Author: Russell Reagan
Date: 09:54:41 04/02/02
Go up one level in this thread
On April 02, 2002 at 05:08:08, Martin Andersen wrote: >"I was totally unprepared to play against a modern playing program. My >experience from the mid-1990s turns out to have been completely irrelevant >today" > >This is a surprise to me. He didn't know that there has been a big increase in >playing strength for chess programs in the last 7-8 years, of course due to >faster hardware but also better software. >Then I would say his score of 3-5 without preparation, is excellent. > >http://www.kasparovchess.com/serve/templates/folders/show.asp?p_docID=20790&p_docLang=EN > > >Martin. the ramble begins... I'm usually torn about what to think about human vs computer matches. It seems like the humans are losing more and more. I think it's more of a lack of attacking the problem on the part of the humans. Just like Gulko said, you have to play a completely different style against a computer, and he himself said that he got better as the match went on because he began to learn how to play better against the computers. This situation reminds me of the evolution of cryptography. Long long ago when we had simplistic encryption algorithms, they seemed unbreakable. After some time, they figured out that there were weaknesses in the algorithm, and were then able to crack the encryption and decrypt messages using that encryption. I think of the German's use of encryption and the enigma machine and how it seemed unbreakable. But there were weaknesses in it as well, and they were exploited. Then we move on to public key algorithms, and at first they seem unbreakable..."we'll never break these" we think. But just like every single time before, there were weaknesses, and people have exploited them, making the "unbreakable" algorithms less and less secure. I think of both problems of attacking cryptographic algorithms and playing chess against a computer to be like scaling a flat wall. If there is no weakness in an encryption algorithm, then the wall is perfectly flat with no place to get a grip anywhere, and you can't climb it. You have an unbreakable encryption algorithm. If there are weaknesses in the algorithm, then it's like the wall has small holes in it, cracks, etc. Places you can use to begin climbing the seemingly flat wall. Even with a small weakness, you can still exploit it. Imagine a tiny pin sized hole in the otherwise perfectly flat wall. We could develop a "tool" to latch on to the tiny hole and climb the wall. I think computer chess is the same way, just like GM Gulko said. The computers have weaknesses just like humans. Until we have complete 32-man tablebases, there will always be a weakness somewhere, no matter how small. I think computers are totally beatable by humans. I think humans are, however, in the "OH MY GOD THAT IS UNBREAKABLE!" attitude that we've had when faced with stronger and stronger encryption algorithms. The other factor is that our best minds (GMs) are not attacking the problem of computer chess as viciously as cryptologists attack encryption algorithms. I think that if they did, we would find that computers are still basically dumb machines. Anyone who has written a chess program knows what a voyage of discovery writing good evaluation functions can be. I think about all of the thousands of ideas I've had for evaluation functions, little things I can add here and there, totally "out there" approaches that seem like they might work. I realize how overwhelming it is with just the ideas I have had, and then I realize how little area I have covered in the evaluation universe. There are probably astronomical ways of approaching evaluating a chess position from a computer's point of view, and my thousands of ideas (that I have no method of testing in my lifetime) are a grain of sand on the beach. Compter chess has so much room for improvement, but right now, computers are still full of weaknesses, even if they are so tiny that 99.999999999% of the world can't see them. I think that's why 99.999999999% of the world will get beat by computers at chess. They can't see the weaknesses, and couldn't exploit them if they could. Top GM's could do it though, with sufficient analysis of the problem of how computers play chess and their weaknesses. Having said all of that, I'm sure it will never happen. What GM really cares about this enough to put his life's work into finding weaknesses in computer chess programs? Who besides him and a handful of other players could even understand the minute weaknesses he discovered? The handful who could understand it probably wouldn't care anyway. And what would we say if he discovered a great many weaknesses in our programs? We'd probably disregard him as "just one person" and go on believing computers are far superior to all humans at chess, which I don't personally think is true, but it's not like anyone can prove it either way, so this, just like most things in life, don't really matter, so let's get back to doing something fun shall we? :) ...the ramble ends. Russell
This page took 0.01 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.