Will computers prevail over humans in chess?

The first game of chess involving a program as one of the players was played in 1952 between Alan Turing’s chess algorithm and his colleague Alick Glennie. Because computers had not yet been invented Alan Turing manually ran his algorithm, taking about 30 minutes per move.

Computer chess programs have come a long way since. In 1997 Deep Blue program created by IBM and run on a dedicated mainframe platform defeated the then world chess champion Gari Kasparov in a match played under “human” FIDE competition rules. Currently the best computer programs run on commercial desktop PC’s routinely beat Grand Masters in short games and are competitive in games played under FIDE competition rules. There are even world chess championships played between computer programs. While FIDE has different rankings for humans and computer programs it is generally recognised that the days of human domination in chess are over. While Deep Blue used brute computing power to defeat Kasparov, the chess algorithms have improved a lot since 1990s and the current crop of programs can compete at FIDE Master level tournaments running even on cellphone CPU’s. In particular, there are two areas where computer programs have an undeniable advantage over human players – openings and endgames.

The openings are essentially a collection of the opening moves which have been played and analysed in the past. Because of the way chess permutations multiply there are millions of different positions possible say 10 moves into a chess game. Chess masters learn many by heart and rely on their intuition to evaluate the ones they have not come across. Computers, with their digital memory, can store, call up and use every single opening ever played as long as it has been catalogued. They can (and do) also analyse various openings looking for statistical trends like: the Black won 60% of games in which they captured the bishop in this position and only 40% in which they did not. Most openings have been analysed this way at least 10 and some openings up to 20 moves into the game. Human brain cannot process so much information and is at a clear disadvantage.

But where computers really come into their own are the endgames. Using their superior computing power chess programs have analysed all possible endgames involving up to 6 pieces (for example: king + bishop + pawn vs king + rook + pawn). Each of these positions has been thoroughly analysed in a sense of finding out the optimal game plan for both sides and the outcome assuming both sides play the best moves. Many endgames involving 7 pieces have also been analysed in a similar manner (assuming no castling). All 5 piece positions (with their game plans and outcomes) can be stored on a single DVD. All 6 piece positions occupy about 1.2 terabytes and will fit on a laptop hard drive. 7 piece positions are too numerous to be stored on mobile memory devices but this will change as the memory density increases. So, essentially, a respectable chess program run on a modern laptop should be able to instantly declare with 6 pieces left  that say the Black will check-mate in no more than 24 moves. This will be done without analysing the position, based solely on the stored library of endgames.

So, essentially, computers start chess games using the in-built library of stored openings. Then, some 10-20 moves into the game (or sooner, if an opponent chooses a little-documented opening), they switch to analysing possible new positions created by their moves. This analysis is generally 10-20 plies (half-moves) deep and takes much more time than simply referencing the opening catalogue stored in the memory. When position reduces to 10-15 pieces the computers start referring to the stored catalogue of the endgame positions. They try to achieve a 6 or 7 piece position which is known to be a win for their colour. When there are only 6-7 pieces left on the board the outcome has already been determined and the game is essentially over.

Where human brain still holds some advantage is in the positional analysis of complex middle games, where intuition plays important role. But, as you can deduce from the above summary, the middle game keep shrinking as more and more openings moves are catalogued and also more end game positions are analysed. The future of the human chess is looking bleak.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: