Wednesday, May 01, 2013

The Importance of Right Makes Might

I have been thinking about feud systems of law enforcement, mostly because of the book I am writing on legal systems very different from ours. Feud is a system of decentalized law enforcement; its essence is that when you wrong someone, he threatens to harm you unless you pay him suitable compensation. Such systems are common in the historical record. Aside from saga period Iceland, which I have written about in the past, they exist in traditional Somali law, among modern Rominchal gypsies, and arguably provided the foundation under Jewish, Islamic and Roman law, all of which contain what I view as fossilized remnants of feud. I have just written a piece arguing that the same logic can be found in modern high tech patent litigation.

Before going on, I should point out that the words "feud" and "feudal," despite similar sounds, have no connection with each other. They describe different institutions and come from different linguistic roots. 

The essential requirement for a feud system to work is that the threat to harm someone for wronging you is more believable if you actually have been wronged than if you have not—that, in other words, right makes might. Without that, threats can as readily be used for extortion as for law enforcement.

Different feud systems get that result through different mechanisms. Among modern Rominchals, both parties know that if you really have wronged me in terms of the norms of our community and I respond by beating you up, my friends will support me and your friends won't support you. Saga period Iceland had a much more elaborate version of the same system  about a thousand years earlier, in which right made might via a court system and explicit law. You wronged me, I sued you, the court found that you owed me a damage payment. If you did not pay I got the court to outlaw you, at which point you had two weeks to leave Iceland. After that it was legal to kill you, tortious to defend or shelter you. The law code and court system were analogous to ours, but enforcement of the verdict was entirely private, via  feud.

For the modern instantiation, consider patent litigation between two high-tech firms. One reason for Apple not to sue Samsung for infringement unless it has a strong case is the risk of losing—but that might not be sufficient, given both legal uncertainty and the market advantages to creating uncertainty that will make it harder for Samsung's phones to compete with the iPhone. But another reason is the risk that Samsung can, and very likely will, counter sue. The accumulation by such firms of a large inventory of patents that they never expect to use may be viewed as the equivalent to a saga period Icelander accumulating weapons and allies, just in case he ever needs them. Ideally, in a world with court error, the combination of court plus feud does a better job of deterring weak suits than court alone.

That mechanism does not work if the suit is by a non-practicing entity, referred to by critics as a patent troll, a firm that accumulates patents in order to use them in law suits, not to practice them. Such a firm is immune from retaliation. 

Which raises a problem, one possible solution to which, borrowed from Periclean Athens, is part of the subject of my new article.


At 2:56 AM, May 02, 2013, Blogger Tibor said...

In the article you state:

"If we assume that the cost of negotiating such a settlement is negligible and that the parties settle for the middle of the bargaining range—the defendant pays the plaintiff his expected return, the probability that he will prevail times the damages he will be awarded if he does."

I think it would be better to say "defendand palys the plaintiff his expected return minus the probability of losing times the absolute value of ligitation costs". While it is quite clear that in the case of a one time game (one lawsuit and expecting none to come) it is best for the defendant to pay the plaintiff the sum you mention, since his expected cost of the ligitation is exactly the expected gain of the defendand and your value is lower, what you mention is not the expected return (of the plaintiff) itself and it is a bit confusing at a first glance.

At 5:15 AM, May 02, 2013, Blogger David Friedman said...

If I correctly understand you, I think I disagree with your version. Whichever way the case goes, each party (in the U.S.) will normally pay his own litigation costs. If they are the same for both parties, as I think I was assuming, they don't affect the location of the midpoint of the bargaining range.

At 12:20 PM, May 02, 2013, Blogger Tibor said...

I see, then the problem is me not knowing the U.S. law :) Here, in the Czech republic, it is the losing side that pays the costs. Or at least I think so (maybe it is only for some cases, or up to some amount). Fortunatelly, I've never taken part in any case and fortunatelly I don't study law. It might be interesting in a common law system, but here my impression based on talking to law students is that most of their studying consists of memorizing legislature that keeps changing anyway. And since the courts cannot change the law themselves and precedents have no legal value, it makes the subject of law much less interesting.

In this case, you're right (since you're assuming the ligitation costs to be equal for both parties).

At 12:48 PM, May 02, 2013, Blogger David Friedman said...


"Loser pays" is the "English system," although it's actually "lower pays what the judge things the prevailing party's legal costs should have been." In America, with a few exceptions, each side pays its own legal costs.

At 1:32 PM, May 02, 2013, Blogger Tibor said...

