What am I up to now?

November 1, 2024

March, 2023

I’ll be in Oxford throughout this month, as revision begins to heat up for the April mega-exams. These three exams — micro, macro, metrics — are the most important procedural requirement for my PhD program. The bar is quite low — we only need to secure a Pass to move on, and everyone in the first class passed last year. None of us know, nor have the courage to ask, what happens if we don’t pass one of the exams.

In once sense, I haven’t studied for an exam since 2018; in a much more real sense, I haven’t studied for an exam ever.I probably studied most intensely for the GRE, which I found wholly unrewarding. My strategies at the moment are rather scattershot — I make Anki decks, I ask ChatGPT to prove the Bondareva-Shapley theorem using linear programming, I pester Qingyu until he explains LASSO estimators. From 10 March until the first exam about five weeks later, I’ll have to be more focused in my efforts. This will be helped by the lack of new material each week. Right now, revising feels like trying to swallow a bucket of water being refilled by a firehose.

Going through old material, ChatGPT has become embarrassingly invaluable. In lecture, I have a crippling inability to ask stupid questions, which can leave me floundering.Sure, maybe no one else knows what it means for a cooperative game to be ‘balanced,’ but they’re not asking, so I must be the dumb one. Better just stay quiet. But ChatGPT can’t judge! I’ve taken to keeping a ChatGPT tab open in lecture alongside the slides, and asking it about points of confusion.It’s helpful to ask ChatGPT to keep its responses shorter. Also, it writes very good quality LaTeX. Then, after a lecture, I’ll review my conversation and use it to fill in my notes. Last term, if I was confused about a definition or step in a proof, I’d mark the slide and come back to it later. This is much more efficient.

Out of some masochistic curiosity, I pasted a recent econometrics problem set into ChatGPT. It would have scored perfectly on all sub-parts of Question 1, and gotten partial credit on most of Question 2. Even when the LLM gets caught on some tangent — on Question 2a, ChatGPT completely mixes up the sample and the population quantities — it is able to provide insight into the structure of the problem. All of the results from 2a were useless, but the model “understood” the setting thoroughly enough to answer 2b,c,d correctly.Neither ChatGPT nor Joseph Levine had any chance on 2e. We both found the algebra irritating and deceivingly complex.

The existential dread from seeing an LLM do well enough to pass a PhD-level econometrics course was… underwhelming. I should be feeling a lot more something than I am. Most of my friends would advocate for “fear” or at least “apprehension.” I’m also confused by the shallowness of my curiosity about how the heck this works.Not that shallow, really. More coming on this soon (i.e., after exams).

I don’t have any opinions worth reading about Sydney not borrowed from Zvi. I have found PoE much more useful than ChatGPT for getting into the weeds of economic theory; PoE lies to me less and cites its sources better. But only being available for iPhone is a dealbreaker for the moment. ChatGPT remains my go-to.

Reading
Eating
Listening

Reading

Especially-good books I read in February:

•”Computing taste: algorithms and the makers of music recommendation,” by Nick Seaver. I don’t know who Kevin Baker is, and I don’t remember subscribing to his newsletter, but the recommendations are consistently top-notch. Kevin recommended “Computing Taste,” and I downloaded it off of the strength of the title. It’s a sociologist’s take on companies which sell music recommendation algorithms, with digressions into the history of music recommendation and more interesting technical problems. Of niche interest.
•”A primate’s memoir: a neuroscientist’s unconventional life among the baboons”, by Robert Sapolsky. A mix between a neuroscientist’s fieldwork memoir and a critical look at the sociology of baboons. Despite Sapolsky’s excellent raconteur-age, I ended up skipping more and more of the memoir chapters towards the end. The baboons steal the show — I was both thrilled and saddened to learn that the species in the book is of Least Concern conservation status, and are considered pests even by primatologists. Sapolsky’s passion for the animals is infectious; one reviewer wrote that “it’s hard not to love what he loves, when he loves so well.” Despite this, baboons continue to… suck? They take all the petty aggression of humanity and turn it to 10, with just enough of the good qualities necessary to sustain a somewhat functioning social group. Loveable nonetheless.
•”Arguing about justice: essays for Philippe Van Parijs”, collection. This book was intended as a birthday present for Van Parijs from his academic friends, a Belgian political theorist most associated with UBI. I downloaded it to get a single essay, titled “If Marx or Freud had never lived?” Then I got sucked in by the table of contents. Other highlights included “Why has Cuban state socialism escaped its 1989? Reflections on a non-event;” “Love not war. On the chemistry of good and evil;” and “Lamentation in the face of historical necessity.”
•”The war that doesn’t say its name: the unending conflict in the Congo”, by Jason Stearns. I wish there were more books like this. The Congolese Civil WarWhich is Not A Thing, according to Wikipedia has been on-going since before I was born; how do we explain decades of low-level fighting? Continued fighting is in no-one’s interest, but governments, rebels, and international organizations continue to instigate. The role of Rwanda in egging on the conflict is particularly egregious. The book has is based on years of field work in eastern CongoGosh I’m jealous. and has strong theoretical grounding. Stearns outs himself as a constructivist early on and I was immediately thrown back to my single year of international relations lectures at St Andrews.
•The most important thing I read this month was Joe Carlsmith’s essays on doing ethics as an anti-realist (1, 2). I’ve been calling myself an anti-realist total utilitarian to anyone who asksNot many people. since at least 2017; I always felt that the anti-realist part of that identity rendered the total utilitarian part a “merely” aesthetic preference. The first essay describes the tension I feel very well, in language I endorse. Carlsmith discusses two reasons why anti-realists should do ethics, one of which I used to endorse strongly (consistency seems important); the other I’m very confused about (money pumps; I read Johan Gustafsson’s excellent book on the topic last year and am making progress.But besides being fascinating, money pumps don’t seem worth doing ethics for. I just don’t think anyone is ever going to find a short, tractable, intransitive cycle and pose it to me. ) Some fun quotes below, to convince you to read. Joe writes well.

“The curve is just: whatever the curve god-damn needs to be, thank you. Or if that doesn’t work: screw the curve. I’m a no-curve dude. Going with my gut; with common-sense; with “what my body wants”; with that special, rich, heuristic wisdom I assume I bring to each decision. Metis, you know? Coup d’oeil.

…

“[T]his approach doesn’t allow for a very simple or compact description of your policy, or your “true values,” or whatever. The curve, if you draw it, seems over-fit; it’s not the sort of line we’d want to draw in stats class. And indeed: yes. But, um… who cares? […] Why would we import map-making norms into a domain with no territory?”

“[Moral Realists] admit the true morality could’ve been way better. They walked out of Plato’s cave and they were like: yikes. If they had their way, they’d go for Utopia, trust them. It’s just that duty (optimality, beneficence) calls.”

Monthly Le Guin

She Unnames Them (pdf) was published in the New Yorker in 1985.H/T ML. The story jumps off from Genesis 2:19:

19 Now the Lord God had formed out of the ground all the wild animals and all the birds in the sky. He brought them to the man to see what he would name them; and whatever the man called each living creature, that was its name. 20 So the man gave names to all the livestock, the birds in the sky and all the wild animals.

The eponymous She is Eve; the eponymous Them being Unnamed are “all the livestock, the birds in the sky and all the wild animals.” Gender and language are central; the philosophy is classical Le Guin throughout — “the issue was precisely one of individual choice, and that anybody who wanted to be called Rover, or Froufrou, or Polly, or even Birdie in the personal sense, was perfectly free to do so.” Le Guin has few other forays with biblical themes; tho other works do touch on religious concepts, including The Word for World Is Forest, which includes themes of spiritualism and environmentalism, and “The Telling,” dealing with an alien but repressive religious culture.

The story is short, less than a page, so I don’t feel bad about spoiling it. Eve goes to Adam, and returns that which him and his father lent to her. Even 80s Le Guin can’t help sounding sarcastic when she writes:

It is hard to give back a gift without sounding peevish or ungrateful, and I did not want to leave him with that impression of me. He was not paying much attention, as it happened, and said only, “Put it down over there, O.K.?” and went on with what he was doing.

Eating

Lots of smoothies and tofu (not at the same time). I made pancakes three times in February, which felt like too few. I miss Mexican food.

Listening

•This John Clayton video is a treat. Gotta be in the top ten of artists I’d like to see live.
•I listened to three recordings of the Oberammergau Passion Play, inspired by this article. The passion play is performed as repayment for a vow made by the villagers of Oberammergau in 1633 to save them from the plague and the Thirty Years War; in return for their salvation, they perform the five+ hour passion all summer in the decadal year. The overture is especially stunning. For reasons separate from the author of the above article, I found the 2022 performance underwhelming; I most recommend the 2010 recording. I didn’t expect supersessionism to take such a large role.
•RAYE has a superb debut/comeback album out, My 21st Century Blues, positively reviewed in the FT. Unlike her massive, polished 2010s hits with the likes of David Guetta and Jonas Blue, the new album is old-fashioned and R&B and gritty and retro pop and addictive.