Classified Intel, Thirteen Bets, $409,881

Here's what prediction markets were sold as: a wisdom-of-crowds mechanism, aggregating dispersed information into prices that beat pundits and polls. Here's what they've quietly become: a place where whoever knows the most wins the most, and for a while, nobody checked whether that knowledge was supposed to be secret.

The arrest of Army Special Forces Master Sergeant Gannon Ken Van Dyke makes that second sentence hard to ignore.

Thirteen Bets

According to the Department of Justice, as reported by Engadget, Van Dyke created a Polymarket account on December 26, 2025. Over the following week — December 27 through January 2 — he placed 13 bets related to Nicolás Maduro. The positions he took included whether US forces would be in Venezuela by January 31, whether Maduro would be out by that same date, and whether Trump would invoke War Powers against Venezuela. He allegedly walked away with $409,881 in profits. The US Attorney for the Southern District of New York announced his arrest on charges that include using confidential government information for personal gain. Per the indictment described by The Verge, Van Dyke was directly involved in the planning and execution of the operation — called "Operation Absolute Resolve" — to capture Maduro.

He didn't stumble into a lucky guess. He allegedly knew the answer before the question closed.

The Infrastructure Problem

What makes this worth more than a single criminal case is the context Engadget quietly drops alongside it: a separate incident in which a hairdryer was allegedly used to manipulate temperature sensors at Charles de Gaulle Airport in Paris, generating winning Polymarket bets on daily temperature fluctuations. French authorities noted the official readings at the airport spiked twice in the past month, and on both occasions, bettors appear to have profited. The temperature sensor has since been moved. Polymarket, per that reporting, did not force anyone to return winnings. The site is still running temperature bets around Paris.

Two stories. Different methods. Same shape: someone with access to information that the market doesn't have, converting that edge into cash. One involves a hairdryer. One involves classified military intelligence and a covert operation to capture a head of state. The scale is different. The logic is identical.

Prediction markets have always had an information asymmetry problem baked in — that's partly what makes them function. But there's a meaningful difference between being well-informed and being classified-briefed. The former is supposed to be an edge. The latter is supposed to be a crime. What Van Dyke's case reveals is that the infrastructure to commit that crime was right there, accessible, with no apparent friction, until the DOJ showed up.

Silicon Valley spent years arguing that prediction markets were just a smarter way to surface truth. The argument had merit in narrow domains. But "smarter" only holds if the participants are drawing from publicly available signal. Once you allow that the best-informed bettors might be informed by things the public legally cannot know, you've built something closer to an insider trading terminal with a cleaner UI.

The government noticed. Eventually. After the operation concluded. After the money was made.

The hairdryer people, at last check, kept their winnings.