4, 2008, Beyoncé and Jay-Z married without publicity...

Answerable seed question:
When did Beyoncé get married?

Entity-swapped unanswerable question:
When did Jay-Z get married? (answerable because the marriage involves both Beyoncé and Jay-Z)

Human annotators do this too but they do it just a little less often than our method does. Human-written questions are also more creative and diverse than a simple method like this. This is a noticeable qualitative difference which we mention in our paper but don't look at in detail, and we don't really know how question creativity impacts question answering systems anyway.

We're also fairly limited in the domains and languages we can work with, because swapping names and antonyms may not work quite so well when dealing with sources of information that are not Wikipedia. Swapping is also very reliant on the quality of tools you have for recognizing entities and finding the right antonym, and also on syntactic features of the English language. If we didn't have high-quality tools for this in English or if we were operating in German, for instance, a lot of things just wouldn't work.

The bigger picture

Recently I've been thinking about broader questions that didn't make it into the paper at all.

Our definition of unanswerability hinges on picking a wrong document in step 1, but what if you had a question answering system that doesn't use the 2-step pipeline at all? ChatGPT doesn't rely on access to a document store or to the internet but simply has all its "knowledge" baked in. In this context, it's interesting to look at unanswerability as a property inherent to the question, e.g., if you had a question that assumes something about the state of the world that isn't actually true.

When was the asteroid that destroyed Australia first spotted?

I've also been thinking about how the world is constantly changing and it is expensive, impractical and plain absurd to up and build a brand new ChatGPT every time a fact changes. There are different, more practical ways to handle this (with varying advantages, disadvantages and degrees of success). I'm curious about how these approaches stack up in the context of question answering, and I've been thinking about the "stickiness" of facts: country capitals change significantly less often than government leaders do. Does this make systems more adaptable to information updates in the latter case?

Quantifying the answers to these questions lets us better understand where these technologies are useful, where a different technical solution might be better, and where we should just keep computers out of the equation altogether. I feel that not enough people are considering this last possibility or thinking critically about the tradeoffs and (social, environmental, etc.) impact of getting a computer to do something, but I'm glad I know and work with many who do. I hope 2024 holds more papers and blog posts about these collaborations - stay tuned!

Minesweeper

2024-04-12T00:00:00Z

Rules: There are mines buried under 10 random tiles in the grid above. To win, you have to uncover every safe tile. Left-click a tile to reveal what's underneath. If it's a mine, you die and have to restart the game by clicking the smiley face on top. Otherwise the title has a number indicating the number of mines that exist in the eight tiles surrounding that tile. Flag a tile as potentially having a mine under it by right-clicking it.

Disclaimer: Nothing good will come from looking at the source. The code is bad because I wrote it 6-7 years ago. All I've done now is to put it on my website and get rid of the jQuery bloat which was cool back then.

Vagrant's Vancouver

2024-05-09T00:00:00Z

This post has been in the making for over 3 years and I need to just get it out now instead of waiting for it to be perfect and complete. Some quick housekeeping:

This list reflects my priorities, which is why it's mostly food
Please look at a map; I include places around the Greater Vancouver Area, not just the city of Vancouver
Some places on my list might have closed since the last time I visited (which for many places was pre-pandemic)
Most places have vegetarian options and are accessible by transit, but I haven't consistently provided accessibility information, dietary restrictions, price points, etc.
Whether you are reading this as a resident of or a visitor to the city, it is worth remembering that you are probabilistically very likely to be a visitor to the unceded (i.e., never given) land rightfully belonging to the Musqueam, Squamish and Tsleil-Waututh First Nations.^[1]

Index

Food

Comfort Western food

Many of these are chains - some Canada-wide, some BC-specific and others just in Vancouver. Other restaurants are standalone. "Comfort Western food" is my catch-all term for what restaurants call comfort food or Pacific Northwest cuisine, or if they just have a menu with burgers and salads and pastas.

Burgoo: My favourite comfort food place. I usually get the after school special (a soup and a sandwich - I get a grilled cheese and tomato soup) or the ratatouille. Great cocktails too.
Burdock & Co: Expensive and really fancy. Apparently they got a Michelin star in 2022!
The Naam: Eclectic vegetarian restaurant.
Tap & Barrel: Great selection of local beers on tap.
White Spot: Another chain with burgers and milkshakes that are truly amazing.
MeeT: Vegan and vegetarian, great cocktails.
The Hub: Good bar food and drink selection. I had the beer Dark Matter for the first time here (it's a great beer - in the same space as porters, stouts and dark ales, but different).
Water St. Cafe: Slightly fancier than most other restaurants in this category (wine glasses are part of the table setting here).
Raglan's: Generous portions.
El Furniture Warehouse: Back in the day, all food used to be 5.95 (and they'd get you with the cocktails, appetizers, etc.) but I don't know if that's still true. Back then it was also called The Famous Warehouse.
The Old Spaghetti Factory: Family-friendly chain (i.e., take your small kids here).
Milestones: It's a chain and it's not that special but it's an option if you want predictability.
Cactus Club: Another not special chain that's an option if you want predictability.

Brunch

Jam Cafe: Yummy and extremely popular all day brunch place. Try their bacon caesar.
Catch 122: Try the huevos rancheros.
Zarak: They have a Caesar flight! This is not a drill!
Rosemary Rocksalt

Cafes and bubble tea

Pretty much all of these are local businesses.

Gallagher's Coffee Bar & Café: My absolute favourite café in the GVA. Great coffee, indoor and outdoor seating with a view of Newport Village with all its local stores and neighbourhood dogs, free wifi, yummy breakfast options.
Renaissance Coffee: Excellent coffee (dare I say better than everything else on this list), located at SFU. Get the 2 samosas and chana masala combo and look out for Mr. Parhar, the friendly owner.
Chatime: Reliably excellent bubble tea chain. My standard drink is a large milk tea with 0% sugar, 50% ice and brown sugar pearls.
Prime's Thai Tea: They have a signature coconut milk and mango drink with coconut jelly that is out of this world.
CoCo Tea: Get the Bubble Gaga just because of the name. It's very summery, with passionfruit, coconut jelly and pearls.
Shiny Tea: Get the strawberry macchiato which has fresh strawberries, milk tea and cheese cream (trust me, it's delicious).
Andrew's on 8th: North Vancouver café with great sandwiches, outdoor seating and mini-golf.
Great Dane Coffee: UBC café with good coffee. Try their grilled cheese. If you're lucky you'll see the owner's great dane!
Juillet Cafe: Cute Burnaby café that won't kick you out if you're there to study or work. They have wifi and a lot of outlets.
Cafe La Forêt: Gorgeous café with a warehouse aesthetic and good pastries.
Fratelli Bakery: Delicious cakes and other baked goods.
The Valley Bakery
JJ Bean Coffee: Chain with lots of locations and reliably good coffee.
Caffe Artigiano
Off The Tracks Bistro: Good breakfast option on Granville Island (but you're spoiled for choice there).
Delany's Coffee: Good West Vancouver coffee chain.
49th Parallel Coffee Roasters: This chain's doughnuts are really good but their coffee isn't.
Chez Christophe
Punk Rock Pastries

Sushi

I won't make restaurant-specific food recommendations in this section because I'm (mostly) vegetarian and therefore not a sushi connoisseur at all. At Japanese places I tend to go for yam tempura, udon and rolls with sweet potato, yam, avocado, mango or tempura inside. These restaurants are ranked in increasing order of fanciness.

Togo Sushi: Least pretentious - food court sushi.
Hong Sushi: As their website says, they "primarily cater to customers on-the-go."
Sushi Bella: Unpretentious sit-down place.
Osaka Sushi: Unpretentious sit-down place.
Sushi Giwa: Mid-level sit-down place I've had many lovely dinners at. I love their yakiudon.
Sushi California: Mid-level to nice sit-down place.
Minami: Fancy and expensive sushi.
Hapa Izakaya

Italian

Vancouver is great for Italian food because of the significant Italian Canadian population on The Drive (Commercial Drive) and in Burnaby Heights. There's no gelato in this section - head to desserts for that.

Cotto Enoteca: Order their tiramisu!
Frankie's Italian Kitchen & Bar: Co-located with Frankie's Jazz Club. With or without the live jazz, you're getting great Italian food from here.
Anton's Pasta Bar: A classic date location especially for students because of the ridiculously cheap prices given the enormous quantities (like 4-5 meals for me). There are long lines outside basically every evening.
Sopra Sotto: All the (wood-fired) pizza here is to die for.
Rosa's Cucina Italiana: Call or go early because this Port Moody spot is extremely popular.
Pizzeria Spacca Napoli: The pizza here is also to die for. I also loved their eggplant parmigiana.
Piva: My favourite Italian restaurant in BC! Have their eggplant sandwich for brunch or their gnocchi for lunch/dinner. They make a killer smoked old-fashioned, if you drink.

Middle eastern

I use this category to combine cuisines where the dishes that I would eat are very similar; basically, every restaurant in this list has hummus and falafel. I indicate how each restaurant describes itself if that information is available.

