Slashdot Mirror
10 Years After Big Blue Beat Garry Kasparov
Jamie found another MIT Technology review story, this time about Chess, Supercomputing, Garry Kasparov, and trying to make sense of just what exactly it all meant when a computer finally beat a grand master. An interesting piece that touches on what it means to play chess, the difference between humanity and machinery and how super computers don't care when they are losing. Worth your time.
in defense of kasparov, big blue also had help from kasparov's previous competitors to look over and recommend moves for big blue to move, so it wasn't really the machine alone that beat kasparov, he was defeated by a supercomputer and a few of his previous competitors.
But what about 'Go'? 'Go' is much harder for computers to play. Let's all talk about 'Go'.
IBM's next chess supercomputer, Big Wuss, is rumoured to care when it is losing.
which is totally what she said
People seem to be very sensitive about computers doing things they think only humans should be able to do. They dismiss defeating a chess grand master or the Turing Test as toy problems.
I did an AI degree in the mid 90s and one of the things we covered was the definition of intelligence. After running through a few unsatisfactory definitions, my conclusion was that people used intelligence to mean whatever could be done better by a human being than anything else...
Actually, my favourite definition of intelligence, partly because of its succinctness, is "productive laziness".
Peter
After it was discovered that IBM was tinkering using chess experts (that is, humans) to tinker with its software between matches, they're personae non gratae in the chess world now.
Back in the early 1990s, I used to play in chess tournaments. I wasn't very good though and I didn't play at a high level, but I did play in official tournaments that the USCF (United States Chess Federation) sanctioned. My goal at the time was to try to make grand master. I gave up because of 2 reasons. The first was that I wasn't very good. I had serious problems in the middle game. My opening play and end game play were sound, but inevitably I would get beat in the middle game through carelessness. The second reason I gave up was because I realized that computers were ruining chess. Keep in mind that I am talking 1990-1993 here (I stopped playing in tournaments in 1993). In the old days, if you learned a chess opening, the moves might go 7 moves deep or so in most openings where the moves for the white and black pieces were known and any deviations from these set moves got you "out of book" as they say. If you deviated on, say, move 4 in a 7 move sequence, the odds were that your move was bad because if it was so good, it would have been known and used by other players and then be part of the book. At this time being "in book" was already starting to change because of computer analysis. Then you could go 10 moves or more in many openings and still be "in book". The amount of time and memory required to memorize these much deeper opening sequences was overwhelming. One day I realized that it just wasn't worth it and I'd rather devote my time and brain power to other things that I actually had some talent for, like learning other languages.
Chess is said to be "solvable". My understanding is that it can be proven mathematically that chess has a finite series of moves. If this is correct, then at some point computers will be powerful enough to be able win every game because they'll be able to analyze every possible opening all the way to the end and only pick the moves that will win. No human will ever be able to duplicate this feat. So it is inevitable that computers will eventually be unbeatable. I think just a few weeks ago Slashdot had an article that a computer program has been designed that is now at the point where it cannot lose at checkers - ever. Checkers is quite a bit less complex than chess and it has only now been solved. Whether it takes 10, 20, 50 or more years to solve chess, the day will come when computers simply cannot be beaten at chess under the current rules.
Should we care? Well, maybe not. Computers are better than humans at a lot of things, like mathematical calculations, so it's inevitable that they will be better than humans at chess. The downside is that once all chess games are solvable, it will ruin chess at the professional level. It will make it almost impossible for any game to be postponed until the next day because once there is a postponement, a player could, in theory, simply use a PC to analyze his game and find a sequence of moves where he cannot lose if he plays them correctly. At that point, there's no more human element in the game - it's simply a matter who can more accurately remember computer analysis. Computers ruined chess for me in the early 1990s. Can you imagine how much worse things are now? And how much worse they will be when the day comes that everybody can use a PC to analyze his game and find a way to never lose? At that point, I suspect that either chess will change to Fischer Random Chess as mentioned in the article or people who would have played chess will simply move on and play the game of go instead. Go is beyond the ability of current computers to solve and even the best computer programs can't beat strong human players.
As far as I know no explanation for the strange uncharacteristic move was given by IBM, and deep blue didn't make any other startlingly non computer like moves for the rest of the tournament. It's a rather interesting puzzle.
Craft Beer Programming T-shirts
Sure. But Kasparov didn't have access to Deep Blue's "previous games", or indeed any information about the system at all. They kept him in the dark. IBM also insisted that there be no game breaks -- not an issue for Deep Blue of course -- but a very *big* deal for professional chess players. But most importantly, IBM's team of chess masters and coders modified the system between chess games after analyzing Kasparov's strategy the previous game. That is, he wasn't playing Deep Blue: he was playing Deep Blue being adapted in semi-real-time by a bunch of human experts. And crucially, IBM hid this fact, knowing that it'd be (rightly) considered highly suspect.
Mine loves to play:
- When Is The Most Irritating Time To Crash
It also enjoys
- Fatal Exception Blocking The Save Function
While it was impressive to have a computer win against the "chess master" it accomplished this task by looking ahead as many board configurations as possible....
There in is why many who play chess don't take this match seriously.
Some flaws, first to play a grand master you need to qualify and play others. Then you enter a tournament and build up to play. This leave a trail of your style of play, your weaknesses and your strengths. A true match, your opponent would study your last games before he moved the first piece!
In this case, it was completely bypassed, placing the single player against machine at a disadvantage. Should it have been a real tournament play, I suspect the machine would have done well but lost. And there was one game I watched where he lost and he was either having a bad day or tossed it.
