Man versus Machine: The last frontier faces final invasion

The Machine is IBM’s Deeper Blue, a R/6000 SP supercomputer, weighing 1.4 tonne. Inside it are 32 nodes with 256 chips working in parallel, giving it an unprecedented computational power and speed. It can evaluate up to 200 to 300 million chess positions a second — or 40 billion combinations in an average three-minute move.

The Man is Garry Kasparov, undeniably the greatest player to play the game till date. A Grandmaster with the best ELO rating ever, he is acknowledged by those who know the game as the best representative mankind could offer: cool, calculating, combative and fiercely proud.

The theory that man is superior to machines is based on the way we “think”, not the modest speed at which we calculate. Machines can certainly compute faster, but they lack man’s greatest attributes: intuition and experience. Whereas Deeper Blue will whir through millions of variations for every chess position, Kasparov’s mind will dismiss many as nonsensical, simply by virtue of the experience and intuition.

Says Murray Campbell, a research scientist who has worked with chess machines: “Deep Blue can see everything in the clear light of day. But when it’s dark, in the black beyond, man’s intuition comes into play. He can guess better… he can smell better and avoid traps.”

But Deeper Blue’s programmers say they have offset this natural advantage for man by improving some of the pattern recognition features in the machine, besides souping up its computation speed. For example, in the contest between Deep Blue, the precursor of Deeper Blue, and Kasparov last year, the computer lost one game badly by letting it’s bishop get trapped at the edge of the board with little power and zero mobility. A good chess player instantly recognises this as a “bad bishop” — in that even though it remains on the board, its lack of mobility makes it less powerful.

Deeper Blue programmers say the new machine has the ability to weigh the mobility and space factors — known in chess parlance as tempo — and recognise good and bad pieces. “Deep Blue was a baby, it did not know too much. This one is going to crush Kasparov like a bug,” says Joel Benjamin, a US Grandmaster who has helped programme Deeper Blue along with Feng-Hsuing Hsu, who made the machines.

But many chess analysts think Kasparov is playing better than ever — almost close to perfection. His ELO score of 2,820 is the highest in chess history. This week, he is still in Moscow preparing for the tie under the formidable eye and comforting care of his mother Klara Kasparova — something Deeper Blue lacks, some of us would think.

But Deeper Blue has no place for emotions and that in fact is its greatest advantage, many observers feel. It’s precursor Deep Blue demonstrated this to great effect during the first game of last year’s contest. In that game, Kasparov sacrificed a pawn and launched a fierce attack against the program’s king. In a human situation, a rival would have felt the pressure and tension of such an attack. But Deep Blue was inhuman and did not know the meaning of discomfort. “The gambit was a complete disaster because the computer really does not care. If the threats are not real, it can see, ” a wounded Kasparov said later.