Inside Spotify and Kalshi's messy breakup

You’ve probably heard quite a few stories about people making bets on prediction markets, then making sure the event happens the way they’ve predicted. You can’t make bets on every number we track for these monthly Garbage Intelligence reports, but you can bet on Spotify streams. For the first time, we have some evidence that these numbers have been manipulated to make money from Kalshi, and Spotify is taking it seriously, so we should too.

Let’s start with the data. On June 29, Malcolm Todd’s song “Earrings” got an enormous bump in streaming numbers, beating its previous highest day by more than half a million streams. Within a few days, Spotify determined that all of these streams were fake and removed them from their records. But we track Spotify data with Kworb, which doesn’t make these adjustments retroactively, so we can see that the streams did jump by around half a million, which were almost all from the US:

A huge red flag for this flood of streams is that Malcolm Todd doesn’t have this kind of fandom. Sure, he’s popular (Ed. note from Ryan — he sounds like Kidz Bop Tame Impala), but nothing about his online presence or promotional cycle would explain such an enormous leap in daily listeners.

So what would explain it? Well, the first person who noticed the leap was Gaeten Dugas, a self-proclaimed “top 100 Kalshi trader” who’s officially promoted the prediction market for years. He posted on X about Todd’s surge as part of a larger accounting of Spotify data that suspiciously overlapped with bets on the Kalshi contract “Who will have a #1 song on Spotify USA in June?” Kalshi would end up paying out bets on Todd, but both they and Spotify told WIRED that they were investigating the matter further and putting checks in place to prevent this kind of manipulation in the future.

Which is surprising to hear from Spotify, considering how much they’ve done already to prevent it. Artificially inflating Spotify streams is a crime, and one that’s been regularly punished. This past March, a man who used a bot farm to stream his AI-generated songs billions of times was found guilty of fraud. A similar case, decided in Denmark in 2024, resulted in prison time and millions in fines for the perpetrator. And that’s pretty unusual.

Since we started tracking social media engagement, we’ve seen signs of manipulation basically everywhere. Instagram just did its biggest bot purge ever in May, we’re still seeing viewbotting on Twitch every month, and X would feel like a ghost town without undisclosed casino ads. These practices are frequently against the terms of service for these apps, but they don’t carry much in the way of criminal consequences.

It’s worth remembering why Spotify is a special case here. The service didn’t grow like a normal social app. It landed with a splash as an above-board alternative to LimeWire and other illegal download sites, backed with the fearsome and fulsome legal force of the entire recording industry. Even if there isn’t much practical difference between Spotify and the other apps we track, faking a listen carries much steeper legal consequences than faking a like or a follow.

Which only matters until you encounter a force more powerful than the law. Kalshi was still illegal in the US less than two years ago, categorized as illegal gambling by Biden’s Commodities Futures Trading Commission. An appeals court blocked the CFTC’s ruling just before the 2024 election, which itself was the centerpiece of Kalshi’s first major marketing campaign. Betting on elections is, and has basically always been, illegal in most of the US — but prediction markets aren’t technically considered betting!

And so on and so on until there’s no more financial regulation anywhere. Prediction markets have a long history, but the modern incarnation led by Kalshi and Polymarket grew out of the infamous Bay Area rationalist community, and hyper-online crypto spaces. Which, if you didn’t know, are both filled with people who genuinely want to overthrow the current socioeconomic order.

That doesn’t really mean anyone who uses a prediction market is automatically a believer in the techno-feudalist “Network State” or whatever. It’s more that the people setting the agenda on these apps are too online to care. Last month, I wrote for Bloomberg Businessweek about how both Kalshi and Polymarket allowed for bets on “mindshare”. That was an AI-generated number that claimed to be an objective rating of online clout — but the rating was so easy to manipulate that the calculators had to change their entire ranking system twice before they started allowing bets. Why would the two biggest prediction markets let you predict a number that was so easy to fix? Because mindshare, that number, was really, really popular on X for a few months. And that beat any consideration about malfeasance, accuracy, or, y’know, responsibility.

This is the bigger reason why, since the Malcolm Todd episode last week, Spotify has started to distance itself from Kalshi. According to WIRED, the company is removing its logos and blessing from Kalshi’s website, even though no one has any hard evidence that Kalshi traders were responsible for this. They’re starting to see how much they stand to lose if prediction markets continue lasering holes in all the structures that society has built up for the sake of social good.

Spotify itself isn’t the most moral or responsible company (just gotta pause writing this to check the window for kamikaze drones). But they profit off the status quo, including the part where you can at least vaguely trust what’s popular online to reflect reality. And Kalshi, along with other prediction markets, are making themselves and their users money off eliminating that.