I used to study AI for a while, and i just wanted to point out how unfair this line of reasoning is. Stuff like this ("Very nice, but it isn't *real* AI, because...") always comes up every time there's some AI break-though being discussed.
1. It's almost trivial to make a program 'learn' from mistakes. Just store some negative value for that specific decision-point. Depends on your definition of 'learning', of course. But the principle is the same in humans and AI
2. Kasparov also adjusted his style (i believe there are certain playing-styles that are beneficial when playing against an AI), and i bet he had coaches and consultants
3. So what?
4. See above.
My point is that every time some AI people actually manage to out-do humans, humans tend to re-define what intelligence is. I bet if you'd tell somebody 100 years ago that a machine would be the world's best chess player, that alone would have been enough to consider the machine 'intelligent of sorts'. But as soon as we know how it works, it somehow looses the right to be called 'intelligent' (mechanical turk). I think this is because it seems to hurt humans that AI shows them that whatever gives us the right to call ourselves 'intelligent' is nothing more than the result of zillions of relatively simple interactions of little protein-machines.
IIRC (its been a while) the best way to determine what language a given text is written in, is amazingly 'stupid': just compare the ratio of how many times the different characters appear. The result is still amazing and should be considered 'kind of intelligent'.
So, just give AI some kudos, accept that there's a lot left to be done, and that the heuristics dint really matter, as long as the result is cool. (and please dont give me none of that Chinese Room Argument crap)
... is Computer vs. Computer
They are fearless, uncompromising, untiring. The games are far more interesting than human efforts. Check out some Rybka vs. ZapZanzibar matches (the number 1 program vs. the number 2 program). Incredible play.
Dennett's article suggest to me that he himself does not know a huge amount about chess. For instance, he writes, "The best computer chess is well nigh indistinguishable from the best human chess..."
Sometimes, but not always. As is well known, computers excel in "random" positions where tactics predominate. That's because they have no concept of "general principles" or strategic goals as human chessplayers think of them - instead, they just calculate furiously and find the move that, against what look like the best replies by the opponent, gives the best "worst-case" outcome after a given search depth. They are programmed to follow the game theory "minimax" strategy, which essentially chooses the best (maximum) outcome if the opponent plays as well as possible (minimum). So in a typical open position with lots of pieces flying around, where there are dozens of variations to calculate, a computer tends to have an accentuated advantage over a human player of similar strength. For many years masters and grandmasters have carefully avoided wide-open positions (like those arising from the King's Gambit, for instance) for that very reason. Playing the King's Gambit against a really strong program looks very much like suicide. You start by giving the thing an extra pawn, which is enough of an advantage for it to win. Then you try to outplay it in its natural environment. It's like fighting a crocodile underwater.
At the other end of the spectrum, there are a few closed positions (i.e. with locked pawn structures) where even very strong chess programs fail to see what a reasonably good human player spots immediately - for instance, "this must be a draw because White's queen can never escape". (However, it might also sometimes happen that a program spots a clever and previously unnoticed way to break that kind of impasse).
Returning to my assertion that Dennett is wrong in saying that "The best computer chess is well nigh indistinguishable from the best human chess," I can immediately think of two classic counter-examples. First, the game in which Deep Junior, with the Black pieces, sacrificed a bishop on h2 and soon after forced a draw. If Kasparov had tried to play on, he risked losing. No one had ever even seriously considered that sacrifice before in the given position, although the general type (the "Greek gift") is one of the most familiar even to beginners. That certainly wasn't indistinguishable from human play, because no human had ever dared to play it. My second counter-example is the way Deep Fritz squashed world champion Vladimir Kramnik flat in the sixth game of their match last year. I was watching live on the Web, and when Deep Fritz played 10.Re3 I thought "Great! the stupid computer is going to get thrashed by Kramnik's ultra-sophisticated play". After some more foolish-looking moves by White, at move 20 I thought the game was definitely going Kramnik's way. But lo and behold! 25.e5! introduced, not so much a tactical melee as the threat of one. Kramnik shuffled his pieces anxiously, on move 30 Deep Fritz grabbed a pawn - and then it was over. Deep Fritz remorselessly ground the world champion down, forcing him to resign in just 17 more moves. In the final position Kramnik, still just a pawn down, could hardly move a single piece. In that game Deep Fritz played the final, technical phase like Bobby Fischer. But it played the attack between moves 10 and 30 better than Fischer could have! Its moves looked like a beginner's, yet they defeated Kramnik.
Strong programs have a big "psychological" advantage over human players, in that they don't have any psychology! Even super-grandmasters like Kasparov and Kramnik, on the other hand, very quickly start to exhibit signs of nervousness after a few games. Eventually, this can assume proportions that start to resemble post-traumatic stress disorder - especially if the human being has had a nasty shock, such as
I am sure that there are many other solipsists out there.
I am both a chess player *and* an AI researcher, so take my nonsense with a grain of salt. :-)
IBM cheated in the spirit of the game. Who defined the spirit? IBM did. They hailed the game as the demonstration that a computer system could defeat Kasparov in a chess match. But a computer system didn't defeat Kasparov: a half-dozen computer systems beat him, each one different from the last. modified by AI researchers and a team of chess masters. And they didn't tell anyone: so far as I understand, it got leaked after Kasparov discovered that Deep Blue wouldn't make the same move twice and that inspired an investigation.
It's one thing to consult with advisors. It's another thing to have advisors heavily modify your brain mid-match. What did IBM prove with all this? Just that Kasparov could only be beaten if they kept changing the goalposts on him.