Jamjar Canteen: Lebanese. Order the makali (deep-fried cauliflower tossed in sweet-sour pomegranate molasses), it is heavenly.
Nuba: Lebanese. I recommend getting one of the sharing options, e.g., Le Feast or Le Grand Feast. They're large and allow you to sample the best of Nuba's stuff.
Anatolia's Gate: Turkish. There is a testimonial on their website that says if the person were on death row this is what they'd want for their last meal. Order the lavash (very big bread) and hummus.
Tamam: Palestinian. Family-run. Exquisite food.
Golden Pita Restaurant: Lebanese. Get a pita wrap.
Babylon café: Middle Eastern. Whatever you eat, get their tea with it (cardamom Alwazah tea)
Chickpea: Middle Eastern. Regardless of whether you go to the food truck or the restaurant, you have to get sabich (fried eggplant) and their hummus (which is what they focused on when they first opened).
Superbaba: Lebanese.
Suren: Iranian/Persian.

Mexican

Original's Mexicano: The avocado tostaditas are fantastic (they're generous with the avocado) as is basically everything else in this place. I recommend the Bloody Aztec to drink.
The Burrow: Previously known as Bandidas Taqueria, this is a Mexican-inspired vegetarian and vegan place. I generally get the Leona Gayle burrito (chipotle tofu, beans, etc.), baked (read: extra cheese), and a jalapeno margarita.
La Casita: You honestly can't go wrong with anything you order here - I tend to go for the quesadillas and a margarita.
Cinco de Mayo: Generous burritos in North Vancouver.
Chipotle Mexican Grill: This is a chain but it's no Taco Bell. Try the sofritas (spiced crumbled tofu) and make sure to get guacamole.
El Santo: A marriage of Mexican cuisine and BC ingredients. This is a loud restaurant and often has live music.
Tacofino: As the name suggests, their tacos are great (and innovative). Each of their locations has a distinct but hipster vibe.
La Taqueria Pinche Taco Shop: They specialize in traditional street taco recipes and have multiple locations now.
Las Margaritas: Lovely atmosphere both indoors and outdoors. Get a margarita!
Guadalupe Taqueria: Located on SFU's Burnaby campus, this place had the best tortillas I've ever had. I used to get the Tree Hugger burrito.

Indian

Tasty Indian Bistro: I've long maintained that this place's name doesn't match its vibe, which is actually quite fancy. In my opinion they have the best Indian food in the GVA. Because I am a paneer fiend, my favourite dishes here are the shahi paneer and the vegetable biryani. The mango lassi is also great.
Agra Tandoori: This place also has great Indian food (I recommend the paneer pakoras) and its Burnaby location meant that it was easier for me to get takeout or delivery from here since Tasty was too far away.
Chettinad Dosa Palace: Hole-in-the-wall joint to get your dosa fix. They also have other kinds of Indian food including chaat.

Desserts

Nuvola Gelato: I lived near this family-run gelato place for 3 years and got into the habit of waving at the guy through the glass every time I walked by. Their offerings are seasonal but try to get the mango sorbet (vegan) and the dark chocolate sea salt flavours if you can. Also their tiramisu is heavenly.
Nero Waffles: Delectable, instagrammable waffles that are loaded with stuff. There are savoury and sweet options and my personal favourite was the tiramisu waffle with its espresso mascarpone cream.
Glenburn Soda Fountain: They specialize in classic sundaes and have a classic soda fountain atmosphere. I got really into their milkshakes.
La Casa Gelato: Classic Vancouver location whose USP is the variety of icecreams they stock - over 200 flavours at a time, including unusual ones like durian, balsamic vinegar, and so on. Trying flavours is encouraged.
Rain or Shine: They're all about natural, local and seasonal ingredients. Their salted caramel icecream is really good (even for a caramel-disliker like me).
Earnest Ice Cream: They're also all about natural, local, seasonal ingredients, and they're a particularly good option for vegans.
Rocky Point Ice Cream
Cartems Donuterie: I'm not a big donut person but I've heard great things about the smoked maple walnut and the earl grey donuts.
Snowy Village: Go to this Korean dessert café for bingsoo and taiyaki.
I.ce Queen: Go here for rolled icecream - half the fun is in watching it be prepared.
Cheesecake Etc.: Only open at night for cheesecakes and live jazz.
Dairy Queen: Getting a DQ Blizzard is a very classic thing to do. After the DQ person makes it they'll typically flip it upside down to show you how cold it is before handing it to you right-side-up. My favourite kind is probably chocolate chip cookie dough.

Ramen

Hot Pot

Boiling Point: Taiwanese hot pot - my favourite!
Happy Lamb Hot Pot: Previously Little Sheep Mongolian Hot Pot.

Chinese

Dinesty Dumpling House: They are famous for their xiao long bao, i.e., their steamed pork soup dumplings. If you're a vegetarian, there is stuff for you to eat here but this unfortunately does not include vegetarian soup dumplings.
Lao Cai: Food from the Chinese province of Xi'an. Get the hand-ripped spicy oil noodles, dumplings and the crispy lotus root salad.
New Town Bakery

Fast food

I've left American fast food chains off this list and limited it to Canadian chains.

Freshii: I often get the Buddha bowl with tofu at the airport.
Fresh Slice: It's pizza! Get what you like.
Uncle Fatih's: They make pizza, but all their pizzas are sprinkled with sesame seeds!
Smoke's Poutinerie: If it's your first time, get the traditional (or veggie traditional) poutine before you experiment with their more fun variations on poutine.
A&W: If you like root beer (nonalcoholic beverage that tastes kind of like toothpaste) that's a classic thing to get at A&W. Get sweet potato fries and chipotle mayo with your burger - they're yummy.

Other

All the singletons are stuck here rather than having their own section.

Chau Veggie Express: Vegan Vietnamese place with great noodle bowls and rice bowls - I recommend the Rickety Rickshaw Bowl.
Brokenrice: Vietnamese.
Chillipepper House: If you're not familiar with Indian Chinese food, go here. I'm a vegetable manchurian ball fan, personally.
Steve's Poké Bar: Authentic Hawaiian poké! I cannot recommend anything specific since you build your own bowl but I will recommend that you are generous with the sauces.
4 Stones Vegetarian: Get as much of these dishes as you can reasonably fit in your tummy: gyoza, the kong pao bean curd rolls, the noodles with sesame sauce, and the Taiwanese deep fried tofu.
Fassil: At this Ethiopian place, everything you order will come with injera, a fermented and soft flatbread. Pick one of the combos so you can try various dishes along with your injera.
Peaked Pies: Australian savoury and sweet pies - has been vetted by a bona fide Australian.
Banana Leaf: Malaysian.
Chad Thai: Thai.
BookKyung Ban Jeoun: Korean.

Books

Most of these places do used and new books and also have good sell-for-store-credit policies. You can place orders for almost anything you want by walking in, phoning or emailing if something isn't available online. Also many of these places have cute non-book merch. Support local bookstores!

Iron Dog Books: My favourite Vancouver bookstore. Indigenous-owned, independent, previously itinerant (they were just a book truck before they opened a brick-and-mortar store). Hilary will, um, read you like a book and give you fantastic recommendations. I like their hand-picked selection of everything but especially queer / sci-fi / fantasy / Indigenous writing. Also go here if you like YA.
Companion Books: I've had good conversations about birds here. Sci-fi / history are big in this store, and paperback bestsellers if that's your thing. They also do "rare and antiquarian" books but this has never been relevant to me.
MacLeod's: This second-hand and rare bookstore looks like it's out of a book - their collection of books isn't just displayed in bookshelves but in several feet high piles on the floor and on tables, with very narrow paths for customers to walk through. There are so many books crammed into this space that MacLeod's has literally clashed with the fire department for a decade over their fire safety.
Pulpfiction Books: Wide selection of used and new books, especially sci-fi and fantasy. They will order books for you at very discounted rates so order your books from here.
White Dwarf: This store specializes in fantasy and sci-fi, and back in the day they had a Basset hound who would hang out in the store.
Massy Books: Also Indigenous-owned and has a great selection of various genres. They also host lots of great events, especially with local artists and authors.
Paper Hound: Small, eclectic bookstore with a poem dispenser and funky shelf names including my personal favourite: Rat as hero.
Little Sister's: This is a bookstore that's very important in Vancouver's queer history. It offered a safe haven for queer people to go in the 80s when homophobia was more mainstream and overt. Over the decades, they've survived multiple incidents of homophobes throwing literal firebombs into the store, a battle with Canadian customs over discriminatory seizures, a 10-year-long trial about this where they sued the Government of Canada (and won), and the deaths of the original founders - Jim Deva and Bruce Smyth. Located in the extremely gay neighbourhood of Davie Village, they stock lots of queer books and other queer stuff (pride merch, clothing, sex toys, condoms and dental dams, etc.)
People's Co-op Bookstore: A volunteer-run member-owned leftist bookstore with a wide selection of genres and a poetry reading group.

Alcohol

This list includes breweries and bars along with beer recommendations with the warning that my taste in beer skews hoppy and dark (IPAs, porters, stouts). My favourite liquor store gets an honourable mention at the end.

