Garry Kasparov and Deep Junior are playing their sixth and final game today and it will be shown live on ESPN2. Is this man’s last stand? Deep Junior is significantly differnet from the other supercomputers like Deep Blue in that it tries to think more like a human, i.e. less materialistic, able to recognize long-term implications.
=“http://www.kurzweilai.net/meme/frame.html?main=memelist.html?m=4%23527”] Is chess soon dead?
Of course not, but it is an interesting read.
Kurzweil’s predicted doom is an old and obvious fallacy.
Cars can “outrun” any human runner easily. Do we still watch track races? Sure. Human vs. human will always be interesting, both to participants and to fans. Machines do a lot of things better than us. Who cares about that in sporting terms?
Computers long ago passed the top human players in checkers and backgammon. A lot of people still play backgammon against other people. Presumably checkers is still played in rural general stores.
Losing interest in chess is a ridiculous premise. (Unless a “perfect play” is found for a player. But then they will change the starting rules like they do for tournament checkers.)
I disagree with ftg. I for one already find chess less interesting knowing that computers can/will beat the top player. Sure there will always be chess tourneys and casual players but I think its lost it’s mystique. Once the idea that strategic genius can still win over tactical bulldozing is finished, interest in chess will take a big hit.
I thought the final game was on Sunday? What time is it on ESPN2?
umm… Now.
How many of the previous 5 games against Deep Junior has Kasparov won?
They just tied the series.
It isn’t like a car vs. a runner at all…Kasparov could’ve beaten the computer. He did infact, beat it one game in the six game series (and tied 4).
I don’t find it any less interesting (actually, I prefer playing GO. and luckily, no computer will ever be made that can beat the world GO champion.
I look at it this way: Kasparov isn’t playing a machine, he’s playing all the people who designed the machine. And those people have used the experience of lots of other chess players in developing the programming for the computer. So, in essence, Kasparov is playing a lot of people, including several Grandmasters. The fact that he’s done as well as he has is a tribute to his skill.
And, of course, if he should get the measure of the machine, it can’t adapt on its own – the programmers will have to do it. We’re not obsolete yet.
While I’m as fascinated by these matchups as the next guy, I honestly don’t understand why anyone would lose interest in chess just because the best human players can’t beat the best computer players anymore. The best human players are such extreme outliers with regards to chess ability that computers have been able to beat nearly all human players for years and years and years already, and that does nothing whatsoever to detract from chess matches between amateurs.
So … if they’ve playes 5 games … and Kasparov has tied 4 games and won 1 … then how can the series be tied?
Really? What makes you say that? What is it about GO that prevents any computer algorithm(s) from being able to beat a world-champion human opponent?
Why is Karparov playing this match?
He isn’t even the world champion.
tracer…I could be wrong, but the series is 1-1-4. They played six games.
Well playing on a 19x19 (standard) board, there are about 10[sup]750[/sup] moves. It would be very, very, very hard to make a computer that works as well as Deep Blue or Deep Junior.
A chess program doesn’t keep the entire decision tree for every possible chess game in memory, any more than a GO program would have to keep every possible GO game in memory.
A typical mediocre chess program calculates the best move-countermove tree out to only a few levels deep, pruning bad decision-branches as it goes; and that’s enough to defeat any player short of a Senior Master. The program running on Deep Junior is a little more sophisticated; I believe it’s using the same chess engine (called “The King”) that’s used in Chessmaster 8000/9000. (Somebody correct me on that if I’m wrong.)
A GO-playing program could be written along similar lines, and given enough memory and processor speed, it too could whup even the greatest GO players.
Out of interest, how are tournament checkers different from casual checkers?
Entering today’s game Kasparov had won 1 game, Deep Junior had won 1 game, and they had tied 3 games, making the “score” 2.5 to 2.5. Then they had another draw today, making it 3 to 3 for the final score.
I was really disappointed, both when it appeared that Kasparov had offered a draw and when accepted the draw offered by DJ’s people. Understanding that anything can happen, it certainly appeared that Kasparov was rolling along and was in a position to be able to win. I think he got rattled by a move near the end that he didn’t expect, but he still seemed to have the overall edge. Must have been lingering doubts about what happened after he lost to Deep Blue. Anyway, after actually tuning in and watching this for nearly 3 hours just to have it end in a draw like that was disappointing.
As for why he was playing, I could be wrong but I think he’s still considered the number 1 player, even though he doesn’t have either of the two major “recognized” world championships right now. The commentators (who really needed to quit trying to make so many analogies to “other” sports) were discussing the impending “unification” of the world championships this year. Kasparov is one of four players competing to have the overall unified championship.
Anyway, I find it amazing that the guy can compete at this level, after 20 years, against the most highly developed computers out there.
I think the double-quotes in that sentence should be shifted one word to the right. 
First of all, the number of moves possible on any given turn is much higher in go than in chess, by a factor of nearly 20 in the opening of the game in fact. If a computer program tries to explore the game-tree to a depth of only 6 half moves (which makes for a pretty pathetic chess program, let alone go), that already means evaluting 64000000 times as many positions as the corresponding chess program.
Secondly, it’s not at all obvious how to prune the search tree in go. Pruning the tree effectively depends on being able to evaluate a given position fairly quickly, which is difficult in go. Heck, beginning players often have difficulty determining that the game is over (i.e., that no more useful moves are possible).
It’s been a while since I checked the state of go software, but I was under the impression that even a mediocre go program was a significant achievement. Splanky is right; it will be a long time before we see Deep Blue’s or Deep Junior’s counterparts for go.
Was there a real reason for the draw? I know nothing about chess but for some strange reason I watched a good amount of this. Then when the moment came I got a phone call! Was he that afraid of losing?
Edward, it seemed that Gary was too afraid of losing. It was more important to him to not lose than to win.
I think humans will be beat at go by quantum computers, which means it will probably be a few more years but not more than ten.