Prediction Markets Promised Better Information. Instead They’re Creating Powerful Incentives to Corrupt Information.

Prediction Markets Promised Better Information. Instead They’re Creating Powerful Incentives to Corrupt Information. from the did-anyone-bet-i-would-write-this? dept There’s a concept in economics known as Goodhart’s Law, often summarized as: “When a measure becomes a target, it ceases to be a good measure.” The idea, originally about monetary policy, has proven remarkably durable across domains. When you attach high enough stakes to a single metric, people stop trying to accurately reflect reality and start trying to game the metric. Schools teach to the test. Banks shuffle risk off their balance sheets to hit capital ratios. Hospitals reclassify patients to improve their reported outcomes. Prediction markets were supposed to be immune to this. The whole pitch — the reason people like me found them conceptually interesting for years — was that because participants are putting real money on the line, they’d have powerful incentives to seek out and act on true information. Financial stakes, the theory went, would filter out noise and bullshit and produce a hopefully decently accurate signal about the probability of real-world events. The wisdom of crowds, sharpened by the discipline of the wallet. What most people didn’t think through was the obvious corollary: what happens when you attach $14 million in stakes to a 150-word blog post by a war correspondent, and the gamblers decide it’s cheaper to threaten the journalist than to accept they made a bad bet? Emanuel Fabian, a military correspondent for The Times of Israel, published an extraordinary account of what happened after he filed a routine item on March 10 about an Iranian ballistic missile striking an open area outside Jerusalem. No one was injured. It was, in the context of an ongoing war, a minor event. He reported it accurately and moved on. Then the Polymarket bettors found him. It turns out that more than $14 million had been wagered on a Polymarket bet titled “Iran strikes Israel on…?” with a clause specifying that intercepted missiles wouldn’t count. Fabian’s report — confirming that a missile warhead had actually impacted the ground — was standing between a lot of gamblers and a lot of money. And so began one of the most deranged campaigns of harassment against a journalist ever (and I’ve seen some pretty crazy campaigns against journalists). It started with polite-sounding emails. A guy named “Aviv” writing in Hebrew, suggesting Fabian’s report didn’t “reflect reality.” Then “Daniel,” a day later, with the same question and a thinly veiled threat: “Sorry for reaching out without a prior introduction, but I assume we will get to know each other well,” he wrote, in a somewhat threatening manner. “I have an urgent request regarding the accuracy of your report on the missile attack on March 10. I would really appreciate a response if possible. There is an inaccurate report from you about the missile attack on March 10, and it’s causing a chain of errors,” Daniel’s email continued. “If you could reply to me tonight… you would be helping me, many others, and, of course, the State of Israel. And along the way, you would gain a good source.” When Fabian didn’t comply, the messages kept coming across every possible channel — email, WhatsApp, Discord, X. Someone fabricated a fake screenshot showing Fabian had agreed to change his story (which he most certainly had not), then circulated it on social media to pressure him further. Someone he knew from another news org even contacted him to say an acquaintance was asking him to convince Fabian to alter the report — and when confronted, that acquaintance admitted he was betting on Polymarket and offered to share winnings if the other journalist could get Fabian to change the story. And then it got genuinely terrifying. After a quiet weekend, someone calling himself “Haim” started sending WhatsApp messages shortly after midnight: “You have exactly half an hour to correct your attempt at influence,” he wrote. “Despite the fact that you received countless inquiries — you insist on leaving it that way.” “If you do not correct this by 01:00 Israel time today, March 15, you are bringing upon yourself damage you have never imagined you would suffer,” he threatened, in a very lengthy message. And “Haim” kept escalating: “You have no idea how much you’ve put yourself at risk. Today is the most significant day of your career. You have two choices: either believe that we have the capabilities, and after you make us lose $900,000 we will invest no less than that to finish you. Or end this with money in your pocket, and also earn back the life you had until now.” After I didn’t respond, as I was asleep, Haim sent me another series of messages: “You are choosing to go to war knowing that you will lose your life as you’ve grown accustomed to it — for nothing.” On Sunday morning, he messaged me again: “You have exactly a few hours left to fix your attempt at influencing [the market]. It would be stupid of you to ignore this.” In short, people who made a bad gambling bet on a prediction market are threatening to kill a war correspondent because his accurate reporting is inconvenient for their wager. They referenced his home neighborhood, his parents, his siblings. They gave him countdown timers. Someone called him pretending to be a lawyer investigating him for “market manipulation.” “If you decide to