I had an interesting email exchange today with John Mackall of the law firm Seed Mackall LLP in Santa Barbara, and he encouraged me to post it on this blog.
On Wed, 24 Sep 2008, John Mackall wrote:
> Dear Prof. Roth:
>
> I read about your kidney market contributions in the most recent
> Harvard Magazine, and since then I have reviewed your blog.
>
> For several years I have speculated about several new ways to approach
> litigation, which must certainly be the biggest and least organized
> "market" in the country.
>
> I have wondered about how game theory might be applied to the
> resolution of thousands of lawsuits at a time, for example. The idea
> would be to select from game theory one or two methods of dispute
> resolution, and then see if it would be possible to identify a set of
> perhaps 1,000 lawsuits that would be particularly amenable to
> resolution in that manner. For example, three party car accidents in
> which four or five large automobile insurance carriers share the majority of the liability. These could be resolved much more efficiently in large groups-100 or
> 1,000-rather than case by case. Cake cutting algorithms or other game
> theory tools would be much easier to use in bulk.
>
> I have also wondered about lawsuits among Fortune 1000 companies.
> There might be cases in which Company A is suing Company B for $10
> million, and Company B is suing Company C for the same amount, and
> Company C is suing Company A for the same amount. Simple chains like
> that might be identified, in which the chain might be resolved much
> more easily than any one link (lawsuit) in the chain. More complex,
> non-obvious chains might also be identified. There are many
> sophisticated analytical systems in math, and in finance there are
> many tools, that could be applied to identify and resolve various
> large groups of seemingly unrelated lawsuits, particularly if there is
> some cohesion to the larger group.
>
> My goal is to put together an academic group that would draw on
> business, law and certain hard sciences-math (such as number theory
> and game theory) and statistics especially-to see if there is some way
> to settle thousands of cases at once in ways that have not yet been
> conceived. The first step would be to identify theoretical
> approaches, and the next step would be to consider whether there are
> "sets" of lawsuits that meet the theoretical requirements.
>
> This idea falls somewhere between grand and grandiose. I am not sure
> where, exactly. If you have any interest, let me know.
>
> Sincerely,
>
> John R. Mackall
> Harvard '71, Stanford Law School '73
From: Alvin E. Roth
Sent: Wednesday, September 24, 2008 11:18 AM
To: John Mackall
Cc: aroth@hbs.edu
Subject: Re: Designing Markets--Litigation
Hi: the idea sounds interesting, but as a natural skeptic (as market designers must be) let me raise two top of the head concerns.
In an insurance pool, while the insurance companies have many cases, the victims have only one, and might not be in a position to accept a compromise that was good on average if it didn't work for them.
The pool of corporate lawsuits sounds much more likely not to have that problem. But (like patent pools), if there's going to be a pool of lawsuits, maybe I should file some extra lawsuits, i.e. there could be bad incentives that would change the pool of lawsuits, even if you had a good way of litigating an existing pool of lawsuits.
So...I think it's an interesting idea, and I'm happy to hear how it develops, but those are my concerns on first blush.
cheers,
al
Date: Wed, 24 Sep 2008 14:19:48 -0700
From: John Mackall
To: Alvin E. Roth Subject:
RE: Designing Markets--Litigation
Al:
Thanks for your prompt response.
Both points have occurred to me. One I can solve. In the insurance cases, I specify a set of multiple defendant cases for a reason: the defendants would resolve the liability among themselves, and leave the plaintiff out of the equation. The resolution would not be of the plaintiff's lawsuit itself, but of the secondary question as to how the damages are to be allocated. They can then present a united defense.
Your second point raises the possibility of gaming the system by filing lawsuits to score points in the game. Yes, this possibility exists, and I think it would be likely in some degree. I don't think it would swamp the overall benefits however.
I would be delighted if you would simply mull this over. Perhaps in the future you might have a student or two that would be interested, and if so, I would be delighted to help in any way.
Sincerely,
John Mackall