Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: How I Learned To Stop Trusting Prediction Markets and Love the Arbitrage, published by orthonormal on August 6, 2024 on LessWrong.

This is a story about a flawed Manifold market, about how easy it is to buy significant objective-sounding publicity for your preferred politics, and about why I've downgraded my respect for all but the largest prediction markets.

I've had a Manifold account for a while, but I didn't use it much until I saw and became irked by this market on the conditional probabilities of a Harris victory, split by VP pick.

The market quickly got cited by rat-adjacent folks on Twitter like Matt Yglesias, because the question it purports to answer is enormously important.

But as you can infer from the above, it has a major issue that makes it nigh-useless: for a candidate whom you know won't be chosen, there is literally no way to come out ahead on mana (Manifold keeps its share of the fees when a market resolves N/A), so all but a very few markets are pure popularity contests, dominated by those who don't mind locking up their mana for a month for a guaranteed 1% loss.

Even for the candidates with a shot of being chosen, the incentives in a conditional market are weaker than those in a non-conditional market because the fees are lost when the market resolves N/A. (Nate Silver wrote a good analysis of why it would be implausible for e.g. Shapiro vs Walz to affect Harris' odds by 13 percentage points.) So the sharps would have no reason to get involved if even one of the contenders has numbers that are off by a couple points from a sane prior.

You'll notice that I bet in this market. Out of epistemic cooperativeness as well as annoyance, I spent small amounts of mana on the markets where it was cheap to reset implausible odds closer to Harris' overall odds of victory. (After larger amounts were poured into some of those markets, I let them ride because taking them out would double the fees I have to pay vs waiting for the N/A.)

A while ago, someone had dumped Gretchen Whitmer down to 38%, but nobody had put much mana into that market, so I spent 140 mana (which can be bought for 14-20 cents if you want to pay for extra play money) to reset her to Harris' overall odds (44%). When the market resolves N/A, I'll get all but around 3 mana (less than half a penny) back.

And that half-penny bought Whitmer four paragraphs in the Manifold Politics Substack, citing the market as evidence that she should be considered a viable candidate.

(At the time of publication, it was still my 140 mana propping her number up; if I sold them, she'd be back under 40%.)

Is this the biggest deal in the world? No. But wow, that's a cheap price for objective-sounding publicity viewed by some major columnists (including some who've heard that prediction markets are good, but aren't aware of caveats). And it underscores for me that conditional prediction markets should almost never be taken seriously, and indicates that only the most liquid markets in general should ever be cited.

The main effect on me, though, is that I've been addicted to Manifold since then, not as an oracle, but as a game. The sheer amount of silly arbitrage (aside from veepstakes, there's a liquid market on whether Trump will be president on 1/1/26 that people had forgotten about, and it was 10 points higher than current markets on whether Trump will win the election) has kept the mana flowing and has kept me unserious about the prices.

Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org