Epistemic status: I don’t think anything I’m saying in this post is completely new (eg see https://www.lesswrong.com/posts/Ap4KfkHyxjYPDiqh2/pascal-s-muggle-infinitesimal-priors-and-strong-evidence and https://www.gwern.net/mugging) apart from maybe the meta-probability framing, but it’s an exercise in
I edited Resistance; Tales from a Post-Antibiotic World, an anthology of short fiction themed around antibiotic resistance. While microbiologists and other scientists have been warning about the dangers of drug-resistant bacteria for years, as far as I know there haven’t been significant reductions in the amount of antibiotics given to livestock, which is a major cause of resistance, and relatively easily reduced. You can buy it on Amazon.
Gaurang Sriramanan and I had some fun messing around with Wasserstein training to build robust image classifiers. (pdf link)