Aug 23, 2026

Medical Tests and AI Text Detectors

I recently listened to a podcast episode where the host and a guest tried to explain the problem with medical tests that produce a lot of false positives, such as various screenings that are used for certain types of cancer. Despite both being smart and capable people, they did an impressively bad job at this, and I suspect they left much of the audience either confused or misinformed. Part of the problem was that they jumped immediately to using terms like sensitivity and specificity, as well as Type-I / Type-II errors, without properly defining what these mean. They also chose to frame this using a Bayesian explanation, using terms like prior and posterior, even though it’s arguably simpler to explain things in frequentist terms. But part of the problem is that this is a legitimately complicated issue to explain to a general audience, especially in audio. ↳

:  large-langauge-models ,  artificial-intelligence ,  sociotechnical-systems ,  statistics ,  measurement ,  evaluation ,  fiction ,  medicine ,  communication

Jul 13, 2026

Reflections on Claude Code

I’m generally a pretty slow adopter of new tech, but in the case of Claude Code, the groundswell of support and enthusiasm was simply too strong to ignore. Having now spent some months exploring how I might work with it, it feels to me like easily the biggest step change, in terms of utility, of any development in the LLM era. For many people, ChatGPT was probably as significant, but that was arguably more about public awareness, rather than dramatic progress. Claude Code, by contrast, truly feels transformative, and it’s honestly shocking that it works as well as it does! ↳

:  claude-code ,  software ,  sociotechnical-systems ,  instability ,  large-language-models ,  generative-ai ,  chatgpt ,  infrastructure ,  anthropic ,  technology ,  optical-character-recognition ,  addiction-by-design

May 9, 2026

64th Ann Arbor Film Festival

I’d been aware of the Ann Arbor film festival since I moved here a few years ago, but I hadn’t known until recently that it’s both a qualifying festival for the Oscars, and North America’s oldest experimental film festival! Running since 1963, it’s a low-key festival that features a wide range of genres and styles over a few days of programming. Being a long-time fan of Wavelengths (the experimental film program that takes placea as part of the Toronto International Film Festival), I was intrigued. Most years it has felt like the festival happened at an especially busy time, but somehow this year things felt much more under control, so I somewhat spontaneously decided to buy the all access pass, and try to see as much as I could. ↳

:  culture ,  art ,  film ,  photography ,  generative-ai ,  documentary ,  animation ,  found-footage ,  archives ,  surrealism ,  criticism ,  ann-arbor

Apr 5, 2026

Moltbook as Totemic Schelling Point

Every so often a story comes along that seems to be part of every conversation. The release of ChatGPT was a prime example of this, given how much it dominated discussions for months after release. More recently, it felt like Moltbook managed to achieve a similar kind of virality, albeit at a much smaller scale. I almost can’t believe I’m writing about this now too, given that I am one of those who felt (and still feels) like the whole Moltbook saga was extremely overhyped, and very possibly misunderstood by many of the people who were most excited about it. And yet, I nevertheless think it’s interesting to reflect on what made it briefly into a kind of engine of public discourse. ↳

:  large-language-models ,  agents ,  sociotechnical-systems ,  culture ,  media-storms ,  computational-social-science ,  coordination ,  communication ,  infrastructure

Jan 31, 2026

Blue Prince

I recently hit a milestone in Blue Prince, which I suspect will end up being a natural stopping point for me. Even though I’m sure I’ll never do everything there is to do in this game, it’s definitely been one of my favourites, and it ended up capturing my interest far more than I initially expected. For anyone who’s not aware, Blue Prince is a puzzle game from last year that has been a major critical (and moderate commercial) success. At a high level, it combines first-person exploration of a mansion involving random tile placement that resets in a rogue-like fashion, with many mysteries and puzzles to be discovered, against the background of a larger narrative. Why am I mentioning this here? I think there are a few things about this game specifically, that make for some interesting points of discussion.

:  games ,  books ,  criticism ,  tonda-ros ,  douglas-hofstadter ,  artificial-intelligence ,  knowledge-infrastructure

Jan 4, 2026

Infrastructural Failures

Susan Leigh Star has noted that infrastructure is often invisible until it breaks. We all know that we are ensconced in a whole network of pipes and wires that provide water, electricity, and other systems upon which we depend, but at least for those of us lucky enough to have reliable service, it is easy to take it for granted that these things just work, until they don’t. Given the increasing interdependence of software and infrastructure, the recent Waymo failure in Los Angeles provided a useful reminder of the types of risks inherent in these kinds of distributed-but-centralized technical systems.

:  artificial-intelligence ,  infrastructure ,  technology ,  sociotechnical-systems ,  software ,  instability ,  governance ,  crowdstrike ,  waymo

Dec 22, 2025

A review I recently wrote that I expect to write again

We are at a strange moment in scholarship, where a powerful new tool (language models) has been handed to people, and existing academic systems are scrambling to figure out how to respond. Even leaving aside the crisis around a potential deluge of AI-generated papers, there are many more conventional questions about what an appropriate use of language models is for domains other than computer science. Especially in the social sciences, my understanding is that established journals are struggling both to recruit qualified reviewers, and to know what appropriate standards should be, in terms of methodological rigor.

:  computational-social-science ,  large-language-models ,  text-classification ,  measurement ,  reproducibility ,  bias-correction ,  science ,  criticism ,  knowledge-infrastructure ,  invisible-colleges

Nov 12, 2025

Language Model Hacking

With the widespread success of language models on many tasks in a zero-shot setting, there has been a huge surge of interest among social scientists in wanting to use them to code or classify documents, sometimes in place of human annotators. Given both the freedom to specify prompts, and the lack of connection to domain-specific training data, a concern naturally arises as to how easily people can manipulate their designs to produce a desired conclusion. This had been on my mind recently, and so I was delighted to see that a couple of recent papers specifically take on this question, both of which conclude that there is indeed considerable latitude to produce a desired finding by manipulating the choices involved. These are extremely useful and important results, although they also open up some questions for me, which I wanted to think through here.

:  measurement ,  evaluation ,  reproducibility ,  manipulation ,  bias-correction ,  large-language-models ,  computational-social-science ,  criticism

Sep 1, 2025

Why Write?

If there is one question that I have returned to over and over in various hand-written journals, it is the basic question of why write anything at all. This could be taken to be asking myself why I am writing, or why others choose to write, or what the purpose or point of writing is more broadly. It could be a question about motivation, or impacts, or addiction. It could be a provocation, to myself or others. But the mystery remains. Why engage in this dangerous business?

:  writing ,  archives ,  large-language-models ,  chatgpt ,  data ,  culture ,  accountability ,  tyler-cowen ,  michel-foucault

Mar 28, 2025

Peeping Tom, Triptych

I don’t go to see much in the way of contemporary dance, but on a whim, I decided to get tickets to the Power Center to see Triptych, a performance developed by the Peeping Tom dance company from Belgium. Comprised of three pieces—The Missing Door, The Lost Room and The Hidden Floor—created in 2013, 2015, and 2017, respectively, the performance ran just over two hours, including a partial intermission. This post is a bit of a break from what I’ve been writing lately, but it feels worth highlighting, if only to help keep at least some attention on artistic endeavours. Just as a warning, I have not refrained from spoiling anything in here, so you might want to stop reading if you’re planning to see it and want the full effect.

:  theater ,  dance ,  surrealism ,  horror ,  performance