miércoles, mayo 22

The Wager That Betting Can Change the World

I had never before attended a business conference with a 28 percent chance of an orgy.

But those were the official orgy odds when I arrived at Manifest, a self-described “gathering of forecasting nerds” that the forecasting start-up Manifold Markets put on last month in Berkeley, Calif.

By the second day of the conference, the odds had risen to 47 percent. And on the third day, they reached 100 percent — because there had, in fact, been an orgy. (No, I was not invited.)

This strange blend of data and debauchery — equal parts Math Olympiad and Burning Man — was the dominant vibe at Manifest, which was held in a converted hotel and populated by a crowd of about 250 tech workers, bloggers, economists, students and assorted wonks.

They were there to celebrate prediction markets, online platforms where users can wager on future events — everything from “Will Ukraine regain control over Crimea before the end of 2024?” to “Will Elon Musk and Mark Zuckerberg have a physical fight in 2023?”

On Manifold Markets, users can create a market on any topic and invite other users to bet on it. Winners get bragging rights along with units of Mana, the company’s play-money currency, which they can convert to charity donations or use on other bets.

Prediction markets aren’t a new idea, nor is the hope that betting could produce useful information. Gambling on elections and other political events was common in the United States during the 19th and early 20th centuries. And in countries where political betting is still legal, odds are often cited alongside polls and surveys as a meaningful data point.

But in recent years, prediction markets have caught the attention of a crowd of Silicon Valley empiricists who believe we can fix much of what ails society by betting on our future the way we wager on stocks or sports games.

These people believe the world is full of bad information — biased news, out-of-touch punditry, loony conspiracy theories. Much of this information is spread by people without skin in the game. (Or worse, people with incentives to lie.) And many people have lost faith in the experts and institutions, such as the government and the media, that once served as trusted referees.

Prediction markets, they believe, offer a better way to search for truth — rewarding those who are good at forecasting by allowing them to make money off those who are bad at it, while settling on the facts in an unbiased way.

In the past year, I’ve heard A.I. researchers wagering on the year that we’ll get A.G.I. — artificial general intelligence, a computer that can do anything a human can — and making side bets on, for example, when an artificial intelligence will win a Nobel Prize, or whether an A.I.-generated movie will be nominated for an Oscar.

Prediction markets have also sprung up around major world events, like the war in Ukraine. And in fields like venture capital and economic forecasting, trend-spotters have started looking at prediction markets for signs of the future.

This year, when a group of South Korean scientists claimed to have created a room-temperature superconductor called LK-99 — a ground-shaking discovery with huge implications, if it was true — prediction markets lit up with bets about whether the discovery would prove to be credible. (It didn’t.)

I went to Manifest to try to understand the appeal of prediction markets, and get inside the heads of the people who are obsessed with them. And while I wasn’t convinced that these markets will be an important part of our future, it’s not a possibility I’d bet against.

The basic idea behind Manifold Markets and similar platforms, such as Kalshi and Polymarket, goes like this: Markets aggregate information. The more information they aggregate, the more accurate they tend to be. And if enough people make enough bets, with enough information behind them, markets can tell you something useful about the future.

Most of us accept this principle when it comes to investing. If the price of Apple stock rises 10 percent one day, or falls 20 percent the next, we assume that it’s because smart investors with access to good information have changed their minds about the company’s prospects, and that it’s not just a random blip.

Research has also shown that betting markets on election outcomes can be more accurate than polls. (Although their recent record has been more mixed.)

But how would markets do at predicting other things? Could you, say, figure out whether Taylor Swift’s next tour will sell more tickets than her last one not by asking music experts or concert promoters what they think but by opening a market that would allow everyone — fans, other musicians, hedge funds, even Ms. Swift herself — to weigh in? And would that market get more accurate over time as new information came in?

If everyone bet on everything, in other words, would our view of the future be more grounded in truth?

That question started percolating in the 1990s among economists who wondered if the internet — which allowed markets to spring up in an instant, and attract participants from around the globe — could bring the idea of universal, real-time prediction markets to life.

Early online prediction markets, such as Intrade and NewsFutures, got some traction by allowing users to wager on elections, sports games, entertainment events and more. But most either shut down or were forced to stop taking real-money bets by anti-gambling laws, which prohibit most kinds of online gambling.

In recent years, though, the idea has been revived by the Rationalists, a movement of cerebral data obsessives who have become a cultural force in Silicon Valley. Many prominent Rationalists are fans of prediction markets, and have encouraged other Rationalists to use them to test their own views.

“Prediction market prices are the means by which a high-functioning civilization knows what it knows,” said Eliezer Yudkowsky, an A.I. safety researcher and prominent Rationalist, who attended Manifest wearing a glittering gold hat.

In the Rationalists’ view, prediction markets are an essential part of a healthy civic ecosystem, and a necessary check on experts and mainstream authorities.

They believe that prediction markets work because they harness the wisdom of crowds, and filter out noise and bias by reducing contentious debates to simple yes-or-no questions. Good forecasters win more bets over time, while bad ones lose money and influence. And everyone learns by watching prices move in real time, as more information is added to the market.

