Date Posted: 8/26/2016

Technology Review and Pacific Standard Magazine recently covered a KDD 2016 paper by Ashton Anderson (MSR), Jon Kleinberg, and Sendhil Mullainathan (Harvard; also Cornell CS '93) that used a dataset of 200 million chess games as a model system for analyzing human decisions and errors. A key feature of the domain is that for chess endgames with small numbers of pieces, the winner under an assumption of optimal play can be determined by lookup in computationally-compiled tables, but even the best human players in such settings have a non-trivial "blunder" probability, making moves that worsen their minimax value.

From the Technology Review article: "How do [skill, time, and difficulty] influence the quality of the decision being made? ... These are hard questions to answer, given the difficulty of setting up a controlled experiment to test them. Indeed, nobody has found a satisfactory way of studying the problem. Until now. [The authors] unveil the first large-scale study of decision making under controlled conditions. For the first time, these guys have been able to study how the quality of decision making changes with the time available, the skill of the decision maker, and the difficulty of the decision at hand."