Eric Chen

Email·GitHub·Google Scholar

Asking the Right Questions

Three panels on choosing which problems matter.

A three-panel comic. First panel: two people stand behind a desk with a stack of papers and a laptop whose screen reads LLM, PROOF COMPLETE, with a check mark. One raises a finger and says, LLMs can prove theorems. Choosing which ones matter takes taste. Second panel: a blindfolded person throws a dart at a spinning wheel labelled OPEN PROBLEMS, saying, Ooh, which one today? The wheel's six slices read High-Dimensional Phase Transitions, Minimax Rates for Optimal Transport, Sparse PCA Thresholds, Adaptive Confidence Bands, Conformal Prediction and Causal Identifiability, and the dart lands in High-Dimensional Phase Transitions. Third panel: a dart sticks out of a manuscript titled High-Dimensional Phase Transitions, OPEN. One person asks, Why this problem? The other, holding the blindfold and a fistful of darts, answers, Our taste is uniformly distributed.
Taste is a prior. Theirs is flat.
DesignAdapted from Martin Saveski
Views since Jan 2026…