One of my students sent me this article about Tuomas Sandholm, a Carnegie Mellon computer scientist who is using game theory to facilitate kidney donations. The article reports on the increasing popularity of paired kidney exchanges, “in which one kidney patient's relative or friend will donate an organ to another patient, whose own relative or friend will donate his organ to the first patient.” Such exchanges are limited by the fact that it’s not always easy to find patient pairs with relatives or friends who happen to be compatible matches.
Sandholm’s contribution was, first, to recognize the possibility of multi-party exchanges (e.g., A donates to B’s relative, B donates to C’s relative, and C donates to A’s relative); and second, to devise a computer algorithm that will allow multi-party exchanges with (almost?) any number of parties (e.g., A donates to B’s relative, B donates to C’s relative, C donates to D’s relative, ... , and Z donates to A’s relative).
Now, this is all great, and Sandholm is doing admirable work. But as I read the article, I couldn’t help lamenting that our current system of organ donation has reduced us to the use of barter. What Sandholm has done, quite simply, is to increase the efficiency of barter. Using his algorithm, we could presumably arrange other kinds of transactions as well: a baker gives some bread to a doctor, who provides medical care to a carpenter, who fixes a cabinet for the baker. Fortunately, we don’t need Sandstrom’s method for that, because we already have a venerable institution that coordinates these magnificently complex trades with remarkable efficiency and without the need for central organization. We call it money. Unfortunately, this mechanism has been outlawed for human organs, thereby forcing us to rely on the far more primitive mechanism of barter.