Kalshi settles bets according to a rulebook and through operator decisions, Polymarket through token-holder voting, while World relies on a data stream: no arbitration panel, no dispute window and no authority to which users can appeal. On September 9, World opened its standalone world.xyz platform to a waitlist of more than one million users — and on launch day, the site crashed under the surge in traffic. Three resolution models are now operating in the real world, and each fails in its own way.
Key facts: more than 150,000 markets (sports, crypto, politics, finance, culture); settlement through Chainlink Data Streams and Chainlink Runtime Environment; non-custodial protocol on Solana with settlement in the CASH stablecoin; no brokerage account or exchange registration required.
None of the components is new on its own — what is new is their combination. World launched inside the Phantom wallet during the summer, and on September 9 opened a standalone world.xyz website, expanding access to a waitlist of more than one million people. More than 150,000 markets have been created since the Phantom integration: sports, crypto, politics, finance, economics and culture. The initial lineup is broad: every NFL regular-season game, seven soccer leagues, Formula 1 races, binary contracts on the 2026 U.S. midterm elections and Federal Reserve interest-rate decisions. Markets for stocks, commodities and weather are also planned.
The mechanics are standard for binary event contracts: each market issues "yes" and "no" contracts priced between zero and one dollar, and the verified outcome settles one side at one dollar. Three structural characteristics distinguish World. First, it is non-custodial: the protocol does not hold customer funds, and assets move only when a user enters a market. Second, orders are routed to liquidity providers on Solana rather than through an off-chain order book. Third, settlement takes place in CASH, a dollar-backed stablecoin within Phantom, with winning positions automatically redeemed directly into the wallet. No brokerage account and no exchange registration: users pay only network fees to open and close positions.
Kalshi: resolution according to a rulebook by a regulated operator. Contracts are settled according to pre-published criteria applied by the exchange, which operates as an officially designated contract market. A specific organization makes the decision, a federal regulator oversees it, and participants have a path for complaints. The operator can correct errors and can make discretionary decisions — these are two sides of the same characteristic.
Polymarket: optimistic oracle with token-holder voting. An outcome is proposed, a dispute window opens, the dispute proceeds to a vote by holders of the oracle's governance token, and the result is finalized on the blockchain. No one can determine the outcome unilaterally — and no one can correct it after finalization.
World: automated data stream with no dispute stage. Chainlink Data Streams provide market data, while the Chainlink Runtime Environment executes settlement as soon as a game ends or an event reaches its specified deadline. The public description explicitly states: there is no human panel, no token-holder vote and no dispute window delaying payouts.
| Model | Who Decides | Error Correction | Where to Appeal |
|---|---|---|---|
| Kalshi | Operator according to rulebook | Possible | Exchange and regulator |
| Polymarket | Token-holder vote | No after finalization | Participation in dispute |
| World | Chainlink data stream | No | Nowhere |
The trade-off across all three models is the same, and it is worth stating as a principle: every mechanism that removes discretion also removes the ability to correct mistakes. Kalshi can correct an error — and can make a discretionary decision you may dislike. Polymarket cannot make an arbitrary decision — and cannot correct someone else's after finalization. World removes discretion more than either of them — and therefore removes the ability to correct errors more than either of them.
For its class of tasks, this design is clearly stronger than the alternatives. Objective outcomes settle instantly: a game ends with a score, Bitcoin's price at a specified timestamp is a number, and a Federal Reserve decision is a published figure. For contracts based on unambiguous machine-readable data, a dispute window is simply a delay. Automation also eliminates the failure scenario that has damaged the reputation of the entire category most severely: disputed resolutions, because a market settled against what most observers believed happened is the one thing capable of destroying trust in a forecasting instrument. The design also scales: administering 150,000 markets since the summer according to individual rulebooks would have been operationally impossible, whereas here resolution has no per-market administrative cost. Notably, Polymarket itself has moved toward oracle-based settlement for price markets — confirmation that the approach makes sense for this class of tasks.
The limits are equally concrete. Ambiguous events have no data stream: a contract asking whether an official has resigned, whether a conflict qualifies as a ceasefire, or whether a statement constitutes an official endorsement cannot be settled by a price feed because the definition itself, rather than the measurement, is disputed. A data-stream failure has no correction mechanism: if the source reports an incorrect or delayed value, or a value that does not reflect reality, the automated system will settle according to it — and without a dispute window, there is no stage at which someone can point out that the input data was wrong before the money moves. Edge cases are resolved mechanically: a postponed game, an interrupted race, a delayed announcement. And there is no one to appeal to: settlement on World is the output of a program that has already executed.
World is listing contracts on every NFL regular-season game, the 2026 midterm elections and Federal Reserve decisions — precisely the categories currently under attack in the United States: state gambling regulators argue that sports contracts constitute betting that requires state licenses, California tribes are litigating under federal Indian gaming law, and a bipartisan bill seeks to ban sports contracts entirely on regulated exchanges. Kalshi and Polymarket hold federal licenses — and are still involved in litigation. World requires no brokerage account or registration, does not hold customer funds and routes orders to liquidity providers on a public blockchain. The optimistic interpretation is that a non-custodial protocol genuinely falls outside frameworks built around intermediaries. The skeptical interpretation is that the same argument was made by offshore platforms before 2022 and ended with settlements and geoblocking. Geographic restrictions are not mentioned at all in the launch materials — the most significant unverified fact surrounding this launch.
Solana is not an accident here: the network spent the year as the dominant venue for memecoins and is promoting prediction markets as the next consumer category. Phantom is the distribution mechanism: a million-person waitlist provides a wallet-native channel that no licensed competitor has, and it cost an integration rather than a marketing budget. Chainlink is the infrastructure bet: the same combination of Data Streams and Runtime Environment was adopted by another prediction market earlier in the year, suggesting the emergence of a standard. An oracle selling price feeds is a utility; an oracle providing settlement for an entire category of markets becomes an arbiter of outcomes, and switching costs increase with every market. If the category standardizes on a single settlement provider, the failure scenario for the entire industry becomes correlated — which looks like efficiency right up until it begins to look like systemic risk.
For incumbent players, the threat is not the technology but distribution: no app, account, deposit or identity verification; and the same characteristics that make World's distribution inexpensive also make its regulatory position unclear. Things worth watching include the composition of the 150,000 markets (how many are settled using unambiguous data and how many depend on disputed definitions), the first genuinely controversial settlement, the emergence of geoblocking, adoption of the model by licensed platforms, and trading volume relative to the number of markets created. The trade-off is the same as across decentralized finance: you give up the ability to appeal and receive access, speed and permissionlessness in return.
The key question to verify: what proportion of the 150,000 markets consists of contracts with disputed definitions. If it is small, the design fits the task and the reputational risk remains with competitors. If not, the first disputed settlement will arrive without any mechanism for handling it, which is worse than either alternative.
“Prediction markets are the most honest way to learn about the future: prices don't lie, because lying there costs money.”
— Robin Hanson, economist and prediction market researcher
