Decision theory
August 23, 2014 — October 4, 2024
The small but crucial insight is that inference is relative to what you do with it; a 1% chance of rain is a big deal if you’re a farmer, but not if you’re a city slicker. A 1% chance of me catching a rhinovirus is no biggie, but a 1% chance of a global nuclear war is a biggie. We don’t want to guess what is true; we want to guess what is useful, and that depends on the consequences of being wrong.
This sounds obvious, but failures to take this into account are ubiquitous even amongst scientists. I’ve seen things.
1 In classification
This simple but crucial case is the most important one for most people.
(Ferrer 2023; Provost and Fawcett 2001; Dyrland, Lundervold, and Mana 2023b; Suzuki 2022).
TBC
2 Incoming
Cassie Kozyrkov frames testing in a decision-theoretic context which I am increasingly convinced is the only sane one.