Our last post explained how blockchain oracles work using @chainlink as an example. Specifically, how DeFi apps fetch price data through Chainlink’s data feeds. Today, we’ll go beyond simple price data and explore how oracles can resolve subjective, human-verifiable truths 🧵👇 Most oracles, like Chainlink, continuously push verified data onchain—e.g., ETH/USD prices—so DeFi apps can fetch it at will whenever they need it. That’s great for objective data that can be pulled from APIs, but what about subjective data that requires human judgment? This is precisely the type of data that enables innovative decentralized protocols, including onchain prediction markets and RWA and insurance protocols. Onchain prediction markets like @Polymarket let users bet on real-world outcomes, such as which candidate won a mayoral election or which team won a soccer match. DeFi protocols like @MakerDAO, @centrifuge, @goldfinch_fi , and @maplefinance tokenize off-chain assets such as T-bills, invoices, and loans. This means they need to verify human-reported facts, such as whether an offchain loan payment was made or how a diversified T-bill portfolio is performing. These facts can’t be fetched via an API in real time; instead, they rely on attestations, audits, and other forms of human verification. In these cases, @UMAprotocol's optimistic oracle comes into play. Instead of constantly pushing verified data onchain like Chainlink does, UMA flips the model: It assumes most people act honestly and that disputes are rare, hence the name “optimistic.” Here’s how it works 👇 When a piece of data needs to be verified—say, “Who won the NYC mayoral election?” or “Did the borrower repay their loan this week?”—anyone can propose an answer to UMA’s oracle. This proposer must stake collateral (usually $UMA tokens) as a bond with their submission. This stake acts as skin in the game: if they lie, they lose it. Once the proposal is submitted, a short challenge window—typically 24 to 48 hours—opens. If no one disputes the answer, it’s automatically accepted as true, pushed onchain, and the issue is settled. However, if someone believes that the initially proposed answer is wrong, they can dispute it by posting their own bond as collateral. If this happens, UMA escalates the dispute to its Data Verification Mechanism (DVM)—a decentralized onchain voting system. The DVM is composed of $UMA stakers who independently review off-chain evidence, including news, reports, or transaction data, and vote on what’s true. These votes aren’t coin flips—they require informed judgment. For instance, to resolve the “Who won the NYC mayoral election?” market on Polymarket, DVM voters check the official election results from the NYC Board of Elections and vote accordingly. That’s where Schelling-point game theory kicks in. Voters are rewarded for aligning with the answer they believe most other voters will consider correct. Since everyone expects the majority to rely on official sources, the safest (and most profitable) move is to vote for the objectively true outcome. This makes the system self-stabilizing: honest proposers earn rewards, dishonest ones lose their bond, and disputers are rewarded for catching lies but penalized for false challenges. In short, UMA replaces continuous data feeds with incentive-driven truth discovery. For prediction markets, it means events can resolve transparently and cheaply. For RWA and insurance protocols, it means real-world data can be verified without relying on a single trusted intermediary.