The Partial Win: Reading the Kalshi Ruling the Way I Read Solidity

The headline said win. I read the court order twice and kept circling one word: partial. I have spent enough of my life inside multi-signature implementations to distrust any summary that smooths over the qualifiers. In 2017, at twenty-five, I stayed up three nights reading Gnosis Safe's Solidity line by line and found twelve logic flaws in their multisig logic — flaws the marketing pages never mentioned, because marketing pages never mention the gap between the promise and the contract. That habit never left me. So when the news arrived that Kalshi and Coinbase had secured a partial ruling in Illinois over sports contracts, my first instinct was not to celebrate a regulatory thaw. It was to ask the question the headline buries: partial against what, exactly, and at whose expense? Here is what the charts will not tell you. A win in a courtroom is not a win in a protocol. One is a statement about jurisdiction. The other is a statement about who ultimately holds the keys. Follow the fear, not the chart — and the fear here is not that prediction markets get banned. The fear is that they get legalized into something that quietly abandons the reason they were interesting in the first place. Kalshi is not a crypto company in the way most of my readers understand the term. It is a designated contract market — a DCM — which means the Commodity Futures Trading Commission has granted it the federal authority to list and settle derivatives. Its product is the event contract: a binary instrument whose payout depends on whether a specific thing happens. Did the temperature in Central Park exceed ninety degrees? Did a particular candidate win a primary? Did a specific team cover the spread? The Illinois case is the latest turn in a longer argument about who gets to regulate these instruments. Kalshi's position, roughly, is that because it holds a DCM license under the Commodity Exchange Act, its sports-related contracts fall under federal jurisdiction, and states cannot require it to obtain separate sports-betting licenses. Illinois disagreed. It argued that sports contracts are, functionally, sports betting, and that sports betting is a matter of state licensing and state taxation. The court's partial ruling leaned toward Kalshi — the judge found that the state's licensing rules may conflict with federal derivatives law — but the transaction-fee dispute remained unresolved, and both parties were ordered to submit proposed injunction terms. That word again. Partial. The fee question left open, the injunction unwritten, the appeal undecided. Let me give you the context the news brief leaves out, because it matters for everything that follows. The CFTC has spent years trying to draw a line between event contracts it considers legitimate risk-hedging instruments and those it considers unlawful gaming. In 2024, Kalshi won a federal case against the CFTC over election contracts — a ruling that stunned the agency and opened the door to a class of products it had spent years trying to keep off regulated exchanges. That precedent matters here, because it established a pattern: Kalshi litigates, Kalshi wins on the federal-preemption argument, and the boundary of what can be listed as a derivative moves a little further. The CFTC's relationship with event contracts has been ambivalent for a decade. It has approved some — the agricultural and financial contracts that resemble hedging — and fought others — the political and entertainment contracts that resemble gambling. The 2024 election-contract ruling forced it to accept that its authority to reject a listed contract is narrower than it assumed. That ruling is the reason the Illinois case exists. Once election contracts were permitted, sports contracts became the next boundary, and the states became the next plaintiffs. What looks like a single lawsuit is actually a sequence, and the sequence is not finished. Coinbase's involvement is newer and, in some ways, more revealing. The exchange is not the license holder. It is the distribution layer — the front end, the user pool, the funnel. When a company with Coinbase's retail reach decides that prediction markets are worth carrying, that is a signal about where the industry thinks the volume will be. It is also a signal about who bears the regulatory risk when the legal picture is unsettled. If Kalshi loses at the next stage, Coinbase loses a product line and, potentially, a story it told its shareholders. The history is worth holding in mind, because prediction markets are older than the blockchain and have died this way before. The Iowa Electronic Markets have run since 1988, small and academic, deliberately capped so they could never matter enough to be shut down. Intrade, by contrast, grew large and offshore and was forced to stop serving United States customers in 2012 after the CFTC sued it. The lesson of Intrade is not that prediction markets are illegal. The lesson is that prediction markets become dangerous to incumbents precisely when they become useful to the public. Kalshi is the attempt to build a prediction market that becomes useful without becoming dangerous — and the Illinois ruling is a step in that project. So the ruling is real, and