Catalog of 14 resolution failures over 18 months at Polymarket and Kalshi affecting over $500M in volume, including the $242M Zelenskyy-suit market, the $120M TikTok ban, and the $47M Cardi B halftime market that resolved yes on Polymarket and no on Kalshi. Groups the failures into four patterns: vague criteria, decentralized oracle capture (UMA token cap below disputed volume), centralized operator discretion, and cross-platform divergence. Useful reference for anyone designing resolution mechanisms or underwriting oracle risk.

Walks through UMA's Optimistic Oracle pipeline as used by Polymarket: assertion, liveness period, dispute, and DVM token-holder voting. Uses the $240M Zelenskyy suit market as a case study where semantic ambiguity ('Is a militarized black outfit a suit?') triggered a disputed resolution ultimately decided by whale token holders, illustrating how the system's 'decentralized courtroom' handles edge cases.

Post-mortem of Polymarket's US government shutdown market, where the market resolved 'Yes' to a shutdown that never happened. Traces the failure to structural issues in oracle design: token holders who can trade and vote, retroactive rule changes, and a corruption cost lower than the value at stake.

Case study of Polymarket's Venezuelan election market, where UMA token-holders overrode the platform's own resolution rules to declare opposition candidate Gonzalez the winner despite official results. Highlights conflicts of interest when oracle voters can also bet on outcomes.