David: I see. I looked up the czech system and I cite (and translate):
"The side of the lawsuit which wins the case is granted compensation of the ligitation costs necessary for an effective claim or defence of rights against the losing side - which then bears all his costs of the ligitation and pays the costs of the winning side as well. That mainly includes the court fee, flat compensation for the so called "palmární" (I don't know how to translate that, that is some law term I see for the first time) costs of legal representation of the winning side, his own legal representation costs, costs of finding, acquiring and creating proofs, including the so called witness fee."

So it seems the system is similar to the UK one - since there seems to be a flat rate for the costs of the other side above which, presumably, even the winner pays his expenses.

At 1:46 PM, May 02, 2013, Anonymous Eric Rasmusen said...

A curious remnant of private enforcement of law is how judgements are collected. I don't know a lot about that, but I think if I win a lawsuit against Smith, it's up to me to get Smith to pay. If he doesn't pay, then I go to court again and ask the judge to let me seize some particular assets of Smith. I'm responsible for finding what those assets are, and if I can't name them, I can't ask for permission to seize them. Somehow the sheriff can but perhaps need not help me to seize assets. Maybe I need him to be present if I have to break into a car or a house to get an asset. In any case, you might find out more accurately than I'm telling it what happens, for your book. It's a hugely important part of the law which academics neglect. For example, I've heard that it's a good idea for a lawyer to ask the defendant if he drove to court that day, and to prove it by letting him see the keys. The reason is that then the lawyer knows the defendant has a car to seize later.

At 1:48 PM, May 02, 2013, Blogger Eric Rasmusen said...

You probably know this essay, but here it is:

"Revenge is a kind of wild justice; which the more man's nature runs to, the more ought law to weed it out. For as for the first wrong, it doth but offend the law; but the revenge of that wrong putteth the law out of office."

Bacon was a lawyer, so he didn't like the competition.

At 1:53 PM, May 02, 2013, Anonymous Anonymous said...

"If we assume that the cost of negotiating such a settlement is negligible and that the parties settle for the middle of the bargaining range—the defendant pays the plaintiff his expected return, the probability that he will prevail times the damages he will be awarded if he does."

I also found this passage confusing. Why is the plaintiff's expected return = "the probablility that he will prevail times the damages he will be awarded if he does."? Is that a standard means of calculating settlements in the US?

At 4:19 PM, May 02, 2013, Blogger David Friedman said...


I am using "expected return" in the conventional sense of "expected value." The expected value of a lottery, a set of outcomes each of which has a probability and a value, is the sum of probability times value over all the outcomes. In the context of gambling, it's what your average return will approach if you take the gamble many times.

At 4:22 AM, May 03, 2013, Blogger Tibor said...

This comment has been removed by the author.

At 4:23 AM, May 03, 2013, Blogger Tibor said...

This comment has been removed by the author.

At 4:32 AM, May 03, 2013, Blogger Tibor said...


That is, if you view the judge's decision to be based on a random distribution with each outcome with it's own probability, then the expected return is simply a weighted average of the sum the court decides the other side has to pay. It is like rolling an n-sided loaded (since the outcomes are not equally probable) dice where each of the sides has a sum of money assigned that has to be paid if the number comes up. Of course, one could see it also in an infinite countable settings where the average would be over an infinite sum or even uncountable infinite settings with infinite amount of different sums, probably with some absolutely continuous distribution and then the weighted average turns into an integral and it could give more accurate estimates that way, but the intuitive idea remains the same as with the "finite sided loaded dice".

Of course the real probabilites are unknown and the sums of money assigned to them as well, but here is some information about them available and so both sides of the dispute can make estimates, so the example mentioned in the article is a little bit simplified. Also, of course the information is never entirely symetrical, one side knows something the other does not and so their estimates of probabilites and damage costs vary, therefore so does their estimate of the expected return. It would be interesting to work out the arguments in this more general framework, but I think it is noted at the end of the article that something like that is to follow.

If I can make a suggestion - it would also be nice to include bayesian inference - the parties know the judge they would be assigned to, can observe his verdicts from similar cases in the past and can make posterior estimates of the probabilities of outcomes and amount of damages he acknowledges based upon that prior information.

Also, it might be interesting to also consider the "English (and Czech)" case in which the loser pays the ligitation costs (or a fraction of them, or what the judge thinks the cost should be which would add another source of randomness).

At 1:46 PM, May 07, 2013, Anonymous Anonymous said...

On the subject of bogus patent-and-copyright lawsuits, you might enjoy this post and the amusing court decision it links to.


Post a Comment

<< Home