The World Computer Championship is also taking place here at the Oval in Turin. Twenty of the world’s top programs are here. It is fascinating to watch the games. Except for the presence of the two machines on the table, it might pass for a casual game between humans. There is a board and clock and the operators make the moves and punch the clocks manually. In between moves, they engage in analysis and good-natured banter.
Earlier today, I sat down for a few minutes with International Master David Levy. Those of my generation may remember his famous bet that no computer could beat him within ten years, which he won in 1978. For a long time, he was the champion of the human race against the machine until he lost to Deep Thought in 1989.
I began by asking him why the human operators were even necessary. Surely, there you could interface the computers and then sit back and watch them go at each other. “We want to preserve a bit of humanity and it’s nice for the programmers to talk face to face.”
I couldn’t help but notice the absence of the best-known program Fritz. “Fritz is a very well-publicized program and it’s not very clear how it would do. It’s not very likely that it would finish at the top, around 6th to 8th is more probable.” It follows that it wouldn’t make good business sense to enter it. (I saw Fred Friedel earlier today and he mentioned that he’s reading my blog; Fred, if you are reading this, feel free to dispute this point.)
How much have computers improved since the days of the Kramnik-Deep Fritz and Kasparov-Deep Junior (a mere three to three and a half years ago)? “They keep getting better every year but it’s hard to say how much. Perhaps 30 ELO points a year. We probably have two 2800 programs and five 2700+ programs here.”
I asked him about open source chess programs and he mentioned two such: Crafty and Fruit. There was a dispute that resulted in a program (LION++ 1.5) being disqualified because it was a modification of Fruit. This would have been acceptable if this had been revealed, credit given, and permission received before entering.
I finished with a discussion of how much of the results are due to hardware and how much due to software. “It’s nearly all software.” I stated the conventional view that most of the advances were due to faster processers and more memory and that while among computers, it might be software that made the difference, in man vs. machine play, it was hardware. He found that much too broad a generalization and said advances had been made in both and it was hard to separate them. So I gave them the hypothetical of today’s software on a Commodore 64. “Of course, it would be much weaker. Perhaps no better than 2000.”
Post a Comment