« Torture vs. Dust Specks | Main | A Case Study of Motivated Continuation »

October 30, 2007


I don't understand how the incentive works as you describe it. The expected cost of harming someone is the same with or without a plaintiff gamble. That is, the expected cost to the harming party of a 50% chance of the complainant doubling claim X is equal to (2X*0.5), or X. It seems the payoffs should be adjusted so that the plaintiff gets a 50/50 chance of, say, tripling the award for this to work as an incentive to discourage small harms.

people don't seem to want law to discourage small harms. But if so, why do we want law to discourage large harms?

The cost of administering a $1,000,000 case shouldn't be 1000x that for a $1,000 case, so the state should focus on big harms.

Assuming visits to the lawsuit randomizing office cost much less than those to court, they'll get lots of trivial cases, doubling up until it's worth the expected cost (lawyers, time, stress, etc.) of going to court. Ambulance chasers will become first aid kit chasers.

In fact, forget expected utility, this is a world of lottery ticket buyers. Joe Gambler will be less scared of being sued when he feels lucky. Sally Stresspill will get ulcers from the fear of losing her house when she bumps into someone. I oppose the motion!

Actually people do want laws enforcing small harms, however, there are many other remedies available to people at that level - laws are just one of them. Social consequences can be worth $50, or $100. Your time isn't necessarily worth what you'll be getting back in damages either, so small harms aren't generally dealt with in a time-consuming manner unless there are circumstances that dictate such an action.

As for large harms, social consequences don't scale nearly so well as people's ability to cause harm, hence, laws.


One-third of my dissertation was on this! I published it here:

Using Lotteries To Expand The Range of Litigation Settlements, 26 Journal of Legal Studies 69 (1997).

Perhaps some of the objections are this: if I harm you by $10, and you successfully double 16 times, I now owe you $655,360. That is unfair to me: I harmed you by only $10, felt in good faith that it wasn't my fault, but here I am losing everything I own because the plaintiff won a 1 in 65,000 shot.

People think courts are there to administer justice, but $655K for a $10 harm doesn't *seem* like justice.

James, neat to see your paper, published before my web essay. You deal with the same mechanism, except for a different purpose. You explore how litigants could achieve mutual gains, but note this won't work for NEV (negative expected value) suits. I was seeing to let law be applied to more harms, and to transform NEV suits into suits that would go forward. Did you not think of this application, or did you not think it a good idea?

Keith, the point is to avoid letting the cost of litigation discourage people from suing.

Recovering, the same mechanisms we use now to punish trivial suits that are large could also be used for small suits.

Phil, I explained you could protect yourself by depositing money that would double when the suit doubled.

Roger, sometimes other social mechanisms deal with small harms, but not always. Why not use law for those other cases?

St. Petersburg paradox, anyone?


See pages 79-80, section III of my "Using Lotteries to expand the range of litigation settlements" article. The section is titled "Asymmetric Information and Negative Expected Value (NEV) Suits."

This paper focuses on reducing litigation costs, but another (unpublished) dissertation essay of mine showed how a plaintiff could get a positive settlement with a suit that has a negative expected value. To understand the logic behind this essay assume:

If the parties go to trial the plaintiff pays $1100 in court costs, the defendant pays $600 in court costs and the defendant must pay the plaintiff $1000. The plaintiff could say to the defendant, if you don't immediately settle the case I will pay $200 of my litigation costs and thus it will then be credible for me to bring the case to trial. I suspect that once I spend $200 you will be willing to settle for more than $200 so my threat to spend $200 is credible. Consequently, you should agree to settle the case now for some positive amount.

Who decides whether the citizen can play his lottery ticket (randomize for a claim)? Someone must judge uniqueness at some point.

This + EY's wankery = me unsubscribing

Great. Now come up with an idea that would stop people from filing frivolous (and malicious) lawsuits.

This proposal seems like you're having your cake and eating it too.

The only useful purpose I see behind this is a deterrent effect - if the $50 in damages I do could cost me $1600, I will avoid doing $50 in damages. If there's some way I can insure against this with, say, a $100 deposit at the beginning of the process, then the deterrent effect is destroyed. If I can't do this, then I could be destroyed by a fairly trivial claim that may not even have been intentional on my part. Because the amount the average individual would need to deposit to break even statistically can be calculated with a high degree of accuracy, there will almost certainly be a lucrative insurance market.

All that's left is that this provides someone with an incentiveto pay $500 to litigate $50 of damages, some small fraction of the time they incur those damages. This litigation is basically subsidized by all the people who insured against loss and had the case thrown out. I'm not seeing why this is a desirable outcome.

If it's an exercise in how to make it possible to litigate such cases, I guess it succeeds, but I don't see why we would want this system, given it has almost no deterrent effect. And given the very obvious and lucrative market in "doubling insurance," I don't see how there could be a deterrent effect.

Not to mention this system seems incredibly easy to game, particularly for plaintiff's lawyers and litigation-obsessed individuals, as well as (possibly) large corporations seeking to bludgeon individuals into submission (corporations might see a LOT of small damages cases, and they could easily pursue all of them and just win big off a few).

First, for what it's worth, I think your link to the original essay is broken - I found it through the Internet Archive here.

I might be missing something, but it seems to me that the downsides of this far outweigh its potential to discourage harming others by small amounts. I can see two serious problems with this idea: the throwing out, by totally random chance, of valid, important suits (by making them worth nothing), and high penalties for inconsequential offenses (curing dandruff by decapitation, as Zappa would have said). Neither of these possiblities should be allowed in a just and reasonable legal system. Furthermore, the fact that such high penalties are possible might provoke an innocent party to settle for a smaller, undeserved sum to avoid the monetary and legal hassle - a whole new arena of extortion to be dealt with. This is particularly a problem as those who would be gambling are unlikely to be those with good legal sense or finances (a wealthy, well-educated lawyer would not need to gamble on a suit).

You might argue that the first issue is not really a big deal, because as you've said, the case wouldn't have gone to court anyway, and the gambling could be optional. However, a proposal that only solves a portion of the problem (those that do gamble, and successfully double) doesn't seem very good to me, especially given the unfairness then to the other party.

I would think that an efficient small-claims court system would alleviate the need for this sort of thing with more justice.

James, I read those pages and I agree there may be cases where a NEV suit might go forward, but neither seem very relevant to my proposal to use lotteries to drastically increase the number of feasible small suits. What say you to that? Most commenters here don't seem to know much law & econ, but if I can't get even you to understand, I'm lost.

Kyle and Psycho, the whole point of the law here is deterrence. So I'm proposing a generic way law could deter small harms.

Do class action lawsuits not acheive a similar effect?


Your idea would indeed create legal incentives for people to avoid hurting others by small amounts, but I'm not sure that this would increase the wealth of society. Consider a simple model:

Assume that the Defendant might engage in an action that causes $100 of harm to the Plaintiff.
Assume that the cost to the Plaintiff of litigating exceeds $100.
Assume that the Defendant gets a benefit of X from taking the action. X>0
Assume that absent implementing your idea NEV suits are never brought

If X> $100 then your idea would destroy wealth because the action is optimal but implementing your idea may cause the parties to pay some litigation costs and could even prevent the defendant from taking the action. If your idea is not implemented then we would achieve the first best outcome for wealth maximization. The fact that the suit is a NEV one means that the harm of the action is relatively small compared to the harm that generates other lawsuits. Thus, actions that create NEV suits are more likely to be efficient than actions that give rise to non-NEV suits.

If X < $100 then your idea might efficiently deter the defendant from engaging in the action. But if the parties can negotiate then by the Coase theorem they should be able to work out an arrangement whereby the defendant doesn't take the action.

We should favor your idea only if we believe that actions that create potential NEV suits are usually inefficient. Also, the higher the costs of litigation, the less willing we should be to favor your idea.

(The link to your full article didn't work for me.)


Actually, your idea with a negligence standard where the plaintiff is only found liable if his action was inefficient would increase the wealth of society. For some reason in the last post I was assuming there was strict liability.


Jim, yes of course but those observations apply to all interactions large or small. If in some area transaction costs are low or most interactions are efficient, allowing suits regarding those interactions can be bad. But surely we don't believe this is the usual case for large harms, or we would not want law. Why would the usual case be different for small harms?

How, exactly, does the money the accused deposits get doubled? Because it would be an easy racket to sue one of my friends for a trivial amount, and then take my chances on repeated double or nothings after he'd placed his deposit (with money I could even lend him).


There is some fixed cost to bringing a lawsuit regardless of the level of damages. The smaller the level of damages the more likely that the cost of using the court system to litigate would be greater than any beneficial deterrence created by the litigation.

Let's say that the social costs of litigation = $10,000 + X(Expected Verdict),
while the deterrence benefit of litigation = Y(expected verdict)
where X and Y are positive but less than 1.

We only want a trial when $10,000 + X(Expected Verdict)>Y(expected verdict). This won't happen when the expected verdict is small.

But under your lottery system we only have a trial probabilistically so this reduces the expected cost of litigation but it might not reduce the deterrence value of suits. So absent a lottery system I think it best that NEV suits are not brought, but I think you might be right that under a lottery system it could be beneficial to have NEV suits.

Your idea reduces the cost of using the court system. By reducing this cost it becomes socially beneficial to have more harms subject to judicial punishment.

James, right that is the idea. But in my experience once people understand and grant that point, they still have strong reservations about using this new ability.


Judge Richard Posner was one of my dissertation advisers. He liked my lottery idea but told me it strongly went against legal culture. I asked him if he would ever use lotteries as a judge. He said no because it would get him impeached.

Wait a sec...

"More generally, you could keep doubling your suit"


"I'd let the person being sued also risk the suit, double or nothing, as long as they showed they were good for the maximum damage amount."


Couldn't a sufficiently rich defendant double it over and over and over again until s/he got "tails" on the coin flip and wiped it out to nothing?

This is one of the best ideas I have heard for a long time!

Diminishing marginal returns set in really quickly for compensation payments. People who sue for damages and win get most of their benefit from psychological effects like having the law validate their complaint, and feeling like the outcome was "fair". And since estimating the monetary value of damages tends to be difficult, many different payment levels can be rationalized as fair. (But not no payment.)

"In fact, forget expected utility, this is a world of lottery ticket buyers. Joe Gambler will be less scared of being sued when he feels lucky. Sally Stresspill will get ulcers from the fear of losing her house when she bumps into someone. I oppose the motion!" and the fact that utility isnt even. I wouldn't take a 50% bet on money cause risk hurts!

The comments to this entry are closed.

Less Wrong (sister site)

May 2009

Sun Mon Tue Wed Thu Fri Sat
          1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30