Mantic Monday 11/1/21
Read the original on Astral Codex Ten →
Summary
One of the meatier Mantic Mondays, with several genuine mechanism-design ideas. The Keynesian-beauty-contest solution to long-term forecasting (predict what other isolated teams will guess) — and its flaw that it penalizes anyone smarter/better-informed than average and requires isolation. A clever teacher-merit-pay-via-prediction-markets scheme: have markets predict a specific student's test scores conditional on teacher A vs teacher B, so market participants are incentivized to notice and price out the biases (race/class confounders) that sink naive value-added models. And Vitalik's prediction-market-as-moderation proposal (upvote/downvote as bets on 'will the mod ban this', so one person could moderate a Reddit-sized site). Plus the Virginia election, SpaceX/college-bubble/Codex Metaculus questions, and the Eisenberg Souljaboy-tweet-market story.
Why this score
Quality 66 · Strong. Strong-floor (66, top Monday band). Multiple substantive conceptual ideas (the beauty-contest analysis, the merit-pay debiasing scheme, the Vitalik moderation mechanism) lift it well above the roundups.
Claude’s paradigm shift 44 · Moderate. Moderate (44). Relays and sharpens others' mechanism proposals; not an original thesis.
Real-world impact 1 · Negligible. 1 — a forecasting roundup; negligible real-world effect.