Start with what a $14m market does to a reporter. In March, the Israeli journalist Emanuel Fabian reported that an Iranian missile had struck Beit Shemesh, confirmed by the military and by video of the explosion. Polymarket had $14m riding on whether Iran would strike Israel that day, and interceptor fragments did not count as a strike under the market’s terms. So strangers wrote to him — politely at first, then less so, then from Discord, X, and WhatsApp, and then through a colleague. Someone had already forged an email in his name saying the story would be corrected. Within days the messages were threats to him and his family: “After you make us lose $900,000 we will invest no less than that to finish you.” Another: “Believe me, you don’t want to be their target.”

Finn Brunton’s essay in the September issue of Harper’s uses that case — and a gallery of others — to make an argument that is not about gambling addiction or state revenue. It is about what happens when anonymous money can take a position on reality.

The pattern is consistent:

  • A $32,000 bet on the US ousting Nicolás Maduro was placed hours before Trump announced it; the bettor collected $400,000. A US soldier was arrested in connection with the wager.
  • A bettor won nearly a million dollars by correctly naming 22 of Google’s 23 top search terms for 2025, having previously won $150,000 on the surprise release date of a Gemini model update.
  • Someone won more than $500,000 betting on the timing of Ayatollah Khamenei’s death, shortly before he was killed by an Israeli airstrike.
  • Another won $16,000 by getting 17 Super Bowl halftime-show performer guesses right hours before the show.
  • One bettor allegedly used a hair dryer to alter the recorded temperature at Charles de Gaulle airport to win a weather bet.

The mechanics are the point. Polymarket and Kalshi accounts are not tied to identity, can be funded from anonymous crypto wallets, and can be multiplied — seven or eight anonymous accounts each landing one improbable guess looks less like insider knowledge than one account landing all of them.

Brunton’s frame is a contest between two men who corresponded on the same mailing list. Robin Hanson designed “idea futures” to synthesize an honest consensus of experts — a machine for collective understanding, the lineage every platform still invokes. Timothy C. May, a retired Intel engineer who called himself a crypto-anarchist, wanted a machine to dissolve trust. His 1993 BlackNet hoax — an anonymous marketplace for buying and selling secrets, paid in untraceable credits — was a proof of concept whose value was indirect: not the leaked secrets but the paranoia, the purge, the inability of any institution to function once everyone inside is a possible seller. May then pushed the logic to what he called “abhorrent markets,” a job board for the liquidation of politicians, and an “anonymous fund to collect the money for such a hit.”

Jim Bell’s 1996 follow-on, “Assassination Politics,” supplied the market design: pick a target with a bounty attached, bet on the date of their death, fulfill the prediction yourself, collect the pot. Crowdfunded assassination in which nobody gives an order. “A Kickstarter for murder,” as Brunton puts it.

His argument is that the elaborate crypto-anarchist machines kept failing — Silk Road’s administrator was caught through a single sloppy mistake; the murder-for-hire sites turned out to be scams extorting their own customers — and that anonymous prediction markets are the version that actually works, precisely because they are simple, legal, and mass-market. Gambling on insider knowledge is intuitive. And when the data a bet resolves against is someone’s reporting, you can threaten the reporter. Even without markets on a death date, a market on whether someone will be a candidate in 2028 provides one surefire way to win a bet that they won’t be.

Hence the essay’s conclusion, which is the part worth carrying around: today’s prediction markets are not the fulfillment of Hanson’s vision but of May’s. “Rather than coordinating expert opinions, they deliver uncertainty and distrust.” A system built to make the future more legible has made the present less so — a solvent that dissolves the capacity for coordination. What is being built, in Brunton’s phrase, is a machine for generating chaos.