go with your ego and not with your head, you are leaving behind dozens of wealthy people from all over the world who will know that you performed market manipulation and stole from them. They know who you are, you don’t know who they are. It took them less than 5 minutes to find out exactly where you live … how often you see your lovely parents … and exactly who your … brothers and sisters are.” Fabian reported it all to the police. The threats stopped almost immediately after he went public. In an interview with Charlie Warzel at The Atlantic, Fabian described the thing that should concern us all: there was a moment, however brief, where the pressure almost worked: Warzel: Did you ever think about changing the story? Fabian: For a split second I did. I thought maybe I could be wrong. Warzel: Like, doubting your reporting? After all, you’re making those calls based on other witnesses and videos online. Fabian: I went and checked again with the military. It was a short item, but I reviewed footage of a large explosion. I had eyewitness accounts—people in the area who saw this massive explosion. And then I thought to myself, Why am I doing this? Triple-checking this minor incident, bothering the military again over an explosion in the woods? I did the reporting, and this was the judgement call I made. I think it was accurate, and I will leave it at that. I don’t need to doubt myself about what I published, especially because this is not something that anyone normally would care about unless they had a financial stake in the outcome. As an event in this war, it is not particularly newsworthy. This missile exploded in an open area. It’s 150 words in the live blog. Fabian held the line — but here’s a journalist who covers an active war zone, confirmed his reporting with military sources, reviewed video footage of an explosion, and still briefly questioned his own accurate reporting because a bunch of gambling addicts wouldn’t stop threatening him. And he’s self-aware enough to recognize what that means going forward: Warzel: Do you think this fiasco will stick in the back of your mind as you continue to report on the war? Fabian: Yes. I think it already has. Since then, whenever I report on something, I feel it in the back of my head: What if the Polymarket bettors are betting on this tweet? Or on whether I’m giving an interview about Polymarket? I’m not obsessing over it. Hopefully I won’t get threatened again. But the thought is there. So much for prediction markets producing more accurate information. What they produced here was a chilling effect on a journalist who was already telling the truth. And what guarantee do we have that the next journalist faced with similar pressure will be as willing as Fabian to tell the gamblers to fuck off? This was always going to happen. Not the specific details — nobody could have predicted that a 150-word liveblog item about a missile in a forest would become a $14 million flashpoint — but the general shape of it was entirely predictable. We’d already seen shades of this in sports betting (which is just prediction markets for sports). Athletes have been dealing with a steadily escalating campaign of harassment from bettors who feel personally wronged when a player doesn’t perform to their parlay’s specifications. Now scale that dynamic up from “some guy lost his parlay on a golf tournament” to “$14 million riding on whether a missile was intercepted during a shooting war,” and you can see exactly where this goes. The stakes get higher, the targets get more consequential, and the threats get more serious. Going from Venmo requests to death threats against war correspondents is an escalation, sure, but it’s an escalation along a perfectly predictable trajectory. And while I opened this talking about Goodhart’s law, this is clearly worse: in the classic case, people game the metric by finding clever workarounds. They don’t usually try to put a gun to the metric’s head. But when the “metric” is a human being — a journalist, an athlete, anyone whose actions or reporting can move a market — the incentive to game the system becomes an incentive to coerce, threaten, bribe, or fabricate. It is not just (as in Goodhart’s formulation) about someone “making a measure into a target.” Someone showed up at the measure’s house and threatened its family. Polymarket, for its part, issued a statement condemning the threats and saying it had “banned the accounts for all involved & will pass their info to the relevant authorities.” Bit late for that. The company also said, with apparently zero self-awareness, that “prediction markets depend on the integrity of independent reporting.” They’re acknowledging that journalists are functionally the oracles their entire market depends on. Which means journalists are “targets” by design. The market’s resolution mechanism requires someone external who has no relationship to Polymarket to report accurately on real-world events, and the market’s financial incentives create enormous pressure to corrupt that reporting. The company has built a system that depends on a thing it simultaneously makes harder to do. In short: the company that built a system on claims of more accurate information is creating powerful incentives for people to falsify it. Fabian raised another point in his interview with Warzel that goes beyond the harassment and into something potentially far more worrisome: Fabian: I think there is a big risk of journalists using insider information to place a correct bet and win. I can tell you as a military correspondent that I’m exposed to confidential information that we can’t report. Now there are ways to exploit that. It wouldn’t surprise me if others have. He’s right to worry. Last month, an Israeli military reservist and a civilian were indicted for using classified information to place bets ahead of Israel’s war with Iran. Prediction markets create a mechanism for anyone with privileged information — journalists, military personnel, government officials — to monetize that knowledge without ever publishing it. And the thing is, supporters of some of these markets argue that’s the whole point. Because people with insider information will bet, they believe that the markets will provide the public better information. But it also creates ridiculously perverse incentives for extraordinarily bad behavior. And the legal system is just starting to wrap its head around these things (for example, also this week, Arizona criminally charged Polymarket’s main competitor, Kalshi, with illegal gambling). As Fabian put it, when Warzel asked whether prediction market companies actually want to combat insider trading: Fabian: I don’t think they really want to combat insider trading. What I’ve heard is that those who bet on Polymarket either know the right answer or are wasting their money. So the incentive structure of prediction markets simultaneously encourages harassment of the people whose reporting the markets depend on, creates a pipeline for insider trading by anyone with privileged information, and — as Fabian’s case demonstrates — can literally cause a journalist to momentarily doubt his own accurate reporting because the financial pressure to be wrong is so overwhelming. I’ll admit that I was once genuinely intrigued by the concept of prediction markets. The original pitch was compelling: because people have skin in the game, real information should flow into these markets more efficiently than it flows through, say, punditry or polling. And we all know how weak punditry and polling has been of late. In theory, this idea of “skin in the game” leading to better information makes some sense. In practice, what we’ve gotten is a bunch of speculative nonsense driven by get-rich-quick bros who, when their bets go south, feel entitled to threaten the life of a war correspondent covering missile strikes. The theory assumed that financial stakes would incentivize finding accurate information. What it failed to account for is that, for many participants, it’s easier and cheaper to try to falsify the information than to accept a loss. Fabian’s advice for other journalists who might find themselves in this situation was to follow in his footsteps of going public: Warzel: Do you have advice for other journalists who may experience this type of betting-market harassment in the future? Fabian: Go public. Don’t let the threats force you to change anything. Be honest. I think that’s the best way. It’s a bit stupid of these people to publicly intimidate somebody who can go and instantly tell 100,000 people what these gamblers are doing. That’s my advice. Because if you were to accept money or change your reporting, who knows how these people might extort you later on. If you change your reporting, it’ll be a mess forever. That’s good advice. But he also pointed out the obvious problem: not every journalist will say no. Not every journalist will have the resources, the platform, or the institutional backing to withstand this kind of pressure. And as prediction markets grow and attach ever-larger sums to ever-more-consequential events, the pressure will only increase. We’ve built a system where people can wager millions on the outcomes of wars, and then express shock when they treat the people reporting on those wars as obstacles to be eliminated. We’ve created financial instruments that depend entirely on the integrity of vastly underpaid independent journalism while simultaneously giving millions of strangers a direct financial incentive to destroy that integrity. That seems bad. And the platforms facilitating all of this respond with a press statement and a few banned accounts, as if the problem is a handful of bad actors rather than the fundamental architecture of what they’ve built. Prediction markets were supposed to be information-discovery mechanisms. When you let people bet millions of dollars on whether missiles hit their targets, what you actually get is a threat-discovery mechanism aimed at the people trying to tell the truth. As Fabian told Warzel: “This is war, not a game.” Unfortunately, someone went ahead and turned it into both. And poured a ton of money into it. Filed Under: gambling, information, journalism, prediction markets Companies: kalshi, polymarket Comments on “Prediction Markets Promised Better Information. Instead They’re Creating Powerful Incentives to Corrupt Information.” It could be even worse If we’re going with the concept of journalists being targets of this sort of thing, or people making bets based on insider information, I do kind of wonder at what point we start seeing people doing something like out of ‘Tomorrow Never Dies’: where someone causes specific events to happen because they stand to make a lot of money from it on said bets. Frankly, the whole concept of this being an excellent method of ‘information-gathering’ strikes me as incredibly dumb. This feels like frankly gambling with extra steps, to quote a phrase.