it is meaningful. But it is a ruling about jurisdiction, not about infrastructure. And infrastructure is where the interesting failures live. The settlement oracle is the unexamined center of the entire product. Every event contract terminates in a settlement decision. Someone or something declares the outcome. For a temperature contract, it might be a National Weather Service feed. For a sports contract, it is the official score, the league's ruling, the box score as finalized. That data source is the oracle, and the oracle is the single most consequential piece of infrastructure in the whole product — more consequential than the matching engine, more consequential than the fee schedule, more consequential than the license itself. I want you to sit with that for a moment, because the entire news cycle has been about the license. The license determines whether the market can exist. The oracle determines whether the market tells the truth. And the oracle, in almost every regulated prediction market I have examined, resolves to a small set of trusted feeds, curated by the operator, with a dispute process that terminates in a committee. This is the same pattern I found in 2017 when I audited Gnosis Safe. The smart contract looked decentralized. The upgrade authority sat with a handful of addresses. Code is law, until the law needs to be changed, and then the law is whoever holds the multi-sig. Prediction markets are now repeating this architecture at the settlement layer. The contract is deterministic. The input is not. And the party who controls the input controls the outcome of every disputed event. Consider what happens when a game ends in a way that admits two readings. A buzzer-beater that is reviewed and upheld, then reviewed again and overturned. A stat correction applied three days after settlement. A league that vacates a result for eligibility reasons. In each case, someone has to decide which truth the contract honors, and that someone is not the market. It is the operator, or a committee the operator appoints, or a feed the operator selects. The market prices the probability of the event. It does not price the probability of the oracle. If you can find a prediction market whose settlement disputes are resolved by a mechanism you would trust with your own capital, you have found something rare. Most of them ask you to trust the same committee structure they criticize banks for. If you want to see how fragile this is, build the manipulation case. Take a market with modest liquidity — a mid-major conference game, a player-prop contract, an outcome that resolves on a single official statistic. The cost of moving that market is small. The cost of moving the underlying statistic is not zero, but it is not enormous either, if the statistic depends on a scorer's discretion or a league office's correction. The contract does not need to be hacked. It needs to be argued about, at the right moment, by the right party, and the settlement committee resolves the ambiguity in whichever direction the operator finds defensible. There is no exploit in the code. There is an exploit in the process, and the process is not on-chain. This is why the contract specification matters more than the contract. A DCM must define, in advance, the source agency for every market it lists — the specific feed, the specific timing, the specific rule for corrections. That specification is the real product. It is a legal document dressed as a technical one, and its ambiguities are where the money hides. I have read enough specifications to know that the ambiguity is rarely an accident. It is the room the operator leaves itself to resolve disputes in its own favor, and no amount of federal preemption changes that. The transaction-fee dispute that the Illinois court left unresolved is the second thread, and I think it is the more revealing one. On its face, it looks procedural — who gets to charge what, and under whose authority. But follow it down and it becomes a question about the economic structure of the entire category. A DCM earns revenue from trading fees. A state earns revenue from sports-betting taxes and licensing. Both are claiming the same economic surface. When the court says the fee dispute is unresolved, it is saying that the two parties have not agreed on who owns the margin on the same transaction. Kalshi wants to run the contract on federal terms. Illinois wants a cut on state terms. The user, standing at the end of the pipe, pays whichever structure prevails, plus the spread, plus the settlement risk. Here is the part that the coverage has not connected. The fee dispute is not just about money. It is about the definition of the product. If the contract is a derivative, it is priced and taxed like a derivative. If it is a bet, it is priced and taxed like a bet. The same binary outcome, the same dollar at risk, two entirely different regulatory and economic regimes — and the difference is decided not by the mechanics of the instrument but by the political question of who gets to call it what. I have watched this pattern before. In 2020, during DeFi Summer, I lost my modest savings and watched friends in my Beijing study group lose theirs when Compound's governance token collapsed. What struck me then was not the loss. It was how the loss was distributed. The people who understood the mechanism exited early. The people who trusted the interface stayed. The interest-rate models that Compound and Aave ran were presented as market-driven, but when I modeled them against real supply and demand, they were closer to arbitrary curves — smooth, legible, and disconnected from the actual pressure of the market. The number on the screen looked like a price. It was a policy. Prediction markets are walking into the same trap, one layer up. The fee schedule looks like a market outcome. It is a jurisdictional outcome. The settlement looks like a fact. It is a committee decision. The probability looks like a consensus. It is a number produced by participants whose ability to exit at settlement depends on infrastructure they do not control. This is why I keep insisting that the interesting question is never whether a product is legal. It is whether the product is legible to the people using it. Let me be concrete about the economics of a binary event contract, because abstraction hides the structure. A binary pays one dollar if the event resolves true and zero if it resolves false. If it trades at sixty cents, the market is implying a sixty-percent probability. The buyer of the true side risks sixty cents to win forty. The buyer of the false side risks forty to win sixty. The operator takes a fee on entry, often a fee on exit, and sometimes a fee on settlement. The spread between the two sides is the operator's compensation for providing liquidity and for bearing the risk that the oracle gets it wrong. Now stack the layers. Federal preemption determines whether the market can operate. The state determines whether it can collect. The oracle determines what happens at the end. The fee determines who profits along the way. At no point in that stack is the user's real exposure to being wrong about the world the dominant risk. The dominant risk is being wrong about the plumbing. I have said for years that the post-Dencun data landscape will not stay cheap. When blob space saturates — and it will, faster than the optimistic models assume — the cost of writing settlement data on-chain will rise, and every rollup that built its economics on cheap blobs will have to reprice. The same logic applies here in a different register. Prediction markets are currently cheap to operate because the legal regime is permissive and the data feeds are subsidized by the venues that need them. Neither condition is permanent. If the fee structure gets reclassified, if the data providers start charging for real-time settlement access, the cost of running a compliant prediction market goes up, and the operators pass it down. The user never sees the oracle bill. The user sees the spread. There is a category error at the heart of how the industry talks about this. A DCM is not a DAO. It has a board, a compliance department, a legal team, and a designated regulator. Its governance is corporate, not on-chain. The people who make the consequential decisions about settlement and fees are not token holders; they are officers and counsel. This matters because the crypto audience keeps reading prediction-market news through a decentralization lens, and the subject does not fit the lens. Kalshi is not trying to be decentralized. It is trying to be compliant, and compliance is a centralized posture by construction. The interesting question is not whether Kalshi is decentralized. It is whether the industry's decentralization rhetoric survives contact with a product that is, by design, centralized and legal. Coinbase's role deserves a harder look, because it changes the risk profile of the whole arrangement. A public company carries disclosure obligations that a private one does not. If Coinbase is distributing prediction-market contracts, and if those contracts carry unresolved legal risk, that risk is supposed to surface in its filings. Watch the next 10-Q. Watch whether the language around prediction markets is expansive or cautious, whether the legal contingencies are described or buried. The words a public company chooses about an uncertain product line are a leading indicator of how the company's own lawyers are pricing the risk. If the disclosure is thin, either the exposure is small or the company is telling itself a story. I have seen both, and the second is more common. There is a competitive dimension the coverage has largely ignored. If prediction markets are legalized on federal terms, they do not merely coexist with traditional sportsbooks. They compete with them, on price and on trust. A sportsbook charges a vig and settles on its own terms. A prediction market charges a spread and settles on a public rule. For a certain kind of user, the second is more attractive, and the incumbents know it. That is why the state-level resistance to Kalshi is not purely a matter of principle. It is a matter of revenue, and the incumbents who depend on that revenue have every incentive to fund the resistance. Follow the money, and the money is in the statehouses. The injunction terms, when they come, will tell us more than the ruling did. An injunction is a concrete statement about what a party may and may not do while the case proceeds. If the court