Living references
One canonical URL per measured fact, updated in place. Each carries its methodology, sample, and measured history — base-rate facts, not predictions.
-
Titles with no checkable threshold dispute far more
On Polymarket, markets whose title states no checkable threshold dispute more than those whose title does, in both samples (high-volume: 26.8% vs 1.4%; chronological: 0.7% vs 0.5%). Read from the title only; the resolution rules were not read. Direction is the fact; magnitude is sample-dependent.
last material change: -
Dispute rank and flip rank diverge by category
On Polymarket, dispute-rate and flip-rate rank categories in opposite orders: Geopolitics disputes most but flips only 11.5%; Sports rarely disputes but flips 32.8% when it does.
last material change: -
Most UMA disputes confirm — they don't reverse
Of 580 resolved Polymarket UMA disputes over ~313 days, only 14.5% reversed the original answer — 85.5% just confirmed it.
last material change: -
A UMA dispute locks capital ~5 days, whatever the category
When disputed at UMA, a Polymarket market runs about two resolution cycles regardless of category — an estimated 104–120 hours locked at UMA's ~60-hour cycle cadence (high-volume sample; magnitude is sample-dependent). What varies is how often a category disputes (Geopolitics 33.9% vs Sports 0.4%), so the per-dispute lockup is roughly fixed while the frequency is what moves the exposure.
last material change: -
Did a Polymarket market's rules change after you bet?
Yes, and it is measurable: over the 90 days ending 2026-09-01, 13,436 rule-change events were read on-chain across 11,930 distinct Polymarket questions and 350 distinct texts, and 8,957 of those questions carry at least one edit outside the boilerplate clarification — a floor, not a census, because 20 of 519 block chunks failed to read.
last material change: