Let’s Replace 50% of the Justice System with Prediction Markets

I swear it’s a good idea

The legal system has three desiderata — that it be prompt, predictable, and just. Unfortunately, our present system largely fails these prongs. The legal system is not fast, with it often taking years for cases to resolve. It is not reliable in its verdicts, with results often informed by jury sentiment or caprice rather than legal principles. Justice is not a consequence of the legal system per se, but rather the law; and if the law is just and the legal system slow and errant, that is no justice. 

An alternative system would be to use prediction markets to decide cases. Rather than have every case go to trial if it isn’t settled, we first have a market on what the outcome would be were it to go to trial. There is a period for bidding, at the end of which some fraction of cases are won immediately by the party which has a greater chance of winning the case. The remaining cases proceed to trial, although they may still settle if they wish. If the case is decided by prediction market, the money is returned with interest (it can be invested in treasury bonds in the interim). If it goes to trial, and you were wrong, you lose the money you put up. Similar markets

Using prediction markets would help with all three desiderata, but especially with capriciousness. Suppose that the even odds of a case being won by the defendant are 70-30. In a world where everything goes to trial, the defendant will lose 30% of the time, despite having a better case. Since payouts are asymmetric, this incentivizes people to bring frivolous suits in order to extort a settlement. In the world where prediction markets are used, you gain nothing for bringing a weak case, rather than being advantaged for it. It’s not even that strange a system, in truth, as it resembles what is already done in civil suits. No one wants to go to trial — it’s too expensive. Instead, the discovery process is really about finding out how strong the other’s case is, and then settling. This is simply another data point, which will make settlements more accurate. 

There are some problems with prediction markets, but I believe they are answered either in this use case, or by specific tweaks to the system. Prediction markets are often illiquid, but here there are two participants with a very strong interest in putting money in. In order for your case to have a chance, the plaintiff must put money in, and so will the defendant to beat them out. The subsidy is built in.

You can allow for verdicts which might find some elements in favor of both parties by having prediction markets for each specific element of the case. Everyone states precisely what argument they seek to prove, and a future trial would say whether or not they proved it. Obviously, when we decide whether the case goes to trial the whole case would go to trial, not just part; but we can allow for split decisions, such as that a company violated section A of a contract but not section B.

The usual objections to prediction markets have been long answered, and I scarcely find them worth answering these days. No, someone would not be able to manipulate the market, unless they were willing to risk losing a ton of money. The most serious concern I can think of would be a mad rush of money into some markets right before the resolution happens. We should expect the amount of money put in to equal the expected gains from winning the case by default judgement; if it comes in late, and the other side lacks a chance to respond, this would return us to the bad world of asymmetric judgements. We could solve this by requiring a minimum amount of time after the last transaction in the market before the case resolves, thus allowing any interested party a chance to respond.

Our present day justice system is kludgy and slow. We should not think it is even close to optimal, and there are a lot of possible solutions. Let’s try something! This can be rolled out on in one district, and if it goes well, rolled out nationwide.