Steamworks: A local (Gastown) brewery with the prettiest cans and bottles. Try the flagship hazy IPA for hoppy goodness.
Yellow Dog: All their beers have dog-related names - my favourite is their Play Dead IPA. They have a huge closed off patio in Port Moody where you can drink in the sun! People used to even bring their dogs but this is no longer allowed because of "local health regulations."
Parkside: Another great Port Moody brewery. I recommend the Dreamboat hazy IPA.
Twin Sails: This is the Port Moody brewery to go to if you enjoy a good stout (though they have other stuff as well). They brew new beers regularly but I distinctly remember having good milk stouts, coffee stouts and oatmeal stouts from Twin Sails.
Granville Island Brewing: The Island Lager is a classic and fits Granville Island's vibe really well.
Main St. Brewing: Good locally-brewed beer outside of Port Moody! Their rotation is always changing so I can't recommend anything specific here.
Alibi Room: The perfect place to go if you want to sample various different local beers; they update their list regularly so there will be new (but still local) craft beer every time you visit.
Good Co. Bars: Previously known as the Colony Bar, I've spent many lovely nights at their location on Main St. Get their beer-battered fries (now "lemon pepper fries" on the menu).
Per Se Social Corner: A pricey place with gorgeous cocktails. I recommend their Grounds for Divorce, a variation on a gin fizz. This was also the place where I was first introduced to dried citrus.
Angry Otter Liquor: This liquor store is a hidden gem with knowledgeable staff and an enormous variety of beer, wine and spirits. Their North Burnaby location fuelled many months of cocktail experiments and I recommend Ungava gin for Canadian (but not local) gin. If you're a beer person and you think you'd enjoy something between a porter, a stout and a brown ale, look for my favourite beer - Hoyne Brewing's Dark Matter.

Museums

Science World: A science museum housed in one of the most recognizable buildings of the Vancouver skyline - a geodesic dome. Great for kids and nerdy adults.
Museum of Anthropology: This museum on UBC campus has a lot of artifacts from First Nations peoples and cultures of the Pacific Northwest, including totem poles outdoors and a few famous Bill Reid sculptures. They also have plenty of stuff from other parts of the world but learning about First Nations cultures and histories is a more Vancouver-specific thing to do here on a visit.
Museum of Archaeology and Ethnology: A small museum at SFU that's affiliated with the university's archaeology department. Their collection focuses on First Nations in British Columbia.
Vancouver Aquarium:
Howard Trottier Observatory: Starry Nights with shared binoculars and telescopes from the community and of course, the observatory.
H. R. MacMillan Space Centre: The real fancy space centre.
Canadian Museum of Flight
Burnaby Village Museum
Vancouver Art Gallery
Steveston Tram: restored tram from the early 20th century.
Britannia Shipyards National Historic Site: collections of buildings turned into a museum about the oldest BC shipyard community.
Beaty Biodiversity Museum
Chinese Canadian Museum

Outdoors

This section covers parks with good scenery and hikes, beaches, birding, and some notable sport-adjacent activities that are fun to do in Vancouver. As Vancouver is in the Pacific Northwest, most beaches are rocky beaches unless otherwise specified. This means it can hurt a lot to walk barefoot, so take sandals.

Lighthouse Park: Tied for my favourite hike, or a close tie with Buntzen lake.
Belcarra Regional Park: Many nice trails including an easy 4 kilometre loop around Sasamat lake and the longer Jug Island Beach hike. White Pine Beach is sandy, but the rest of Belcarra's beaches are rocky.
Burnaby Lake Regional Park: The park where I've spent the most time. Very easy trails but it can also get busy. Lots of cheap thrills in terms of birding, including a vagrant mandarin duck!
Buntzen Lake: The trail around the lake is tied for my favourite hike. This is relatively flat compared to other Vancouver hikes, but still extremely hilly for someone from a flat place. It still takes the average person (and me) about 3 hours. The Diez Vistas trail is harder and longer and while some of the 10 vistas are fantastic, others are a bit disappointing.
Quarry Rock: A not very pretty hike in Deep Cove (North Vancouver), until you get right to the top where the views are incredible. Until then you're going basically straight uphill though.
Stanley Park: A Vancouver classic. You can walk the seawall or rent bikes.
Burnaby Mountain Conservation Area: Lots of great trails including the Velodrome trail and the Pandora trail. The park in particular is very popular at sunset and has a number of carved wooden poles by Japanese aboriginal people.
Rocky Point Park: The shoreline trail is beautiful and goes through some great marshy areas and forested bits and has really photogenic boardwalks.
Burns Bog
Queen Elizabeth Park
VanDusen Botanical Garden and Bloedel Conservatory
Deer Lake Park
Vanier Park
Cates Park
Capilano Suspension Bridge Park: Unlike most other parks, this one has an entrance fee, and the highlight is of course the suspension bridge.
Dr. Sun Yat-Sen Chinese Garden: A very peaceful place with traditional tea experiences you can pay for.
Other birding-focused locations: ƛ̓éxətəm Regional Park (pronounced tla-hut-um; previously Colony Farm), Reifel Bird Sanctuary, Deboville Slough, Maplewood Flats Conservation Area, Minnekhada Regional Park
Beaches: There are many to choose from. Jericho Beach is my favourite sandy beach. English Bay, Spanish Banks and Sunset Beach are other options that are also relatively sandy. Rocky beaches include Ambleside Beach and beaches in Belcarra.
Kayaking: You can rent kayaks in Deer Lake, Jericho Beach, Main St (near Science World), Deep Cove, Port Moody, and Cates Park. This is in order from (what I think is) easiest to hardest for first-timers.
Water biking: Imagine a bicycle on pontoons that you ride on the water - that's a water bike. It's ridiculous but also really fun to do with friends in good weather because you can have conversations while water biking.
Lights to the Lodge: Snowshoeing in the evening with LED lights on Cypress mountain.

Events, venues, festivals

Look here if you're into music, film, theatre, etc.

Vancouver International Film Festival: The festival happens at venues all around town. They screen films at the VIFF Centre year-round as well.
Vancouver International Jazz Festival: Performances at venues around town, some ticketed, some free, some on outdoor stages. They also have year-round events.
Bard on the Beach: Shakespeare festival where plays are put on in large tents set up in Vanier Park. It's not just traditional performances, e.g., a spin on Coriolanus where they made the character a woman and used it to explore themes of gender and violence. Pro-tip: if you aren't familiar with the play, read the synopsis of the original before you go. Pro-tip 2: go to the pre-show talks regardless of your familiarity with the original Shakespeare because the talks focus on the production you're about to watch.
Vancouver Hot Chocolate Festival: Festival where cafés, bakeries, chocolatiers, etc., all come up with signature hot chocolates and you can go around town sampling them!
Vancouver Mural Festival: New murals are commissioned every year in different neighbourhoods of Vancouver and you can use their online maps to plot a path and go look at murals.
Queers and Beers: When I used to go it was a small meetup that would take over most of a bar. I think they've grown bigger since!
Vancouver Queer Film Festival: Queer film and filmmakers at venues all around town for a month! They also have year-round events.
Vancouver Pride: Year-round events but of course it reaches fever pitch during Pride (July/August in Vancouver). Pro-tip: look out for the Unicorn Ball in the spring (my favourite Vancouver event) and the Dyke March, not just the parade.
Biltmore Cabaret: Music venue.
Fox Cabaret: Music and event venue.
The Cobalt: Music venue.
Rio Theatre: My favourite indie movie theatre.
The Cinametheque: My second favourite indie movie theatre.
Frankie's Jazz Club: Jazz and blues venue.
Vancouver Mural Festival: Happens every year and you can go around watching murals being created during the festival or just do a mural crawl using the map at other times of the year.
Imperial: Theatre, music and event venue that has hosted Vancouver Pride Society's Unicorn Ball in the past, one of my favourite events.
Nerd Nite: Drink and learn! Nerdy talks for a general public, one of my favourite things to attend, especially when people I knew were presenting.
Food Truck Festival: Food trucks galore! Look out for the grilled cheese food truck, that's my favourite.
Vancouver Symphony Orchestra: They do plenty of classical music, of course, but the Movie Nights with the VSO series is my favourite. You get to watch a film with the VSO playing the orchestra live. It's the most incredible experience and I cannot recommend it enough.
Playland: Amusement park right in the middle of the city with Canad's oldest rollercoaster, a wooden rollercoaster.
Twilight Drive-In: A drive-in movie theatre - Vancouver's only one! It's a bit far out but it's such a fun and canonically North American (in my mind) experience.

Getting around

Public transit: Skytrain (light rail), buses, and the seabus, which you should definitely ride between Waterfront and North Vancouver if you're into transit. Public transit is a great way to get around if you can. You can use a credit card to tap on at Skytrain and seabus stations and on buses, or get a Compass Card at a Skytrain or seabus station (not on buses).
Driving: I recommend using a car share company for short trips, e.g., Modo for round trips or Evo for one-way trips. Both options tend to be cheaper than longer-term car rental companies like Hertz.
Ferries: False Creek Ferries has trips within Vancouver, along False Creek. BC Ferries handles trips between Metro Vancouver and places that are further out, e.g., islands, central BC, northern BC, etc.
Air: Apart from Vancouver International Airport, you can also do a heli-tour with SKY helicopters or a seaplane trip with Harbour Air, which is a delightful experience.
Taxis: I recommend using good old taxi services like Yellow Cab or Bonny's Taxi.

Donations and volunteering

If you want to make material donations or donate your money, time or skills, this is the section for you. My recommendations are places I've donated to before, and I chose them because most of them work to address current social, financial and structural inequalities that stem from historical oppression. They do this through a combination of direct aid, free or discounted access to services like medical or legal aid, food or shelter, etc. The two notable exceptions to this on my list are Canada Learning Code (where I previously volunteered to teach kids how to code; these were mostly affluent kids, often with one or two parents in tech already), and Value Village (which is just a secondhand store and one that is very important for people who are not well-off).

