Rebutting random walk: Survival probability updates rationally over time

erikbryn · x · 2026-08-29

Erik Brynjolfsson argues against the random walk hypothesis with a counterexample: assume an extinction risk window T with probability p each year, and no risk after T. Each year we survive, we gain information and should monotonically update our expected value of survival. Such rational patterns do not appear as random walks.

Related event: Hanson's Random Walk View Claim Meets Counterargument(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →