Lee (aka pumpedlunch)

How UMA Oracle Works: The Genius of Paying People to Tell the Truth Imagine you and your friends made a bet about whether the principal would wear a blue suit to the next assembly. Problem: how do you prove it actually happened? Who decides if it was really blue? Who settles the argument when someone claims "that was more navy than blue"? This is the oracle problem. In prediction markets, millions of dollars are on the line. You need someone -or something -that can look at reality and say "yes, it happened" or "no, it didn't" And that someone can't be bribed, can't be lazy, and has to be right Welcome to UMA Oracle. It's not one person. It's not a computer algorithm. It's actually something smarter: it's a system that pays people to be honest The Genius Idea: Optimism With Incentives Here's how @UMAprotocol works -and it's beautifully simple: Step 1: Someone proposes an answer Event resolved? Someone puts up money (~$750) and says "Biden dropped out on July 21, 2024. Resolved: YES" Step 2: The challenge period For 2 hours, anyone can dispute it. Think you can prove them wrong? Put up $750 of your own money and challenge them Step 3: If nobody challenges... it's done 95% of the time? Nobody challenges. The answer stands. Game over. The proposer gets their money back plus a small reward (~$50-100). Done in 2 hours Step 4: If someone challenges... actual voting happens Now it escalates to UMA token holders -thousands of people with real money at stake. They vote on what actually happened. Simple majority wins. If you voted wrong? You get slashed (lose a tiny % of your stake). If you voted right? You get rewarded Why This Is Brilliant 1. It's Expensive to Lie Think about the math: to move a market falsely, you'd need to: -- Propose a lie (lose $750) -- Watch someone challenge it (they're $750 in too) -- Lose the vote in UMA (someone gets your money anyway) So now you've lost $750+ trying to manipulate a market where maybe you could've made $1,000 if you'd won Compare that to traditional systems where one person makes the decision. There? You only need to bribe or threaten one person. Here? You'd need to control thousands of token holders. Good luck 2. It Scales Efficiently 11,000+ markets have been resolved. Gas costs? Like $0.01-0.05 per operation. That's basically free. Traditional centralized oracles? They'd charge $100+ per market resolution. UMA does it for pennies And 98% of resolutions happen without any disputes. You know why? Because people know disputing costs them money, so they don't bother unless they're actually right 3. It Handles Real-World Messy Questions Chainlink Oracle? It handles prices. Asset ABC = $123.45. Binary. Clear UMA? It handles reality. "Will Zelensky wear a suit in June 2025?" "Will a peace deal be reached?" "Did the storm cause more than $10B in damage?" These aren't price feeds -they're events, narratives, judgment calls The Real-World Examples That Prove It Works The Biden Withdrawal (July 2024) Biden announced he was dropping out of the race on X. The market resolved in 2 hours. No disputes. Why? Because it was obvious. There's video. Official statement. Everyone could see it. Proposing "No, Biden didn't drop out" would be insane -you'd lose $750 instantly This is the Schelling Point in action. Everyone converges on the obvious truth because lying costs money and truth costs nothing Result: Perfect resolution. Market settled. Life moves on When It Worked Even When People Tried to Game It Another market: "Will Biden drop out?" before he actually did The price hit 80% probability to "Yes" weeks before the official announcement. Why? Because people with information about what was happening in the campaign -real insider talk, credible rumors, smart analysis -were buying "Yes" contracts They weren't lying. They were just earlier than everyone else. And that's fine. That's what markets are supposed to do: reward accurate information. Result: The market worked like a truth-finding engine. Money followed reality How It Actually Protects Itself The Bond System: -- You put up money to propose: $750 gone if you're wrong -- Someone challenges: now they're $750 down if they're wrong -- Suddenly everyone's thinking hard about what's actually true The Voting System: -- If you vote wrong in UMA: 0.1% of your stake gets slashed -- Doesn't sound like much, but imagine having $10,000 staked. That's $10 gone per wrong vote -- Do that 10 times a year carelessly? That's $100 in losses. Now you pay attention The Whitelist (MOOV2): -- Only proven truth-tellers can propose on big markets -- Your first proposal? It's public. Your accuracy record is visible -- Everyone can see: "Oh, this person's been 95%+ accurate over 100+ proposals." Worth trusting them Result: The system creates incentives for honesty The Philosophical Beauty Here's what's actually genius about this: You don't need to trust that people are good. You don't need to trust institutions. You don't need to trust anything except math and money Compare: -- Traditional oracle: "Trust this company. They're reliable" -- UMA oracle: "Lie and lose money. Tell the truth and make money. Choose" The first requires faith. The second doesn't. It requires only rational self-interest, which is something you can count on Why This Matters for Prediction Markets Polymarket can exist and grow because UMA solved the resolver problem. Without it? -- Every market needs a centralized judge -- That judge can be corrupted, sued, pressured -- The whole thing collapses into litigation With UMA? -- Markets resolve through economic incentives -- Thousands of token holders can't be bribed together -- The system is resilient 11,000+ markets resolved. 94%+ accuracy. Billions in value secured The Bottom Line UMA Oracle is like asking "how do we get thousands of people to tell the truth about something?" Answer: Pay them to. Punish them for lying. Make it all transparent. Let the community watch Is it perfect? No. Will there be edge cases? Absolutely. Do whales sometimes try to game it? Yes But 98% of the time, it works flawlessly. And the 2% of the time it doesn't? The system learns and improves That's not just smart engineering. That's actually kind of beautiful

Disputes are low AND volumes are high.

the biggest hypocrites i see in the prediction market sphere everyday are the people that aggrandize the market's power to surface truth in complex situations and yet vilify resolution mechanisms that make that possible prediction markets would be boring if you could only bet on questions that resolve via API the bigger vision of disrupting news, surfacing truth, economically incentivizing the collective to provide information is what's thrilling and exciting about them this requires resolution mechanisms that are scalable, flexible and transparent, all of which centralized resolution is not centralized prediction markets are fine in their own right, but you don't get to preach about the high mission of prediction markets if what you're building prohibits it

Across is now live on Solana. Bridging to and from @solana, powered by intents.

We're coming home. 🇺🇸

One day, there may be more AIs than humans creating, pricing, trading millions of polymarkets on everything (h/t @VitalikButerin). A unique benefit of onchain prediction markets is the ability for AI agents to plug in. Multiple groups quietly print on Polymarket with agentic workflows. As AI ushers in Superintelligence, Polymarket ushers in Collective Intelligence. It's happening.

@X @Polymarket

Not every IP conflict needs a courtroom. With @StoryProtocol, disputes surface, and UMA helps resolve them through open, community-enforced truth.

SYSTEM UPDATE: > Re-try logic for P4_TOO_EARLY condition has been restructured. > All execution attempts are now retained; data loss eliminated. > Previous retry mechanism caused elevated P4 rates — expected to normalize. > Output: reduced P4 errors, improved proposal log integrity, increased accuracy. ⭕️ USER INTERFACE: Re-run instances display in separate tabs. ⭕️ SYSTEM BACKEND: API and database modified — full run history preserved; legacy compatibility maintained.