November 7, 2014

No set-up, just diving right in. Thoughts on the underlying similarity between New York City's affordable housing policy and pairwise kidney donation programs. You were warned!

The allocation of subsidized housing in New York City operates according a random lottery for spots in a queue which is then re-ranked according to preferences for certain groups, e.g. the disabled, households with children, or municipal employees. It's an instance of an algorithm that implements a program that implements a policy that implements an ethical determination about how to distribute common resources fairly. Things begin to get complicated when you pit a (1) disabled (2) veteran, say, against a (1) disabled (2) municipal employee. The rules that resolve these comparisons will emerge from the technical details of the system, but they may or may not jibe with our intuitions about who should get what.

Let's keep that in mind and turn to now to pairwise kidney donation. Briefly, the way it works is: Patient A needs a kidney transplant, and finds a willing (living) donor, call her Donor A, but, due to blood-type or other factors, no transplant is possible. At the same time, Patient B, who also needs a kidney transplant, finds a different willing donor, Donor B, but again, no transplant is possible. But Donor B matches to Patient A and Donor A matches to Patient B. We make the match and achieve utility. Very simple.

Higher order exchanges are possible too. Donor A matches Patient C, Donor B matches Patient A, and Donor C matches Patient B, say. We can extrapolate from this some total latent utility among the set of all patients and donors. It's sobering and quite sad to think about, given the number of patients who, in the real world, die each day for want of a kidney transplant. (From 2008 to 2010 in the United States, the number was about 4,700.) Kidney paired donation programs exist in order to orchestrate these matches, which become--logistically, analytically, and ethically--quite complex. Two donation pairs means scheduling four surgeries simultaneously, three means six, etc. and don't forget the fact that something like 20% of these procedures won't go as planned, for medical or other reasons. In 2012, Alvin Roth won a Nobel Prize in economics for his work on optimizing paired donation outcomes.

Anyhow, as in the case of allocating affordable housing in New York City, we can see the chain: an algorithm implements a program that implements a policy that implements an ethical determination about how to distribute common resources fairly. I love that!

October 24, 2014

This is going a little slowly, but that's okay. Still lots of very basic stuff to figure out about the terminal. The funny thing is, I recognize Pine from 1994, when I got my first email account. Brings back unexpectedly vivid memories of making the daily trip to the basement of Connecticut Hall so I could check email. I remember a command that transformed whatever you wrote into "Swedish Chef" language. I think the command was "chef."

October 21, 2014

Here's my old place. I don't miss it really.