Dispute Database
Overview
A disputed-market record is a row from Polymarket's dispute history. A Brierly-classified hard-failure outcome is the narrower editorial class covering 50/50 resolutions and escalated overturns—not a premature bot proposal, spam challenge, or rejected dispute where the rules held.
That distinction matters because raw dispute counts mostly measure process activity. A 50/50 result may also reflect a correctly executed void clause or event-execution problem, so the classification is evidence for review—not proof that every record reflects bad drafting.
Key findings
Failure taxonomy
The case-study taxonomy describes recurring ways settlement language or adjudication can fail. Select a type to filter the detailed examples below.
● Source dependence
Resolution hinges on specific named sources reporting (or failing to report) an observable fact. The world and the source diverge — the source wins.
● Scalar carve-out
Numeric thresholds, rounding, first-print-vs-revision, and boundary handling. The number moves after settlement, or the boundary itself is ambiguous.
● Definitional vagueness
A key term ('suit', 'ban', 'invasion', 'go live') is undefined or defined differently than traders assumed.
● Disclosure timing
The event happened, but confirmation landed outside the market window — or the window itself was read in conflicting ways.
● Oracle manipulation
Token-weighted oracle voting is captured or steered by economically interested whales.
Searchable disputed-market census
filter by category, dispute outcome, hard-failure label, rating, and market textDetailed sourced case studies
high-signal examples from both venues—not the full 1,595-record censusMethodology and limitations
The census is a text-free generated export derived from public Polymarket/UMA history and a private source corpus. It includes one apparent duplicate and lacks identifiers sufficient to prove every row is a unique market. Brierly's hard-failure label is editorial. Dispute escalation is a lower bound on contested settlements; informal disagreements may never reach the oracle. Reported trading volume is not open interest, loss, or capital at risk. The detailed cases include Kalshi examples but do not form a Kalshi census.
Read RuleScore methodologyCite the database
Brierly-authored case-study data is published under CC BY 4.0. Preserve source attribution and the distinction between the generated census and detailed examples.
Detailed case-study notional totals exclude overlapping aggregate rows and cases with unreported volume. Sources remain attached to each case page.