Deep Blue was the first computer to beat a grand master in a chess tournament. There’s controversy over some human intervention between games, but what’s interesting to me is what they had to do to create Deep Blue.
The difficulty in building a program to compete in world-class chess was that the raw calculations are too big. Starting each game with 32 pieces total, many of which have multiple options each turn, the number of possible sequences that might follow from any one move in a situation quickly becomes astronomical.
Grand masters don’t have to analyze every move. For instance, they might not consciously consider that pawn sitting in the corner. Instead, they know from experience which positions to pay attention to.
The breakthrough for the computer programmers was when they discovered ways to program this kind of processing into the computer, this brain-like function, instead of having the computer try to crunch the numbers for every possible sequence.
In other words, to make the computer smarter, they had to teach it how to ignore some possibilities in order to focus on the ones that matter.
