Why teeth
August 2026
I did not choose teeth. A friend brought me into Zhang Rui's project group, and I stayed.Zhang Rui is an associate professor in the medical school. For a while I just did whatever needed doing. Then I started sitting in on the group meetings with her graduate students, and at some point I started saying things in them rather than only listening.
What I actually do is hard to put in one line. I built a small web tool for Dr. Kang so she could place landmarks on panoramic radiographs and measure the distances between them. I helped a medical graduate student run a discriminant analysis, because she had the data and the question and not much programming. Most of it I learn while doing it, roughly one step ahead of needing it.
The common thread isn't dentistry. It's that I like building things with people who are not like me — different fields, different ages, people who know something I don't and need something I can make. Teeth are just where that happened to land.
The inflation problem
Making things has become very cheap. A working prototype is now an afternoon. A plausible-looking analysis is now a prompt. What has not become cheaper is knowing whether any of it is true.
It behaves like inflation. The supply went up, so each unit is worth less, and the scarce thing is no longer production — it's whatever certifies that a result means what it claims to mean.
The bottleneck moved. It used to be making. Now it is believing.
So reproducibility isn't a bureaucratic virtue to me, or a box the journal makes you tick. It's the part that still costs something, which is exactly why it's the part worth doing carefully. If someone who isn't me can run it and get what I got, the work meant something. If not, I produced output.
Outside this: I play Paradox games, which is a slightly embarrassing route to a real question. Every federal arrangement designed to hold a multiethnic state together seems to have failed.Austria-Hungary, the first Yugoslavia, a dozen constitutional drafts nobody remembers. The European Union did not.A looser version of the same idea, with fewer instruments of coercion. I don't have an answer. I'm not certain the two are even the same kind of object.
The world changes constantly and is also very stable. Over the last two years language models have gotten into nearly everything I do, and if you asked me what is actually different I would struggle to name it. The tools changed. The problem didn't.
Notes are unedited and occasionally wrong. Corrections welcome at mail@edmundji.com, and recorded on the errata page.