A note on the Downtown Eastside: This is the area of Vancouver around the intersection of Hastings St and Main St. It is known for high rates of homelessness, drug use, mental health issues, crime, etc., a lot of which are consequences of austerity-based policies introduced in the Harper era that led to many people being deprived of social support they needed. Withdrawing these support mechanisms disproportionately affected certain demographics of people who are already at risk due to intergenerational trauma (from colonization, family history, etc.), which is why the area has, for example, a high proportion of homeless Indigenous people. The Downtown Eastside is thus a focus area for many of the organizations on this list which work to try and fill gaps in social support. They treat Downtown Eastsiders like human beings, which many tourists and even local residents fail to do. Please treat your fellow humans with respect and if you are willing and able, donate to these organizations generously.

Downtown Eastside Womens Centre: The DEWC accepts donations (I've mostly donated clothing before) and they also take volunteers. They provide a space for women in the Downtown Eastside to hang out and get health checkups and food and such.
Atira: Atira runs the DEWC and various other projects to help women who have experienced or are experiencing abuse, homelessness, poverty, etc. Although I find their wording on this a bit clunky they are trans-inclusive in their definition of women.
PHS Community Services Society: I read Gabor Maté's book about addiction which highlights many stories of people dealing with addiction in the Downtown Eastside, the "worst" of whom (i.e., those who are rejected by all other community programs) are still taken in here and provided with shelter, food, therapy and such. This organization provides a model for what addiction care should look like, unlike the war on drugs policies of various Vancouver mayors in recent history. They're doing essential work and deserve all the financial support you can give them.
Indian Residential School Survivors Society
Quest
Vancouver Overdose Prevention Society: Addicts don't deserve to die. I love an addict. Please donate mnoney and/or supplies to the OPS so they can continue to save lives in the Canada's overdose crisis.
Qmunity: This organization builds queer community in Vancouver by running youth groups, providing consultation and financial help with accessing services like gender-affirming procedures (name changes, etc.) as well. They're worth donating to and also checking out if you're a queer person (especially a queer youth) looking for community.
Rainbow Refugee: This organization helps queer refugees seek asylum in and/or immigrate to Canada, leaving behind countries where they are persecuted for being queer. They were the beneficiary of Perform for Pride 2017, where I was co-emcee.
Value Village: If you want to buy anything secondhand, this is a great bet that isn't the Salvation Army. They also accept most items as donations, from clothing and furniture to kitchen stuff. Look at their website for details.
Canada Learning Code: I volunteered here for a few years teaching kids to code. There are lower- and higher-commitment volunteer options; I opted for the former, i.e., to be a mentor (walking around and helping individual kids with problems) rather than a session leader (crafting a 3-hour program for a session to teach kids how to build something in code). Chances are higher than average that you code if you know me, so this is a place to consider volunteering, and you don't have to work with kids - there are also sessions to teach adults how to code, in various languages and with various goals.

Shopping

For food and neat things.

Miscellaneous

Granville Island: Great for a day trip involving food, a public market, art and other events, shopping, etc.
Steveston: Steveston is a fishing village that is featured often in film and TV. There are several good museums and it's a nice place to walk by the Fraser river.
FlyOver Canada: Some might say this is a bit overpriced but I think it's really fun. You "fly" over Canada in something like a 4D movie.
Big Bro's Barbershop: Trans hair place!
B-Bombshell Salon: Geeky hair place!
Adrenaline Studios: Tattoos and piercings and such.
Womyns Wares: Sex toys, etc.

Further out

These are nearby cities that are good for a day trip or longer.

Whistler (Mogul's Coffee House, La Cantina, Cow's Ice Cream)
Victoria (Victoria Bug Zoo, Hoyne Brewing Company, Miniature World)
Seattle (Zeitgeist Cafe, Left Bank Books)

Now closed

RIP some of my favourite places that are now closed.

The District
The Wallflower
Table153
Soirette
Charlie's Mexican Restaurant and Catering
Oysi Oysi
Zeta Cafe
PopBar
Godiva Chocolatier
Thai Café
Yaletown Distillery
Browns Books
Albion Books
Cafe Deux Soleils
Ingrid's Village Cafe
Hakkaku Ramen

If you aren't a visitor by this definition, thank you for having me on your land for 6 years. I miss it. ↩︎

Protect a disabled NLPer; don't come sick to EMNLP!

2024-10-08T00:00:00Z

Although I haven't previously talked about it online, many of my colleagues and friends know that I have a chronic lung condition. In my childhood, colds could escalate into hospital visits, and as a teenager, I couldn't do the same sports as other kids. As an adult, it's been mostly controlled, except since COVID appeared. I am at higher risk to catch it and also to have complications^[1] from it, so my life has changed a lot since 2020 - I rarely eat at indoor restaurants, I don't go out to indoor bars and clubs anymore, and I mask on public transit and other indoor spaces like conferences where everyone is sharing air.

Unfortunately, at my first in-person conference (EMNLP 2023 in Singapore), there were a lot of sick people who showed up anyway, which was clear from the coughing and sneezing you could hear in every room. As conferences are crowded and in closed spaces with poor ventilation^[2], they provide perfect conditions for illness to spread and for disabled scientists like me to end up with additional disabilities.

This year, I will be attending EMNLP again (in Florida this time) to present four papers^[3]. I like having opportunities to talk to people about my work and learn about theirs, which is why I want to attend EMNLP in the first place, despite generally minimizing my conference travel. People in the NLP community can take two simple steps so I can be an equal participant:

Please don't show up sick, whether you have a cold, the flu or COVID.
Please wear a mask (as I will be), as all of these are also transmissible before you start to show any symptoms. In fact, your wearing a mask protects me better than my one-way masking.

Hope to see you there! I'll be the one looking like this, plus or minus the sunglasses:

See this story about the effects of COVID reinfection and this Nature review article on long COVID ↩︎
In about 10-20 years I feel very confident that high-performance air filtering will become a standard in most buildings and public spaces, just like other sanitary services like clean water and sewage systems, which will greatly reduce transmission of infectious diseases and remove other allergens I am sensitive to. ↩︎
A TACL paper I led on pronoun fidelity with language models, a main conference paper I co-led with Arjun Subramonian looking at "democratization" in machine learning and NLP, a main conference paper Marius Mosbach led on the impact of interpretability and analysis work on NLP, and a (CRAC) workshop paper I led on pronominal bias in coreference resolution. ↩︎

(My!) class on concepts in NLP

2024-10-11T00:00:00Z

I have an up-to-date reading list on GitHub, and a more formal and citable write-up about this course here:

Teaching and Critiquing Conceptualization and Operationalization in NLP
Vagrant Gautam.
Preprint, 2025.

I'm thrilled to be teaching next summer at Saarland University, and not just as a co-instructor on an existing course! I designed a seminar to get Masters students comfortable with conceptual work, something that I'm really enjoying doing these days (see our work on intersectionality, names, and democratization). The course is research-focused and reading-heavy (for computer scientists) and I'm looking forward to the discussions we'll have, because I think conceptual work is critical to think about where we are and where we're going in NLP. I've chosen a selection of concepts in subfields that I'm familiar with—fairness, interpretability and reasoning—but I welcome additional suggestions for conceptual/critical papers. If you're an author of one of the papers I've selected and would be interested in joining our discussion, I'd be thrilled to hear from you.

Syllabus

Introduction

NLP papers commonly use various abstract concepts like "interpretability," "bias," "reasoning," "stereotypes," and so on. Each subfield has a shared understanding of what these terms mean and how we should treat them, and this shared understanding is the basis on which datasets are built to evaluate these abilities, metrics are proposed to quantify them, and claims are made about systems. But what exactly do these terms mean? And, indeed, what should they mean, and how do we measure that? These questions are the focus of this seminar on defining and measuring abstract concepts in NLP.

In two week cycles, we will cover various concepts in NLP research from the selection below. We will read papers that analyze or critique how a given concept is used, and then we will use this as a lens to read, discuss, and critique three recent NLP papers that use that concept. We will also try to reimagine how we would run these projects and write these papers in light of what we have learned.

Course Requirements

Present a concept and lead a discussion about it (a sample will be given of what this should look like) – 35%
Engage with presentations about other concepts – 30%
Write a report designing an NLP project involving the concept that is either novel or a reimagination of one of the papers we saw in the seminar – 35%

Learning Objectives

This course will help you acquire / practise the following skills, among others:

How to read and critique papers (both interdisciplinary and more conventional NLP papers)
How to critically evaluate aspects of conceptualization (defining an abstract concept) and operationalization (creating empirical measures of the abstract concept), towards answering questions like: What does X mean? How do we translate our conceptions of X into something that we can observe and measure? How, concretely, can we measure it—with data and metrics and gold labels and an evaluation strategy? What should X mean? How do our choices in defining, conceptualizing and operationalizing X lead to gaps in how we make claims about X?
How to design NLP projects in ways that address critiques and push the discipline forward

General Reading

List of Concepts

We will choose a subset of the following topics based on class size and student interest: robustness, stereotypes, interpretability, explainability, paraphrases, bias, race, emergent abilities, gender, generalization, names, reasoning. I may add further topics, including understanding, intelligence, accountability, memorization, trust, etc., depending on interest. Each concept is listed below with a paper that does a critical or conceptual analysis (sometimes with a survey) of it. In addition, other papers that use the concept are listed here for us to discuss and evaluate.

