The optimistic oracle design assumes that honest disputers will act as a check on bad proposals. But the final backstop—the DVM vote—is where the real power lies, and it is a pure plutocracy .
Bloomberg’s analysis of blockchain records and past votes revealed that the “crowd” deciding Polymarket’s most contested outcomes is really just a tiny group. Among more than 6,400 participating addresses, nine whale wallets controlled roughly half the total voting power and sided with the winning outcome in nearly all disputes .
Some wallets hold so much sway that their votes can single-handedly decide cases. In multiple documented incidents, a single voter or coordinated group deploying millions of UMA tokens—sometimes representing 25% or more of total voting power in a round—was able to flip a market’s outcome just before finalization .
The anonymity of these wallets compounds the problem. Unlike Kalshi, which resolves disputes internally with clear lines of accountability, Polymarket outsources final authority to an unregulated, pseudonymous jury . Polymarket’s user agreement explicitly states that the platform “assumes no responsibility for resolving disputes related to trading contracts”
.
What makes this structure genuinely dangerous is the incentive misalignment. Voters frequently hold positions in the very markets they are judging. More than 60% of active UMA voters are linked to Polymarket trading accounts . A voter holding a directional bet or a position on both sides of a market can vote to steer the outcome in their own financial favor—effectively acting as both judge and a financially interested party.
The consequences have been concrete and expensive. Controversial UMA resolutions affected over $30 million in market value in 2025 across markets like the Ukraine minerals deal, the Fort Knox gold audit, and the UFO declassification market . The Iran ceasefire dispute brought even greater scrutiny, not only for its $280M+ volume but for suspicious trading patterns that triggered congressional investigations
.
Awareness of the problem is not new. UMA introduced a “MOOV2” update that restricted voting eligibility to just 37 addresses, an attempt to improve accountability. But the update failed to solve the core concentration problem—it simply formalized a slightly larger oligarchy . Polymarket’s own broader revamp of the voting process has been delayed
.
The deeper issue is structural. The optimistic oracle assumes disputers will act as a protective check, but when whales control the final DVM vote, that check becomes functionally meaningless . As one analysis put it: “decentralization in name does not mean decentralization in practice”
. Until the voting mechanism itself is fundamentally re-architected—or until Polymarket brings oracle functions in-house, which it is currently exploring through a proposed POLY token—the world’s largest prediction market will continue to rely on a truth-arbitration system controlled by a handful of anonymous wallets
.
Comments
0 comments