grants a narrow injunction, Kalshi keeps operating in Illinois and the legal question is deferred. If the court grants a broad one, the product is frozen in that state and the economics of the case shift entirely. The scope of the injunction is the difference between a business that continues and a business that pauses, and the parties will fight over every word of it. The headline will call the outcome a win or a loss. The injunction will decide whether the win is worth anything. There is a cleaner way to say all of this. The Illinois ruling is a fight over who gets to define the product. The oracle and the fee are the two places where that definition actually lands. Everything else — the headlines, the press releases, the celebratory threads — is commentary on the fight, not the fight. Let me now do what I always do when I read a ruling like this: I ask what the losing side would have to prove, and I ask whether the winning side has the infrastructure to survive being right. If Illinois appeals and wins, Kalshi's sports contracts in that state face a licensing requirement that may be impossible to satisfy alongside its federal obligations. The product does not get banned; it gets stranded. If Illinois loses and the preemption holds, Kalshi's contracts become a template, and other states have to decide whether to fight or to fold. In that world, the interesting question becomes whether Kalshi can scale settlement across a national footprint without centralizing the oracle into a single point of failure that regulators can lean on. That is the question I would be asking if I sat on Kalshi's board. Not can we win. Can we survive winning. Here is the angle I have not seen anyone take, and it is the one I believe. The prediction-market industry is celebrating a victory for decentralization. It is actually a victory for centralization, and it is being sold to you as the opposite. Think about what federal preemption accomplishes. It replaces a patchwork of state rules with a single federal regime. A single federal regime is easier to lobby. It is easier to capture. It is easier to standardize — and standardization is exactly what the incumbents want, because standardization favors the largest operators with the best legal teams. A world of fifty state regimes is messy, expensive, and hostile to scale. A world of one federal regime is clean, efficient, and friendly to whoever gets there first. Kalshi is first. Coinbase is the distribution. Together they are building the compliant prediction market that regulators will eventually treat as the reference implementation. And a reference implementation, by definition, is a centralization point. Follow the fear, not the chart. The fear is not that prediction markets get shut down. The fear is that they get standardized, and that the standard is written by the two companies with the most to gain from writing it. The crypto-native answer to this is Polymarket — on-chain, offshore, transparent, and priced by a market rather than a committee. I am not naive about Polymarket. It has its own oracle problem, its own dispute mechanism, its own governance captured by a small set of holders. But it is architecturally honest about the fact that settlement is a social process. It puts the dispute in public. Kalshi puts the dispute in a committee room, and the federal preemption it just won is the legal foundation for keeping it there. The pragmatism test is simple. If the goal is a prediction market that anyone can use without permission, federal preemption is not a step toward that goal. It is a step toward a market that anyone can use without permission, provided they are willing to be settled by a committee they did not choose, on a contract defined by a jurisdiction they do not vote in, at a fee set by an operator they cannot replace. That is a real product. It may even be a good business. It is not decentralization. It is the same centralization we left the banks to escape, wearing a binary contract as a costume. And there is a second-order effect that deserves its own paragraph. If Kalshi's model becomes the template, the capital that flows into prediction markets will flow through compliant rails. That capital will demand the same things compliant capital always demands: predictable settlement, disclosed counterparties, audited oracles. Audited oracles are centralized oracles. The market will not get more decentralized as it gets more legitimate. It will get more centralized, and it will call that maturity. If you can read the roadmap of any compliance-first crypto product, you already know where this ends. It ends with a committee. The Illinois ruling matters, and it will be cited for years. But the thing worth watching is not the ruling. It is whether the settlement layer gets audited the way the matching engine gets audited, and whether anyone asks who holds the keys when an event resolves in two directions at once. We are watching prediction markets become infrastructure. The question is whether they will be infrastructure that tells the truth, or infrastructure that tells the version of the truth its operator can defend in court. Follow the fear, not the chart. The chart says the market is growing. The fear asks who settles it when the market is wrong.