Matthew Yglesias wonders whether the algorithms that good chess-playing computers employ when they play chess are similar to the way good chess-playing humans play. Chess programs assign specific values to contingencies on the chess board - for instance, each piece will have a value, and these values might be altered by which other pieces are present. Humans might think in terms of values, but everything is more fluid, and positional (strategic) considerations often trump material or tactical advantage.

As I discussed in an earlier post, the cutting edge programs grandmasters have to face are carefully adjusted so that contingencies have the best values possible - and these adjustments are made by grandmasters involved on the programming team. So, maybe it's fair to say these computers think like great players, but only because they've been programmed with the whole body of chess knowledge and intuition - arrived at not through calculation, but directly from the head of a grandmaster, probably through a laborious trial and error process.