Some even believe that prediction markets could keep extremists and conspiracy theories at bay by raising the stakes of fringe views. QAnon believers who insist that Democrats are harvesting the blood of children may balk at the idea of betting next month’s rent on it — which would prove, to anyone watching, that they weren’t that serious.

“We live in this delusional world full of things that people are cheering for,” Mr. Yudkowsky said. “And if they had to bet money, boy, would they back off quickly.”

In the betting-filled utopia the Rationalists envision, leader boards would rank pundits by the accuracy of their forecasts, and we’d pay attention to only the provably prescient ones. Businesses would track prediction markets to figure out which products to build, or which competitors to worry about. Governments would lean on prediction markets, not polls or lobbyists, to figure out which policies to pursue. And contrarians with unpopular (but correct) views could make gobs of money betting against the odds.

Of course, there are giant obstacles to that future. Prediction markets don’t work well if few people use them, or if participants all have identical information about something. (For example, you wouldn’t learn much from the prediction market “Will the sun rise tomorrow?”) They don’t work for more subjective or hard-to-measure questions. (Who decides, for example, if an A.I. has surpassed human intelligence?)

Experts have raised other issues with real-money prediction markets — that they could allow rich people to distort public opinion by betting huge sums of money on their preferred outcomes, that they can encourage illegal or immoral behavior, that insider trading could spoil them.

But if these problems could be overcome, fans believe, these markets could bring logic and intellectual rigor to a world that badly needs it — similar to the way short-sellers on Wall Street think their ability to bet against a company’s stock provides a necessary check on corporate mismanagement.

The Rationalist revival has put wind into the sails of start-ups like Manifold Markets, which was initially funded by a grant program run by Astral Codex Ten, a Rationalist blog that has promoted prediction markets. (It also received $1 million from the FTX Future Fund, the philanthropic arm of the bankrupt crypto exchange whose founder, Sam Bankman-Fried, is a fan of prediction markets.)

Most prediction markets are still tiny, by stock market standards. (Manifold Markets has about 43,000 users, according to the company.) But defenders say they’re still good enough to be useful.

“It’s roughly true that in every domain in which we’ve been able to compare data from a prediction market to alternative forecast mechanisms, the market has done better,” said Justin Wolfers, a professor of public policy and economics at the University of Michigan.

“If you want to predict which horse is going to win the Kentucky Derby,” he added, “you’re better off tracking the betting odds than asking the experts.”

Austin Chen, 28, a Manifold Markets co-founder, told me that even though the company used fake money, its prediction markets were well calibrated — that is, when the site’s users predict a 70 percent chance of something happening, it actually happens roughly 70 percent of the time.

Mr. Chen is a true believer in prediction markets. (He even created one before proposing to his wife.) And he said that even though individual markets could be wrong, he believed that prediction markets, on the whole, were good sources of wisdom.

What they are not, at least where real money is concerned, is legal. This year, the Commodity Futures Trading Commission rejected a proposal by the prediction market start-up Kalshi to allow its users to wager on which party would control Congress — saying that allowing users to gamble on elections would be “contrary to the public interest.” That agency also fined Polymarket, a cryptocurrency-based prediction platform, $1.4 million for offering unregistered options trading last year.

Most people I met at Manifest dismissed these concerns, and thought most real-money prediction markets should be legal. (Some drew the line at markets related to the deaths of public figures — which they said could encourage assassinations.) And few expressed any moral qualms about a world where gambling on everything is encouraged.

“Prediction markets are a lot better than gambling at a casino, or betting on sports,” Mr. Chen said. “The betting serves a higher purpose, of helping the world get better information.”

For two days, the crowd at Manifest — mostly men in their 20s and 30s, who all seemed to know one another from Twitter — crowded into rooms and a sun-drenched courtyard to hear from speakers like Nate Silver, the founder of the forecasting website FiveThirtyEight, and Robin Hanson, an economist at George Mason University who is considered by some to be the godfather of prediction markets.

Aella, a Rationalist sex researcher and writer, led the crowd in a “spicy live polling” session that required participants to sort themselves according to intimate personal details. (For example, how many psychedelic drugs they had taken, or how polyamorous they were.)

Then they partied. At night, there were poker games, wrestling matches, a Magic: The Gathering tournament and a karaoke contest. Mr. Yudkowsky debated the left-wing YouTube streamer Destiny. Richard Hanania, the conservative commentator, signed copies of his book on wokeness. A shirtless man did acro-yoga near a fire pit.

And, of course, they made markets — lots and lots of markets, mostly jokes about the conference itself.

“Will Grimes show up at Manifest?” (Answer: No.)

“Will anyone walk around in a fursuit at Manifest?” (Answer: Yes.)

“Will Jimmy Carter die during Manifest?” (Answer: Thankfully, no.)

There are downsides to running a conference this way, to say nothing of a society. But at Manifest, where the money was fake and the mood was exuberant, nobody felt much like hedging.

“I spend my time on Manifold that I used to spend on Twitter,” said Joshua Fleming, 28, a civil engineer from San Diego. “It’s kind of a more fun way to follow the news.”

Mr. Fleming estimates he makes dozens of bets a week, on topics including politics and gaming. He likes earning Mana, which he can then donate to charity. But mostly, he said, he likes being proved right.

“There’s a competitive aspect to it,” he said. “I feel good when I win.”