Hacker News Reader

Comment by __MatrixMan__ | Hacker News Reader

I hope we manage to leverage prediction markets to actually achieve goals rather than just making a casino out of it.

For instance, if you spot malware in a commit you could bet heavily against it being merged, and that would attract the maintainers' attention, and they'll see what you see and not merge it, and you get paid for the code review--that money would come from whoever bet that it would get merged, which you could require be the author of the malware. I haven't worked it out entirely but it seems that there are opportunities to build games that reward dilligence and transparency and penalize deception and spam.

View on HN

It’s not clear to me how you’d get anyone on the other side of that market. Anyone who is capable of understanding the wager would see the same thing you did, and not bet in its favour. Anyone who can’t understand the code will just pass the bet by and play some other market they do understand. The malware author is presumably not someone you can compel to do anything, never mind pay out a losing bet.

The way I imagined it, the maintainers wouldn't be considering a commit whose author was not willing to bet on its success.

This thought experiment took part in a world where the web was significantly worse than our own: hoards of malicious AI's and precious few humans trying to not be mistaken for a malicious AI. Of course a pre-existing trust relationship is much better, but ideally there'd be a way for untrusted authors to make it through to a real human somehow. Attaching money to the commit would be one way to do that.

However, your bet against it being merged incentivizes contrarian behavior by the maintainer in order to claim your wager for himself.

Similarly, betting that a public figure will still be alive a month from now is functionally equivalent to putting out a hit on him.

I assumed that the maintainer chose to associate PR-creation with a prediction market because they wanted a way to reward reviewers. I suppose there's nothing stopping them from using it to instead prey on reviewers... but that's not exactly a recipe for attracting motivated reviewers to your project.

As far as using prediction markets to put a hit out on people... How does that differ from how it's done now, except that the flow of money becomes part of public record where it's more likely to appear as evidence against you in court?

Anyhow, like most online media, moderation will be necessary. If we can't game-theory our way to a nice solution, we can just do the normal thing and ban the activity we don't like. I just think that where possible, designs that align incentives are likely to be more successful. Prohibition is generally ineffective, expensive to maintain, or both.

Alternatively, you can bet a few tens of thousand that your enemy is going to be alive in a few weeks and let the market handle the rest.

Why not just bet heavily against and then inform maintainers? By just betting on it instead it makes you look like you, or someone you know planted the malware

That is what I meant to say, that you'd inform the maintainers along with your bet against the commit. In this thought experiment I assumed that the maintainers are already being spammed by AI so heavily that the bet is necessary to get their attention. (Neal Stephenson had something similar going in in Anathem, he called them "bogons".)

In the case where you're betting heavily in favor of a commit, maybe because you've reviewed it and think it's good, maybe because it contains malware you want to inject... you'd be attracting reviewer attention to that commit because if they can talk the maintainers out of it they end up with more of your money.

Probably the best strategy for a malicious committer would be to sneak through a low value nothing-to-see-here commit, because the low bet would not attract extra reviewer attention, so the maintainers have to set it high enough that it still incentivizes review.

I don't want to live in this world, by the way, I'm just afraid we might have to.

I was thinking of a similar idea to this, but for news/tweets/posts. As a consumer of media, I might decide to only read media with $x staked, so AI media factories need to be willing to stake that much to reach audiences, and will get penalized when they are wrong…

I imagine a hard problem is building a system to resolve these markets.

Reply on news.ycombinator.com