Polar Bears and Operations Research
The Sep-Oct 2008 issue of Interfaces contains an article concerning the listing of polar bears as an endangered species, along with three commentaries. The recommendations in favor of the listing are based on polar bear population forecasts that depend in part on forecasts of summer sea ice in locations inhabited by the bears. The Interfaces article, written by Armstrong, Green, and Soon, calls into question the forecasting used in two of the studies that informed the decision.
There were a number of things I found interesting about the article and each of the commentaries. First, the basis for evaluation in the article is a set of forecasting principles from a project started by the lead author. The studies are rated in terms of percentages of the principles applied and contravened (i.e. acted counter to). I like the idea of principles for forecasting, not to mention for other realms, such as modeling and data mining. They would seem to offer a good check against methodological blind spots and preconceived notions emanating from one’s belief system.
Along those lines, one of the principles contravened in the studies according to Armstrong et al. is “Make sure forecasts are independent of politics”. So something else I found interesting reading the paper and commentaries is that I could not help thinking that the writers’ beliefs were close at hand, despite the cool logic of the arguments. The commentary by Murphy is most open about this:
… we all operate with belief systems, and those belief systems lead us to self-select where we stand on issues … Our beliefs lead us to the point where we know our side is right.
Reminds me of some of the issues that came up in the “Doing Good with Good OR” Energy & Environment session at INFORMS 08. (See this post (3rd paragraph) for more on that.) In any case, if you are interested in polar bears, evaluation of forecasting or the place where science and belief systems might collide, this issue is worth checking out.