Wednesday, July 19, 2006

"Secrets of the Expert Mind"

This article by Phillip Ross is in the August 2006 Scientific American, and will be available free online next month.

Ross summarizes research which suggests that chess experts have huge stores of knowledge of the game, gained by decades of organized and diligent study of the game (rather than by playing a lot). He suggests that experts play better not because they analyze more, but rather primarily because they use this accumulated knowledge to recognize patterns of pieces on the board, and to recognize the most promising moves from such patterns. They can then restrict their analysis to the more promising moves. Non-experts, even when they analyze in depth, waste effort in analyzing scenarios with low likelihood of panning out.

He suggests that modern chess masters are actually better than those of 100 years ago because the modern masters have more accumulated knowledge as a result of the raising of the bar due to more competition, to better and better organized learning materials and tools.

It is interesting that there are very reliable ways of rating and ranking chess players, but the better rated player does not always beat the lower rated player. What is reliable is the portion of games that the better player will win against the lesser player, given the differences between their ratings.

Would that we had a similar rating scheme for our leaders, be they in the political, economic or social sphere. It would be nice to know we have a clue on who to believe, since these guys make so many mistakes.

I am also concerned that we so often ignore the research on decision making. As a result of this research, we should know that there are some ways to get decisions out of groups that work better than others. Unfortunately, those who organize advisory committees seem seldom to use this knowledge, or indeed to be aware of its existance.

So too, as a result of research on decision making and prediction, we should know that there are ways to train people to make better probability estimates and to avoid common errors of judgement. Unfortunately, I don't know of many places where such training is actually applied to improve decision making.

No comments: