"My opponent left a glass of whisky 'en prise' and I took it 'en passant". - Henry Blackburne | SINCE 2007

Thursday, May 20, 2010

Can Chess Be Solved

by rjlipton

Computers play great chess—can they play perfect chess?

Ken Thompson is one of the co-inventors of UNIX, perhaps one of the greatest programmers who ever lived, and won the 1983 Turing Award for his important work.
Today I plan on having a guest author, Ken Regan, discuss Thompson’s work on chess playing programs. It is Ken on Ken, or rather Ken on : when Ken (Regan), played at the Westfield (NJ) chess club as a teen in matches against distant clubs, he knew Ken (Thompson) not by his famous programming handle but as the “club techie” who relayed his moves by telex.

I once visited Bell Labs and spent some time talking to Ken T. on his chess program. I asked why he was doing so well—his program was winning a lot. He said his program had, probably, fewer bugs than the competition. In those early days chess programs were often buggy, and a program with a bug is likely to be a program with a chess weakness. The reason, Ken T. explained the programs were so hard to get right was simple: All chess programs looked at a huge set of positions and then generated one move. As long as the move was legal the programmer had no idea the program had made an error—unless the move led to an unexpected loss.
The rest of this is the work of Ken R., who by the way is an international master with a rating of over 2400. With, of course, lots of help from Subruk. READ THE FULL ARTICLE

No comments: