Slashdot Mirror
Chess: Man vs. Machine Debate Continues
Frederic Friedel sent in an interesting submission. It's an interview with the current world's chess champion, Vladimir Kramnik, in which they talk about the upcoming year in chess competitions, but also get into [Deep Blue] and where computer chess playing is versus several years ago, with a comparison between Deep Blue and Fritz. If you want more info, check out Chessbase for additional news.
It's been well known since, well, before I was born, that a computer could easily trounce a human in any game involving only tactics. For example, many fourth graders in this country have programmed a BASIC script to create a tic-tac-toe player that will never lose.
Therefore, it's not particularly novel that computers can beat people at tactical games. The only thing interesting that I see arising from these onging "human versus machine" chess matches is the proposition that strategy can be broken down into millions of tiny tactical evaluations.
This begs the question: is the strategy that a human chess player would use also based on these millions of tiny tactical evaluations, only so subtle that he's not aware they're going on in the vast electrochemistry of his brain? Or is strategy discernable from tactics in a human mind, but simply a subset thereof in a computer?
The sole interesting conclusion I draw is that if it can be proven that strategy is something different to man and machine, then a hybrid approach might allow us to solve problems in ways we've never dreamed of. Whether that hybrid approach would involve implanting computers in our minds, making computers that can function like minds, or simply working really well with computers, I leave to you.
"Beware he who would deny you access to information, for in his heart he deems himself your master."
Grandmasters can in fact tell whether their oponent is a computer, sometimes even after playing just a single game, and certainly by the end of a match. In fact, I believe Kasparov lost to Deep Blue precisely because he counted on the computeresque behavior of his opponent when designing his strategy. If you read the article, you will learn that Kramnik can tell computer programs apart by their style, and that he thinks Fritz is becoming more human-like in its behavior, from which I infer that he can still identify its style as computeresque on some level.
So, the test you propose has already been carried out, and the machines "failed". This may have more to do with the fact that the people who write chess playing programs are more concerned with the programs' ability to win than they are with the programs' ability to emulate the playing style of humans. If humans could calculate better [Note: "calculate" has a precise technical meaning in chess] or chess playing computer programs were slower and considerably more stateful, their respective styles might be much more similar and your test, therefore, be met.
My own belief is that the ability to play chess well, let alone the ability to play chess in the style of a particular grandmaster, is not an accurate or even adequate measure of intelligence, so I will not be particularly hurt when the day comes on which computers at last surpass our chess playing skills, just as they have surpassed our (numerical) computational skills.
Even I can tell.
by the style of play, humans usually have clear strategies, computers dont, they usually just tactically try to beat you, using lots of tricks and traps, they dont have REAL plans so its easy to know its a computer if the computers every move is generic.
If you use Linux, please help development of Autopac
Let me interpret for the original poster:
If you have a tree who's root is the current board position and the leaves are all "mate" or "draw" positions (where the mates give you the game) then it certainly WOULD be possible to force a draw or win for yourself in every game. Starting from some arbitrary board position, of course you couldn't, because there are then board positions which will not/can not happen. But FROM THE BEGINNING, a program could enumerate every possible move, and eliminate those which end in its loss.
Now, that said, it's not true that from a given board position, there is any move that will guarantee a loss. In some positions this is true, but not for any given position (take the first move of the game, for example. Presumably you can win no matter what first move you make.) The early game would likely need to be a series of moves to bring you to a board position in which the rest of the game is deterministic based on the tree. This is because eliminating all initial moves that can result in a loss will eliminate all initial moves.
Getting to that deterministic position is not guaranteed, either. If it was, the entire match would be deterministic based on the computer player's moves. So there would still be strategy involved. But the computer could still look ahead to prevent moves that result in mate (for example, the 4 move mate that's such a common ploy against new players). And in doing so, it could look for ways to get to a board position that's in its lookup table.