Monday, September 19, 2011

Robot Negotiators

With computers already able to replicate a number of routine tasks performed by lawyers (e.g., basic will drafting, document review in discovery), can negotiation be far behind?  A recent article in the Economist talks about how game theory can help design computer programs that, for example, can help couples negotiate divorce settlements. All the parties need do is secretly assign their own personal values to the assets that need to be distributed in a divorce, and the machine will spit out the solution for them, saving them the trouble of figuring out how to conduct the tricky back-and-forth negotiations that would otherwise be necessary.

There are also on-line programs available to assist parties in conducting various kinds of business negotiations as well as settlement of legal disputes. These programs enable each side to input their bottom line numbers without revealing them to the other side, allowing parties to find out relatively quickly and painlessly if an agreement can be reached.

Does this mean, as a post on ADR Prof Blog asked, that we don't need to teach negotiation skills to young lawyers anymore? Does it mean that mediators will soon be replaced by computers? Perhaps I should be worried about these innovations, but I'm not. In fact, I think it's probably a good thing that some of the game-playing aspects of negotiation can be emulated by computers. Those are the parts that make a lot of participants uncomfortable anyway, and lead some participants to question the legitimacy of mediation. Some people distrust mediation because the process has no rules, and because its results seem to depend too much on the negotiating skills of the participants and their attorneys. In contrast, the court system at least aspires to produce results that accord with the law, and protect litigants' rights. (Of course the results in court are also affected by the respective skills of each sides' attorneys, and a lot of other extraneous factors, but that's a whole other story.) What we're talking about here is whether the process of mediation itself can be made more predictable and fair. If the game-playing aspects of negotiation can be more regularized, for example by teaching everyone some basic negotiating skills, or by turning over some of the back-and-forth to a machine, participants might feel reassured that the playing field has somewhat leveled, and results made more routinized and fair.

If we decide that it would be a good thing to take some of the mystique out of negotiations, does that mean that mediators and parties' attorneys are soon going to be out of a job? I think not. Even if we could program machines to weigh all of the values that people assign to their interests, and produce a fair result, professionals are still needed to help the participants identify the goals that are most important to them in the first place, as well as to help parties evaluate the costs and risks in pursuing their claims through whatever means are appropriate. Discussing the Economist article, a post on Associate's Mind glosses over this problem, suggesting that it is inevitable that computers will simply take over the whole negotiation process:
What if parties could merely input points of data at the beginning of filing a law suit, and let a computer decide what is a fair outcome?  Would clients prefer that to years of discovery and legal bills, all purely for jockeying and better positioning during settlement negotiations? (emphasis supplied)
I'm sure many clients would prefer that a black box provide the answer to their conflict, rather than years of expensive litigation. But the rub is in that "merely." To suggest that inputting the "data" relevant to a conflict is a trivial task is like suggesting that all a baker needs to do to produce a perfect cake is merely to combine all the ingredients into the pan, and then leave it to the oven to produce the correct result. The oven's part in that process is important, of course, but it is purely mechanical. The part that needs human creativity and ingenuity lies in figuring out the right proportions and the right ingredients, before the baker even puts the cake in the oven. That part resembles the story that the client tells the lawyer at the beginning of the case, which the lawyer then has to form into the initial pleadings. The dispute resolution process is even more complicated than that, because it requires further rounds of re-analysis and re-processing as parties to a dispute and their attorneys learn more facts, and are compelled to take the other side's view of the dispute into account in trying to predict how the legal system might resolve it.

How are computers supposed to tell people what matters to them the most, or what is fair, or what is really bothering them about the other side's conduct, or how another group of human beings (the jury) is likely to evaluate their claims? Answer: they can't. Perhaps someday we will invent a computer that can do all that, but until then, human beings will still play the most important roles in the dispute resolution process. That is the role that lawyers and neutrals have always played and will continue to play: helping clients sort out what is important and what is relevant; helping them present their story in an effective way; and trying to predict how the legal system is likely to evaluate their claims. So it's fine with me if computers can be programmed to manage the game-playing aspects of negotiation. Because the more interesting parts, the parts that require human help, are the parts where people "merely" input all of the "data" that is important to them. And where more humans are needed to figure out what all that "data" might mean.


(photo: KAIST)

1 comment:

James F. Ring said...

Thanks for the thoughtful post and the link to our website, http://www.fairoutcomes.com. I agree that – at least with respect to the game-theoretic bargaining systems that can be accessed via our website – lawyers play and will always continue to play a critical role in helping a client understand what data they should enter into our various systems. Lawyers are trained in law school, and learn through real-world practice, to look at a case from both sides and attempt to identify what would constitute a fair and reasonable outcome from the perspective of their clients given the particular facts and circumstances of the case. In the vast majority of cases opposing counsel arrive at similar assessments of what would constitute a reasonable settlement, as is evidenced by the fact that the vast majority of cases settle. What the systems offered on our website serve to eliminate is a role that many lawyers don’t like to play but often feel compelled to play, a role that not only gives rise to a very negative public perception of lawyers, but that also gives rise to a fair amount of self-loathing and depression amongst legal professionals. More specifically, our systems eliminate any incentive or excuse for a lawyer (or a party) on either side of a case to try to bluff or to posture about what it would consider to be a fair and reasonable outcome. There is no question that, in so doing, our systems wipe out one of the central roles that have historically been played by lawyers, a role that causes litigation to be dragged out for years and at great expense. But the elimination of that role through the use of our systems is a development that is being welcomed and embraced, and that should be welcomed and embraced, by everyone, including lawyers themselves.