18th February 1996

Garry Kasparov lost his first match to the Deep Blue computer, which can consider more than 500m forward chess positions per second. Kasparov then won the second game, but won't beat any Deep Blue mark II. From the average position, you have some 30 options for your next move, most of them dotty. After one white and one black move, there are thus 900 possible patterns of the board. By white's fourth-next move, there are more than a billion. The computer can envisage all of them. A Kasparov can consider at best 500, a huge shortfall.

If, as a computer, you try to peer further forward, watch out. During the next 70 moves, there will be more possible board patterns than atoms in the universe. If all humans since Adam had done nothing but play chess, with no pauses to eat and few to sleep, we might not yet have played through all possible games. Kasparov won last week's second game by offering the computer some short-term temptations, which should give him longer-term advantages beyond the computer's timespan. Since later versions of Deep Blue can effortlessly increase that timespan, our touchiest human geniuses are eventually bound to be overwhelmed in logical deduction by slivers of silicon.

That will cause humans' greatest and silliest debate. Kasparov sounded to me petulantly protectionist when he prayed he would beat the computer so as to help continue "human control over such fields as music, art, literature". I don't know how soon a computer, searching each second through 500m alternative next lines in a musical score, will compose sweeter sounds than Mozart, or, searching each second through 500m alternative brush strokes, outdazzle Leonardo da Vinci. I will love the greater art when they do, although cross friends who are artists call this anti-human. But computers will much earlier cause revolutions in less lofty information industries, such as newspapers, genetics, healthcare and education.

Tomorrow's computers will set pupils questions, then deduce from their answers what their learning patterns are. They will follow changing experience's rules on what the best ways and speeds of imparting new information and searching questions are. Some trade-unionised teachers may be cross.

When I wrote the biography of Princeton mathematician Johnny von Neumann (1903-57), whom I regard as the key inventor of the modern computer, his private papers showed me his dreams for his machines. He never imagined them grasping 500m concepts a second, thinking maybe 10,000 more plausible, but hoped even this should improve simple information industries such as education and research. His forecasts for the private sector (eg, on laptop computers) have been overfulfilled, but ­ fresh from Los Alamos ­ he assumed even the public sector would by the 1990s provide limitless cheap energy through nuclear fusion, and have managed to control the weather.

If it didn't, this would be because the social sciences (economics, sociology, politics) lagged so far behind commercial exploitation of post-quantum physics. Princeton's social scientists said their fields were weakened by the human element, by uncertainty about how different people would react to different incentives. With his research into the computer and the brain, Johnny rudely told them they were "utterly mistaken". He could use his computers in physics because physicists had proved this follows from this.

Johnny hoped his computers might help in free-market economics. He assumed we would by now have databases to show the disasters when farmers' prices or minimum wages were set too high; when politicians decree that exchange rates try to stay fixed through each stage of wildly swaying trade cycles; and the wastes when tomorrow's high-tech industries such as health and education remain ruled by Brezhnevian state control instead of by dynamic changes in the markets.

In your guts, you know that's nuts. But Johnny also feared his computers would be used for macroeconomic models (like the Bank of England forecasting two years ahead), before there was any scientific evidence on what would happen next week. A year ago the Bank's macroeconomic model said interest rates should go up. Last week it agreed they had been right to come down. Don't giggle in church.

As Deep Blue resigned last Monday, the whole audience rose to bid Kasparov "bravo". I am a yob supporter of the other side. As we wait for sweeter music than Mozart's and saner economics than Eddie George's, I chant: "Grow, Deep Blue, grow".