Robustness: Beyond generalization: a theory of robustness in machine learning^[8]

Stereotypes: Stereotyping Norwegian Salmon: An Inventory of Pitfalls in Fairness Benchmark Datasets^[12]

Interpretability: The Mythos of Model Interpretability: In machine learning, the concept of interpretability is both important and slippery^[16] and Against Interpretability: a Critical Examination of the Interpretability Problem in Machine Learning^[17]

Explainability: Explanation in artificial intelligence: Insights from the social sciences^[20]

Paraphrases: What Is a Paraphrase?^[24]

Bias: Language (Technology) is Power: A Critical Survey of "Bias" in NLP^[28]

Race: A Survey of Race, Racism, and Anti-Racism in NLP^[33]

Emergent abilities: Are Emergent Abilities of Large Language Models a Mirage?^[37]

Gender: Theories of "Gender" in NLP Bias Research^[41] and Gender as a Variable in Natural-Language Processing: Ethical Considerations^[42]

Generalization: A taxonomy and review of generalization research in NLP^[45]

Names: Stop! In the Name of Flaws: Disentangling Personal Names and Sociodemographic Attributes in NLP^[49]

Reasoning: Back to Square One: Artifact Detection, Training and Commonsense Disentanglement in the Winograd Schema^[53]

Arjun Subramonian, Xingdi Yuan, Hal Daumé III, Su Lin Blodgett ↩︎
Christina R. Steidl, Regina Werum ↩︎
Clémentine Fourrier ↩︎
S. Keshav ↩︎
Jason Eisner ↩︎
Maureen A. Carey, Kevin L. Steiner, William A. Petri J ↩︎
Josef Fruehwald ↩︎
Timo Freiesleben, Thomas Grote ↩︎
Shiyue Zhang, Vishrav Chaudhary, Naman Goyal, James Cross, Guillaume Wenzek, Mohit Bansal, Francisco Guzman ↩︎
Lalchand Pandia, Allyson Ettinger ↩︎
David Esiobu, Xiaoqing Tan, Saghar Hosseini, Megan Ung, Yuchen Zhang, Jude Fernandes, Jane Dwivedi-Yu, Eleonora Presani, Adina Williams, Eric Smith ↩︎
Su Lin Blodgett, Gilsinia Lopez, Alexandra Olteanu, Robert Sim, Hanna Wallach ↩︎
Akshita Jha, Aida Mostafazadeh Davani, Chandan K Reddy, Shachi Dave, Vinodkumar Prabhakaran, Sunipa Dev ↩︎
Moin Nadeem, Anna Bethke, Siva Reddy ↩︎
Eddie Ungless, Bjorn Ross, Anne Lauscher ↩︎
Zachary C. Lipton ↩︎
Maya Krishnan ↩︎
Jing Huang, Zhengxuan Wu, Christopher Potts, Mor Geva, Atticus Geiger ↩︎
Asma Ghandeharioun, Avi Caciularu, Adam Pearce, Lucas Dixon, Mor Geva ↩︎
Tim Miller ↩︎
Sarthak Jain, Byron C. Wallace ↩︎
Sarah Wiegreffe, Yuval Pinter ↩︎
Yanai Elazar, Shauli Ravfogel, Alon Jacovi, Yoav Goldberg ↩︎
Rahul Bhagat, Eduard Hovy ↩︎
Jan Philip Wahle, Terry Ruas, Frederic Kirstein, Bela Gipp ↩︎
Jan Philip Wahle, Bela Gipp, Terry Ruas ↩︎
Saket Sharma, Aviral Joshi, Yiyun Zhao, Namrata Mukhija, Hanoz Bhathena, Prateek Singh, Sashank Santhanam ↩︎
Su Lin Blodgett, Solon Barocas, Hal Daumé III, Hanna Wallach ↩︎
Mihir Parmar, Swaroop Mishra, Mor Geva, Chitta Baral ↩︎
Maarten Sap, Swabha Swayamdipta, Laura Vianna, Xuhui Zhou, Yejin Choi, Noah A. Smith ↩︎
Maarten Sap, Saadia Gabriel, Lianhui Qin, Dan Jurafsky, Noah A. Smith, Yejin Choi ↩︎
Seraphina Goldfarb-Tarrant, Rebecca Marchant, Ricardo Muñoz Sánchez, Mugdha Pandya, Adam Lopez ↩︎
Anjalie Field, Su Lin Blodgett, Zeerak Waseem, Yulia Tsvetkov ↩︎
Tom Bourgeade, Alessandra Teresa Cignarella, Simona Frenda, Mario Laurent, Wolfgang Schmeisser-Nieto, Farah Benamara, Cristina Bosco, Véronique Moriceau, Viviana Patti, Mariona Taulé ↩︎
Maarten Sap, Dallas Card, Saadia Gabriel, Yejin Choi, Noah A. Smith ↩︎
Nicholas Deas, Jessica Grieser, Shana Kleiner, Desmond Patton, Elsbeth Turcan, Kathleen McKeown ↩︎
Rylan Schaeffer, Brando Miranda, Sanmi Koyejo ↩︎
Jason Wei, Yi Tay, Rishi Bommasani, Colin Raffel, Barret Zoph, Sebastian Borgeaud, Dani Yogatama, Maarten Bosma, Denny Zhou, Donald Metzler, Ed H. Chi, Tatsunori Hashimoto, Oriol Vinyals, Percy Liang, Jeff Dean, William Fedus ↩︎
Sheng Lu, Irina Bigoulaeva, Rachneet Sachdeva, Harish Tayyar Madabushi, Iryna Gurevych ↩︎
Peiyu Liu, Zikang Liu, Ze-Feng Gao, Dawei Gao, Wayne Xin Zhao, Yaliang Li, Bolin Ding, Ji-Rong Wen ↩︎
Hannah Devinney, Jenny Björklund, Henrik Björklund ↩︎
Brian Larson ↩︎
Andreas Waldis, Joel Birrer, Anne Lauscher, Iryna Gurevych ↩︎
Aniket Vashishtha, Kabir Ahuja, Sunayana Sitaram ↩︎
Dieuwke Hupkes, Mario Giulianelli, Verna Dankers, Mikel Artetxe, Yanai Elazar, Tiago Pimentel, Christos Christodoulopoulos, Karim Lasri, Naomi Saphra, Arabella Sinclair, Dennis Ulmer, Florian Schottmann, Khuyagbaatar Batsuren, Kaiser Sun, Koustuv Sinha, Leila Khalatbari, Maria Ryskina, Rita Frieske, Ryan Cotterell, Zhijing Jin ↩︎
Elron Bandel, Yoav Goldberg, Yanai Elazar ↩︎
Niklas Muennighoff, Thomas Wang, Lintang Sutawika, Adam Roberts, Stella Biderman, Teven Le Scao, M Saiful Bari, Sheng Shen, Zheng Xin Yong, Hailey Schoelkopf, Xiangru Tang, Dragomir Radev, Alham Fikri Aji, Khalid Almubarak, Samuel Albanie, Zaid Alyafeai, Albert Webson, Edward Raff, Colin Raffel ↩︎
Swaroop Mishra, Daniel Khashabi, Chitta Baral, Hannaneh Hajishirzi ↩︎
Vagrant Gautam, Arjun Subramonian, Anne Lauscher, Os Keyes ↩︎
Sandra Sandoval, Jieyu Zhao, Marine Carpuat, Hal Daumé III ↩︎
Haozhe An, Rachel Rudinger ↩︎
Fatemeh Torabi Asr, Mohammad Mazraeh, Alexandre Lopes, Vagrant Gautam, Junette Gonzales, Prashanth Rao, Maite Taboada ↩︎
Yanai Elazar, Hongming Zhang, Yoav Goldberg, Dan Roth ↩︎
Vagrant Gautam, Eileen Bingert, Dawei Zhu, Anne Lauscher, Dietrich Klakow ↩︎
Zhaofeng Wu, Linlu Qiu, Alexis Ross, Ekin Akyürek, Boyuan Chen, Bailin Wang, Najoung Kim, Jacob Andreas, Yoon Kim ↩︎
Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed Chi, Quoc Le, Denny Zhou ↩︎

Studying names, pronouns and gender with large language models

2024-11-19T00:00:00Z

Pronouns and names

Vagrant told Vagrant's friend that Vagrant's landlord made Vagrant's landlord's famous onion soup, and that Vagrant's landlord's famous onion soup tasted great.

That was probably painful to read, right? That's why we have names and pronouns!

Vagrant told Vlad that xyr landlord made his famous onion soup, and that it tasted great.

We use names to refer to a specific individuals directly in conversation, and we use pronouns as shorthand to stand in for a noun or a noun phrase, including names. Especially when talking about people, this is useful so we don't have to repeat someone's job or their relation to us over and over again.

We even have conventions around when to use different types of referring expressions while referring to the same person. Consider the two examples below, which are identical except for a single highlighted word:

Tomorrow I'm going to catch up with a friend I haven't spoken to in a while, Xanda. She works at Harvey Mudd and I've always been a fan of her cool hair colours.

Tomorrow I'm going to catch up with a friend I haven't spoken to in a while, Xanda. She works at Harvey Mudd and I've always been a fan of Xanda's cool hair colours.

The first clearly sounds better, doesn't it? That's because I already switched to pronoun mode for Xanda when I told you she works at Harvey Mudd and so you expect that to continue, even if you didn't know that consciously. When I instead switch back to using Xanda's name in the second example, it's unexpected and can even be a bit confusing.

Pronouns, names and gender

If you have no linguistics background, you probably didn't consciously know that you had these intuitions about when to use a name and when to use a pronoun. Indeed, both names and pronouns do a lot of linguistic heavy lifting without us paying much attention to them. Consider the following sentence:

Rory ran into Jess and he told her about his work.

Who told whom about his work? You might have inferred that the 'he' user is probably a man, that 'Rory' is more likely to be a man's name than 'Jess' (often a nickname for Jessica) is, and therefore that the most likely scenario is that Rory told Jess about his work. All those inferences and connections happened in the timespan of you reading that sentence.

We make these inferences about social gender because names are often assigned or chosen in gender-associated ways, and because third person singular English pronouns^[1] mark grammatical gender which is also often associated with a social gender. Our inferences about social gender via names and pronouns are actually very important to create and follow conversations about people in a way that disambiguates different individuals.

However, the heuristics we use to infer gender can sometimes trip us up, since neither names nor pronouns map one-to-one to gender. You will likely already know this if you know queer/trans/non-binary people.^[2] More broadly, we're more likely to infer wrongly when we lack context about the people and the situation. For example, if you've watched Gilmore Girls like I just did, you might recognize the names in my previous example and realize that Rory (Lorelai Gilmore) is a girl and Jess (Jess Mariano) is a guy.

Pronouns, names, gender, and natural language processing

So, to summarize, names and pronouns help us keep track of people in conversations, without having to specify someone's relation to us over and over again, we have a lot of complex conventions in how we use them to create and follow discussions about individuals, and they frequently signal things about gender, which can sometimes trip us up.

As someone who works on natural language processing (NLP, or making computers do things with language), some questions that I'm interested in are how NLP systems deal with names, pronouns and gender. I find this interesting from a number of perspectives, including

Fairness, i.e., wanting systems to treat people fairly and not cause harms, including misgendering and misnaming people
Linguistics, i.e., how do systems create or understand coherent discourse involving the use of names, pronouns and other referring expressions (e.g., someone's occupation or relation to us)
Meta-level research practices, i.e., when we want to measure things about "gender" or "race" and instead use pronouns and names as a stand-in for that, how much do we miss?

Three of my papers this year explore the answers to some of these questions. I hope to write a post about each of them for which this post will serve as a general introduction, but for now, here are the questions each paper tackles:

When do names say something about a person's gender, race, and other sociodemographic characteristics, and how should this inform our use of names to infer these characteristics?

Stop! In the Name of Flaws: Disentangling Personal Names and Sociodemographic Attributes in NLP
Vagrant Gautam, Arjun Subramonian, Anne Lauscher, Os Keyes.
Workshop on Gender Bias in Natural Language Processing, 2024.

How well do large language models reuse pronouns for an individual when we're talking about two people who use different pronouns?

Robust Pronoun Fidelity with English LLMs: Are they Reasoning, Repeating, or Just Biased?
Vagrant Gautam, Eileen Bingert, Dawei Zhu, Anne Lauscher, Dietrich Klakow.
Transactions of ACL, 2024.

How reliably can large language models map different kinds of pronouns to the right people? How should we map pronouns to (social) gender?

WinoPron: Revisiting English Winogender Schemas for Consistency, Coverage, and Grammatical Case
Vagrant Gautam, Julius Steuer, Eileen Bingert, Ray Johns, Anne Lauscher, Dietrich Klakow.
Workshop on Computational Models of Reference, Anaphora and Coreference, 2024.

In English, it's just third person singular pronouns that mark grammatical gender. In German, nouns and adjectives carry this information too, and Italian even marks gender on verbs. ↩︎
I use xe/xem and they/them pronouns and I chose the name Vagrant, an English noun with no obvious gendered associations, to convey my not-a-girl-and-not-a-boy-ness. ↩︎

2024 in Review

2024-12-26T00:00:00Z

Hello from Vancouver, where it is surprisingly warm! I'm spending Christmas with loved ones and reflecting on what's been the fullest year of my life so far. Here's what happened.

Work

I was on 7 papers (3 first author, 2 co-first, 3 second), I gave 10 talks, I did a research visit in Hamburg, and the Heidelberg Institute for Theoretical Studies awarded me their Independent Postdoc fellowship! To start this position in 2025, I have to wrap up my PhD a year early, so I'll probably be writing my thesis for the foreseeable future.

Despite managing to do everything I did this year, I somehow have more imposter syndrome than a year ago! Objectively, I know more things because I've read lots of papers and books, and I've even created new knowledge. But all of this has also revealed that there are things I didn't even know I didn't know, and in so many different areas. As Weizenbaum has said, I "stand in awe before the depth of [my] subject matter." I am humbled by it.^[1]

At the same time, I am enjoying research more than ever, because learning is just so much fun. It helps to have collaborators who are as knowledgeable and generous as mine—particularly Anne, Arjun, Marius, and Zeerak.

Non-work

My evenings, weekends and holidays have been just as busy. I got to show eight visitors around Saarbrücken this year, and I travelled to see people I love in other places. I also moved, after which I've been cooking more, watching more TV, and making and drinking lots of coffee with our new espresso machine.

2024 was a year of some interesting firsts, such as my first cat cafe, fondue, and bath bomb. I also tried bouldering for the first time and it felt horribly unnatural (so I might do it again). One friend taught me how to drive manual, and then I completely forgot their lesson by the time I drove another friend's dad's tractor for the first time out in the German countryside. I also tried his automatic lawn mower after. I was pretty awful at all of this but everyone was quite nice about it. This year I also bought a sketchbook and made some mediocre visual art, mostly with ink and watercolours. There were also some firsts that I was actually good at: I watched Grey's Anatomy (yes, all 19 seasons), I finally managed to donate blood, and I visited a bunch of places on vacation for the first time -- Cochem, Trier, Heidelberg, Kaiserslautern, Kiel and Schwerin in Germany, and New York, Brussels, Ghent and Bruges outside. I hit two big milestones this year beyond firsts -- finishing all my name change bureaucracy after 2.5 years, and finishing the Duolingo German course after 3 years and change.

I've been worse than usual at tracking what birds I saw this year, but here are the lifers I remember: In northern Germany I saw my first tufted ducks, horned grebes and barnacle geese, and in southern Germany, I saw my first goldcrest and bullfinch. I also saw a flock of white ibis while in Miami for a conference. Some birders and I made plans to ditch the conference in favour of birding on a couple of mornings, and we saw a night heron, brown pelicans, common and boat-tailed grackles, a Muscovy duck, and a blue-gray gnatcatcher, most of which are lifers for me. One nice thing about this year has been the opportunity to share birds with other people more than usual. In both Saarbrücken and Vancouver, I've enjoyed showing people moorhens, Egyptian geese, tits, cormorants, coots, pintails, long-billed dowitchers, green- and blue-winged teals, scaup, and mergansers. Apart from these live birds, I saw hundreds of dead ones in natural history museums in Basel, New York, Kiel, Berlin, and more, and found typos in at least two museums, which made me feel smug.

Intentions for 2025

My work goals are clear: I want to write fewer first author papers, but do more risky work. After my PhD, I want to think and learn more about advising students, picking and evaluating research ideas, and reinventing myself and my research as fields and funding priorities change. One explicit intention I have is to read more mathy and theoretical machine learning papers. Here are some example paper titles: Sparse Continuous Distributions and Fenchel-Young Losses, Wrapped β-Gaussians with compact support for exact probabilistic modeling on manifolds, DAG Learning on the Permutahedron.^[2] I find it impossible to read this kind of work at the moment, because I'm unfamiliar with the conventions and feel intimidated by all the scary-looking new words. But that's also how I felt about critical theory, philosophy, etc., before making a concerted effort to read work in these domains over the last few years. I'm only semi-ashamed to admit it took me 2-3 years to understand the word epistemology; you have to start somewhere.

Outside of work, I hope to continue to find joy in trying new stuff and being bad at it, spend enough time resting and tending to my relationships, and see more people and places on vacation.

Happy almost new year, and see you on the other side!

I can't recommend this enough: Weizenbaum, J. (1972). On the Impact of the Computer on Society. ↩︎
I'm so sorry I can't understand half your work, Vlad. ↩︎

Faculty job interview questions

2025-04-18T00:00:00Z

General

Who are you? Tell me about yourself
Why do you want this position?
Why do you want to be a faculty member?
What do you know about our organization?
What do you bring? What is your greatest strength? What are your relevant strengths?
How do you see your role evolving in the next five years within the academic community?
What sort of equipment would you need?
Who are you looking to hire right out of the gate: post-doc/technician/grad student? And what kind of person would you look for?
Can you summarize the contribution you would make to our department? / How do you see your work fitting into our department?
How will you set yourself apart from your current PI?
Why are you interested in the department?
How quickly would you want your lab to grow?
What size do you envision for your lab, and makeup of graduate vs undergraduate?
What kinds of students/trainees do you envision recruiting to your research program?
What attracted you to apply for this position?
Tell us about your background as it relates to this position, pointing out key reflection points or moments where you either: 1. grew or developed a new skill or, 2. needed to create or repair a relationship with a colleague.
Provide an example of where you needed to collaborate with your colleagues on a project and what was the result. Did you face any challenges?
Tell us about a time when you have taken steps to ensure that everyone in your class or lab felt included. Describe the situation, the actions you took, and the outcome.
Tell us about a time when you responded to a colleague or student who made an insensitive remark. Describe the situation, the actions you took, and the outcome.
Why did you choose this profession/field?
Describe how your career has been enhanced by exposure to diverse people, places, or experiences.
Have you ever had a great idea but been told that you could not implement it? How did you react? What did you do?
What are three leadership competencies you possess? Please provide an example of when you have used these to provide effective leadership during a difficult time.
Have you ever experienced a situation in which you had difficulty getting people to accept your ideas? What was your approach to handling these situations?
What practices do you implement during conflict situations and are they successful?
Describe a situation in which you did all the right things and were still unsuccessful. What did you learn from the experience?
Tell us about your preferred work environment.
Describe your ideal job.
What things have you done on your own initiative to help you prepare for your next job?
Can you describe how you go about solving a problem? Please give us some examples.
Tell us about a time when you had to deal with a conflict at work. Describe the situation, the actions you tool and the outcome.
Describe a time when you helped to resolve a conflict between two colleagues or students.
What are one or two of your proudest professional accomplishments?
Do you have any concerns that would make you have reservations about accepting this position if it is offered to you?
What do you think most uniquely qualifies you for this position?
On the basis of the information you have received so far, what do you see as the major challenges of this position and how would you meet them?

Research

Tell me about your research interests.
Please give us a 5 minute presentation of your research plan.
Describe your current research (or your dissertation). What is the broader significance of your research?
How do you see your research fitting in with the department?
Where are the points of uniqueness for you compared with your PI or others in your field?
What is your research agenda?
How would you situate your work in the field?
Tell me about your research plans for the next two/five/10 years?
Do you think of your work as translational or having translational applications?
How do you contribute to the community outside of your institution?
Can you tell us about a project that you're currently working on?
How do you stay current with the developments in your field of expertise?
Could you provide an example of a successful research project you have led or been a part of?
Please tell us about your research interests and how you see them fitting with the
department's priorities.
What grants have you received and what plans do you have to fund your research?
What types of resources would you require to successfully continue your research agenda?
How would you involve undergraduate and graduate students in your research?
What changes do you see on the horizon in your area of research?

Mentoring/recruitment

Can you describe your approach to mentoring students?
How would you encourage students to pursue a major in our department?
How do you motivate students to become more engaged in their own learning process?
How would you recruit students to take courses with us?
How would you recruit graduate students?
How would you mentor graduate students? What experience do you have?
How do you see adapting your research so that it fits on short-term scale of Honors Thesis work or summer student?
How do you contribute to a collaborative work environment within your department?
Our university is working hard to integrate more undergraduate students in research activities. How could you contribute to that mission?

Teaching

Tell us about your teaching philosophy.
How do you manage a classroom?
How do you handle academic dishonesty?
How do you adapt your teaching to different learning styles?
How do you assess and measure student learning?
Describe a course you have taught in the past and how you measure the students' learning.
How do you incorporate the use of technology in the classroom?
What classes will you be able to teach with little preparation?
What is your experience working with diverse student populations?
How does your research inform your teaching?
Describe your teaching goals and/or challenges?
How do you implement inclusivity in your classroom structure?
Can you discuss a time when you had to adapt your teaching style to accommodate different learning preferences?
Describe your worst teaching moment. And the best?
What strategies do you use to engage students who are less participative in class?
How would you handle a situation where a student challenges your authority or expertise in the classroom?
Describe how you frame your syllabi theoretically.
In what ways do you incorporate student feedback into your teaching practice?
What methods do you use to ensure that you are an accessible and approachable faculty member?
Describe how you would develop a new course or curriculum in your area of expertise
How would you teach?
- a first-year seminar
- an Honors class
- a writing-intensive class
- an upper-level course in your field
- an introductory or survey class
- a hybrid advanced undergrad/MA level class
- a graduate-level course
How do you engage students in extracurricular education?
How do you feel about teaching classes online / in the evening?
What is your approach to giving feedback to students on their academic work?
Can you describe a course that you would want to teach?
What strategies do you employ to ensure academic integrity in your courses?
Discuss a time when you had to modify a pre-existing course. What changes did you make and why?
Describe your teaching style and teaching philosophy.
What do you think are the most important attributes of a good teacher?
Describe your experience in teaching and mentoring students from diverse cultures and backgrounds.
Tell us about a time when you have taken steps to ensure that everyone in your class or lab felt included. Describe the situation, the actions you took, and the outcome.
You're faced with the following situation. You've started teaching a class and ten minutes in you notice that some of the students are distracted. What would/did you do?

Service

What institutional issues particularly interest you?
How will you contribute to the community at our department and institution?
All faculty are required to participate in service activities. How do you see yourself getting involved in service on campus?
Would you be willing to serve as the faculty mentor for [insert student organization here]?
In our department, the department chair role rotates every three years. How would you begin preparing to be department chair?
Tell us about what leadership experience you have and what leadership activities you'd like to undertake at our university.

Funding

Where do you think your research fits in terms of funding sources?
What do you see as sources of funding?
Can you provide an example of a grant or funding proposal you have written and how it benefited your research or department?
Describe your experience in writing and preparing grants. What do you consider the critical factors in this process?

Interpersonal

Discuss an ethical dilemma you have faced in your teaching or research and how you resolved it
How will you balance research and teaching moving forward?
How do you handle disagreements or conflicts with colleagues?
How would you support students who are struggling academically or personally?
How do you balance your teaching, research, and service responsibilities?
What experience do you have with interdisciplinary teaching or research?
Tell me about a time you dealt with a difficult situation
Tell me about a time you failed
Tell me about a time you disagreed with a colleague
Can you give an example of a time when you had to handle a difficult student or situation?
How do you measure your success as a faculty member?
Can you describe a time when you collaborated with other faculty members on a project?
How do you respond to feedback from students, colleagues, or supervisors?
Can you describe a time when you innovated in your teaching or research?
How do you balance your personal and professional life?
Do you have any questions?

Interviewing to be a professor

2025-04-22T00:00:00Z

I recently interviewed for a (tenure-track assistant) professor position in computational linguistics—in Vancouver, even!^[1]—and although I didn't get the job, this has really been the professional highlight of my year so far. Briefly, the timeline was:

November 2024: Submitted application materials
December 2024: Got long-listed
January 2025: Had a 30-minute, structured Zoom interview with 6 professors; got short-listed for an in-person campus visit
February 2025: Got flown out for a 2-day campus visit (including meetings with undergraduate and graduate students, curriculum committees, 2 lunch interviews, a dinner interview, and 2 talks - a research-focused colloquium talk and a pedagogy talk/discussion)
March 2025: Rejected (womp womp)

Since I benefited greatly from others generously sharing their materials with me (and giving me feedback on mine), I am sharing all the materials I applied and interviewed with^[2], with the institution redacted:

Cover letter (2 pages)
Research statement (4 pages + references)
Teaching statement (3 pages)
Curriculum vitae (as of 2024)
Research talk slides (40-45 mins talk + 15 mins Q&A)
Pedagogy talk slides (15 mins talk + 45 mins discussion)

I also previously posted this list of interview questions for faculty jobs that I compiled and used to organize my thoughts before the long-list interview and the campus visit, with the hope that it's helpful.

The absolute best part of the whole experience was getting to have good old-fashioned nerdy fun: talking phonology with a phonologist, doing literal back-of-the-napkin syntax, getting to introduce a student to crip linguistics, learning about effort-based grading, etc. It was particularly gratifying to do a job talk in Vancouver, in front of academics I've admired for years and people I care about. Getting shortlisted has totally upstaged my PhD defense for this reason, because it likely won't be timezone-friendly for Vancouverites 9 hours behind German time.

Beyond the fun it was also really useful to have the experience for potential future applications, and putting materials together early helped me clarify the overarching story connecting all my research and see which parts of my application were weakest. Once I'm done with my thesis and start my postdoc in a few months, I'll have a chance to apply all that to my research program going forward!

Home! Here's a list of things to do in Vancouver if you ever visit. ↩︎
Note that I applied at the end of 2024, so the application materials reflect my work / CV at that point in time and are already outdated. ↩︎

NLP Concepts Seminar: A Retrospective

2025-07-15T00:00:00Z

I have an up-to-date reading list on GitHub, and a more formal and citable write-up about this course here:

Teaching and Critiquing Conceptualization and Operationalization in NLP
Vagrant Gautam.
Preprint, 2025.

The summer semester is nearly over, and that means my seminar on defining and measuring abstract concepts in NLP is also almost done! The class exceeded my expectations: My students were very active in discussions, bringing very different perspectives to class from their personal, professional, and academic experiences, and we all learned a lot. And according to the course evaluations, the students seemed to really enjoy this class, in spite of the fact that I assigned four (4!) times as much reading as most other seminars here.

I previously posted my concepts syllabus; in short, it's a class I designed to explore abstract concepts such as interpretability, bias, etc., in terms of:

How they are currently conceptualized and operationalized (i.e., defined and measured) in the field -- by reading 3 popular/recent papers per concept
What these choices might miss -- by reading 1 interdisciplinary critique / conceptual paper

After some general reading on conceptualization, operationalization, and how to read papers, the concepts we ended up covering were: Names, interpretability, reasoning, stereotypes, paraphrases, emergent abilities, explainability, and bias. I presented names, interpretability, and stereotypes, and the rest were paired student presentations.

What went well

Assignments

Before each session, I had students submit a short (3-5 bullet points) graded assignment reflecting on the four readings. I explicitly allowed them to write more if they were feeling particularly inspired, and I consistently had many students take me up on this right from the start. In my instructions, I asked students to synthesize ideas across all the papers and critique the concept papers or point out what the authors did especially well. Some students had trouble with this at the beginning, instead just summarizing what each paper did (e.g., some students went too low-level with their critique, so I gave them feedback to think of bigger, more fundamental problems with how a concept was being treated in a paper). But by the end of the term, all of my students were submitting solid assignments: Sometimes they brought up points that I know other people have written about, and I enjoyed sending them paper recommendations in my responses to their assignments. They also made many connections in these assignments that I hadn't thought of and that were really fun for me to engage with.

Discussion

Since it was a seminar, I wanted the class to be discussion-heavy. I also wanted the discussion to be high quality, and the assignments were my approach to achieve this, encouraging students to do the readings even if they weren't presenting. Overall, I was very happy with the quantity and quality of the discussion -- students made connections across sessions and concepts, and connected their own experiences (personal and professional) to the topics we were discussing. Discussion-heavy classes can sometimes be challenging for students who are shy or less confident in themselves, so I tried to ensure that everyone got a chance to speak. I also explicitly told student presenters that they would be graded on their classroom management and on allowing everyone to participate in discussions, because I believe this is an important skill, regardless of whether it's in a classroom in an academic setting or running a meeting in industry. This worked exceedingly well, and a third of my student evaluations explicitly mentioned it ("Welcoming atmosphere, easy to participate in discussion," "[The lecturer] always made sure that everybody was included in the discussions," "Learning atmosphere during our sessions allowed me to speak freely and engage in discussions which is sometimes difficult for me"), which I was thrilled to hear.

Cohesion of the readings

Interpretability, explainability, and reasoning were concepts that worked especially well, in my opinion. Some concepts also overlapped (e.g., interpretability and explainability, as well as names, bias, and stereotypes), which was really nice for making connections across sessions and concepts. For explainability, we read a critique of AI explainability that surveyed social science papers on explanation from various disciplines, which helped ground our understanding of explanations in the other papers. We also read 'Attention is not Explanation' and 'Attention is not not Explanation', which worked well together because the latter is a response to the first, and itself a critique. Interpretability was another concept with two critique papers, 'The Mythos of Model Interpretability' and 'Against Interpretability', which provided great perspectives to ground our reading of Patchscopes and RAVEL, two rather technical papers. Reasoning, on the other hand, didn't have an explicit critique paper, but it still worked well because we covered a range of types of reasoning (pronominal, spatial, logical, mathematical, etc.), as well as recent approaches in the literature (chain-of-thought prompting and DeepSeek's reasoning chains). This diversity gave us a good bird's-eye view of work on reasoning and helped us find similarities, differences, and missing pieces.

Pain points

Connection between assignments and sessions

The assignments and sessions weren't well connected in a few different ways. First, the student presenters never had access to their peers' assignments, because I wanted students to be able to do their readings and submit their assignments right before class instead of significantly earlier. Secondly, this also sometimes meant that some really great points from the assignments were never discussed in class, just because they happened not to fit with the discussion prompts the presenters came up with. Since I had access to all the assignments as the instructor, I was much more aware of this gap than the students were, but at least one student did tell me that they felt there sometimes wasn't space in the sessions to discuss thoughts they had written down in their assignments. So next time, I want to try to have a tighter connection between the assignments and the sessions -- perhaps by moving the assignment deadlines up to incorporate parts of them into the session, or by having an open-ended component to the sessions where students could bring up observations we hadn't already addressed.

Cohesion of the readings

There were a couple of papers that I found simply boring or a slog to read, which I would remove next time, and one conceptual paper (What is a paraphrase?) that should either be replaced with a critique paper or removed altogether. At a higher level, since I prioritized student preferences for presentation dates, concepts that are closely connected ended up being far away from each other on the schedule. This made it hard to disentangle the significant overlap between interpretability and explainability, for example. Next time, I'd love to combine related topics, extend the time we spend on them, and add the following readings which I mentioned in class and would provide excellent additional context to our discussions:

Explainable AI: Beware of Inmates Running the Asylum Or: How I Learnt to Stop Worrying and Love the Social and Behavioural Sciences^[1]: A short and sweet version of the more comprehensive Miller survey paper I assigned
Mechanistic?^[2]: A history of mechanistic interpretability that provides much needed context to understand this subfield
The elephant in the interpretability room: Why use attention as explanation when we have saliency methods?^[3]: A nice critique of using attention as explanation that is parallel to existing readings on attention as explanation and nicely links interpretability and explainability
Learning to Faithfully Rationalize by Construction^[4]: The 'Attention is not explanation' and 'Attention is not not explanation' authors later collaborated on this paper, which would be nice to round out the whole thread

Workload

According to the course evaluations, roughly half the students thought the requirements and workload were just right, and half thought it was a little demanding. This was something I was concerned about, but since student satisfaction was so high on the overall metrics and since not a single student selected the "too demanding" choice in any of these questions, I'm not sure whether I should actually make any changes here. I'm reluctant to assign less than 4 papers per concept because I still feel that it's necessary for the kind of high-level discussions I wanted us to have about a concept, so perhaps having fewer concepts and spreading them out over a longer period would be something to consider. As for my workload, the individualized and extensive feedback I was giving was fun but far too time-consuming to scale to a group larger than 10 students. I struggled to find time to do it amid submissions and other work, until after submitting my thesis.^[5] So if I wanted to run a class like this on a larger scale, I would need to restructure the class as well as my feedback.

Summary

Overall, I think the seminar went swimmingly. The students who did all the readings read 34 papers and got a new appreciation for some pressing problems in NLP. They turned into skeptics and good conceptual thinkers by the end of the course, paying attention to both the social and technical aspects of NLP, and how decisions about conceptualization and operationalization can impact all of our science as well as public perceptions of NLP technology and tools. At the beginning of the course, I told them I would feel like I had succeeded at running this seminar if I could get them to come up with new ideas that I hadn't thought about or read about before; they did this over and over again at every session, and I'm so proud of them!

Tim Miller, Piers Howe, Liz Sonenberg ↩︎
Naomi Saphra, Sarah Wiegreffe ↩︎
Jasmijn Bastings, Katja Filippova ↩︎
Sarthak Jain, Sarah Wiegreffe, Yuval Pinter, Byron C. Wallace ↩︎
Oh, yeah, I submitted my thesis! ↩︎

Notes from Germany, part 2

2025-10-23T00:00:00Z

Soon after moving to Saarbrücken, I wrote my first post about the move, Notes from Germany, part 1. Four years later, I realized I never did a part two. Since I just moved to Heidelberg, now felt like a good juncture to write it. This was a larger scale move than the last one despite being within Germany, because it involved transporting a whole apartment full of furniture and stuff. But it felt a lot easier, partly because I could make my German roommate handle phone calls in German, and partly because my own German is so much better now. That's what the rest of this post is about (my German, not the roommate).

Between Saarland University and DeutschAkademie, I've taken a bunch of German classes^[1] in the last 4 years, and recently finished a B2.2 course. I've also done Duolingo basically every day since I moved here, and in about 20 days I'll hit a 1500 day streak on Duolingo, but after that I'm going to quit. I've been through the Duolingo German course twice, once the normal way and once on legendary mode, but I definitely maxed out on its usefulness a long time ago. In my case, all of this training translates to a level where I can talk about my politics and my work, do all my various medical appointments (GP, skin, gyno, dentist, etc.) exclusively in German,^[2] and—my personal favourite—tell off a man who tried to make fun of my mask until he apologized. Basically, being able to speak German at this level has massively improved my quality of life in Germany. I'm not confined to my little research bubble and can talk to real people in my community about real things. And it also lets me connect with Germans at a much deeper level, even ones with whom I talk English most of the time.

As a nonbinary person, however, German has been kind of challenging. The obvious issue I keep running into is that German is aggressively gendered and has no convenient singular-they equivalent. "Wie gendert man dich?" (How do I gender you?) is a question I regularly get, to which I have no perfect answer. Also, all the possible solutions (pronoun avoidance, alternating genders, neopronouns like dey, xier, per, etc.) are ones I learned from non-binary Germans, not the German classroom. It is well-established that the language classroom is full of various cultural ideologies, including what kinds of genders, relationships and family structures exist and are intelligible to other people, ideologies that leave no room for people like me. The curriculum also typically assumes that learners have a sense of belonging to a single country that they're from and know things about, whereas I'm a literal Vagrant with a fraught relationship with like three different countries I'm plausibly "from." Most of this isn't necessarily very important for immediate needs like getting groceries, and I feel lucky to have had wonderful teachers who were open to hearing that I was non-binary and did not "correct" the way I choose to gender myself in German. But there just isn't room in a status quo classroom to teach me queer German slang, or to help me express myself in a way that reads as queer to other queer people, which is the German I need long term.

The other set of language ideologies at play as an L2 learner are related to standardization. We get taught high German, which does next to nothing for my understanding of the gardener or my roommate's parents or my friend's conversations with his football team, because all of these people speak in various dialects we get no training in. And when we speak, whether it's in the classroom or outside, we get praise for approximating the "standard" accent as closely as possible. Don't get me wrong, I'm only human and I love a compliment (and recently I've been getting a lot of them on my German!). I just feel sad as a linguist about the hierarchy it reflects, where my more convincingly "native" accent in German gets me more points than someone who sounds different but can express themself in a much wider range of circumstances than I can.

Once again, I have so much respect for L2 English learners at a level I really didn't have before moving to Germany and learning German. I have a whole new appreciation for this highly memed scene from Modern Family:

I'm funny and smart and sexy in English, and in German, I'm a shadow of myself. It's probably a big part of why I still want to move back to Vancouver despite all the linguistic and social (and professional!) progress I've made here in Germany. Until then, though, Heidelberg is a beautiful city, and I have to admit I'm enjoying it much more than Saarbrücken.

Viele Grüße an alle meine Lehrerinnen! ↩︎
This can actually be kind of involved, because as I said to my GP recently, I'm almost 30 and I have 10,000 new mystery pains each week. She thinks my posture is the culprit... (She's probably right). ↩︎