[an error occurred while processing this directive]
BBC News
watch One-Minute World News
Last Updated: Tuesday, 18 November, 2003, 17:48 GMT
Humanity counts in chess battle

By Robert Plummer
BBC News Online

World chess champion Garry Kasparov is preparing for the fourth and final match in his latest attempt to beat a computer at the game - but he has a long track record of playing non-human adversaries.

Garry Kasparov
Garry Kasparov has striven to avenge his defeat by Deeper Blue
"Machines are stupid by nature": that was Garry Kasparov's disparaging verdict on his first encounter with a computerised opponent.

It was 1996 and Mr Kasparov, the world's number one human exponent of the game, had just beaten the IBM Deep Blue computer - although he admitted it was a tough match.

Even so, he had spoken too soon, because the following year, he lost in a return bout with the upgraded and renamed Deeper Blue.

Some grand claims have been made for that contest, with some chess buffs hailing it as a moment to rank with the 1969 Moon landing as a milestone in human history.

For the first time, a machine had beaten a person in a mental struggle that Mr Kasparov himself described as "species-defining".

Fighting back

But although the uninitiated might have regarded it as inevitable that a computer would one day trounce a chess grandmaster, Mr Kasparov has performed well in more recent matches against non-human opponents.

Chess-playing computers have tended to be better at tactics than strategy
In January and February this year, he took on another computer program, Deep Junior, which had not been beaten by a human in two years.

In the six-match series, he won one, the computer won one, and the other four ended in a draw.

Mr Kasparov has also fared well in his current battle with the X3D Fritz program - securing a draw in game one and a win in game three, although he lost the second one.

So how exactly does a computer play chess? And what can a top player do to outsmart the machine?

Complexity

Artificial intelligence experts and computer programmers agree that chess-playing computers have tended to be better at tactics than strategy.

The classic way to design a chess program relies on what experts call "brute force" - that is, using massive processing power to work out as many potential moves as possible and analyse their consequences.

The problem is that even for a computer, chess is a fiendishly complex game.

Garry Kasparov and Deep Junior
Mr Kasparov's battle with Deep Junior ended in a draw
From any average position on a chess board, there are 38 possible moves.

To analyse all the implications of just six moves by each player would require considering nine billion billion possible positions - more than any current computer can handle.

The programmers of Deep Blue worked around this by using techniques that allowed the computer to eliminate obviously bad moves.

This allowed the program to get by with the ability to consider just 100 million positions a second (doubled on the upgraded Deeper Blue) - plus a huge database of mistakes to avoid.

Unfortunately, this method did not encourage aggressive chess play.

"Brute force" programmes found it hard to cope with human strategies that involved consciously sacrificing chess pieces in order to gain an advantage later on.

Achilles heel

Deep Junior and Mr Kasparov's latest opponent, X3D Fritz, represent a change of approach.

They pack less processing power, but use "smart" software to pick out the moves that have most potential.

The difference is noticeable. In its fifth match with Mr Kasparov, Deep Junior surprised spectators by sacrificing a bishop in the early stages.

Chess pieces
Human chess players can still think further ahead
But computers' weakness at strategy is still an Achilles heel that the chess-playing human can exploit.

Mr Kasparov outwitted X3D Fritz on Sunday by building a wall of pawns early on.

In this way, he played a "closed" game that restricted the number of possible moves and made strategic thinking all the more important.

Analysts point out that machines still have sheer relentlessness on their side.

They never have an off day and they do not suffer from lapses of concentration.

Human beings may be fallible, but Mr Kasparov's track record shows that at their best, they can still defeat computers by planning ahead.




SEE ALSO:
Kasparov even in virtual contest
17 Nov 03  |  Technology
Kasparov fights computer to draw
12 Nov 03  |  Technology
German chatty bot is 'most human'
20 Oct 03  |  Technology
Kasparov draws against computer
08 Feb 03  |  Technology
Kasparov clears key second match
29 Jan 03  |  Technology


RELATED INTERNET LINKS:
The BBC is not responsible for the content of external internet sites


PRODUCTS AND SERVICES

News Front Page | Africa | Americas | Asia-Pacific | Europe | Middle East | South Asia
UK | Business | Entertainment | Science/Nature | Technology | Health
Have Your Say | In Pictures | Week at a Glance | Country Profiles | In Depth | Programmes
Americas Africa Europe Middle East South Asia Asia Pacific