About 12 years agoit took a mid sized multi-node parallel linked supercomputer and specially designed software to give Kasparov pause. Is the computer hardware and chess software available to retail consumers powerful enough these days to accomplish this?
An Intel chip in a PC now has more power than Deep Blue did in 97.
The Core 2 quad processor is rated at 49 GFLOPS while Deep Blue was at 11 GFLOPS.
Now a lot depends on the software and how well it can utilize this speed.
Deep Blue had a team of programmers continuously tweaking it (before and between games, not during) with the input of several chess masters. IIRC Kasparov objected strongly that Deep Blue had been programmed to beat a single person (Kasparov) and not programmed generically to play anyone.
I doubt off the shelf software even today could beat a top flight grandmaster but it will likely beat most people.
I always thought that they were cheating and he probably did too. A supercomputer combined with a whole team of programmers and chess experts isn’t a very fair matchup for anyone but Kasparov still did pretty well despite that.
Pardon my ignorance, Shagnasty, but how can one cheat at chess? Aside from shouting ‘look over there!’ and switching some pieces around, of course.
Kasparov asserted that a human was involved during the second game, which IBM denied. They did tweak the software between matches, allowing the Deep Blue to avoid a trap in the final game that had worked twice previously. The film Game Over: Kasparov and the Machine suggested that it was likely a publicity stunt involving humans on the computer side, to try and gain stock market value for IBM. The claims haven’t been substantiated however, though I’d agree it was done for publicity.
More recent chess computers have focused on software programming, rather than hardware iterations. Deep Fritz which took on the then grandmaster Vladimir Kramnik in 2006, winning 2 drawing 4, used a PC with two Intel Core 2 Duos (8 million iterations per second vs. Deep Blues 200 million/s, but often searching to greater depth for a single move).
Where there are more minds looking at a move and it’s results and potential responses, fewer mistakes will be made. I can beat any 6th grader I’ve ever played, but I can’t beat three of them working together.
Ah, that seems simple enough. Thanks for the info, Captain Awesome and Harry 1945.
I suspect Kasparov could beat an infinite amount of sixth graders at chess, so numbers dont always scale
The way I interpreted the word “cheat” was that they were cheating at the challenge, not at chess. The challenge was see whether “a computer” could beat “a Grandmaster”. As shown in other responses, it’s highly questionable as to whether that’s what was being tested in the Deep Blue versus Kasparov games.
Also, in the second match between Kasparov and Deep Blue, IBM set a grueling schedule to tire Kasparov out, since of course the computer doesn’t get tired. And on top of that, most chess analysts seem to think that Kasparov was just having a bad week at the time, another thing that doesn’t happen to computers.
These days computer programs frequently beat grandmasters, including ‘top-flight’ ones, and these are programs you can buy off the shelf (like Fritz or Rybka). I happened to be reading the game score from this match this morning…Fritz trades a couple of wins with (then) World Champion Vladimir Kramnik.
Don’t have a cite for this, but I recall hearing on a radio program recently that computers haven’t made more progress against grandmasters because the humans have learned to exploit the weaknesses in the way computers play.
Do you think you could have beaten Josh Waitzkin at that age? Just curious, not trying to pick a fight or anything.
I was being beat by a Radio Shack chess computer when I was a kid back in the late seventies! I presume the software has gotten stronger since then.
I’m also seeing a Web site out there: How many 6th graders could you beat…
It’s funny because I read somewhere it’s quite the opposite - They don’t do computer vs human matches anymore because the computer always wins (or draws) so it’s not very interesting for anyone involved.
I’m not saying you’re wrong - in fact it seems more likely that people would figure out how to exploit the AI. I’d be curious to learn which is correct.
Hey, Capablanca at age 5 would probably have wiped the floor with most of us, and that was after learning by watching his dad play (no instructions).
One of the problems with the chess playing games is that they typically use a brute force method of checking a zillion moves and picking the one with the best chance of winning. Not very smart, but powerful.
A more interesting challenge is the game of Go which can’t really be tackled in this same way. The computer needs to attempt to find the same patterns the human can see. For this game there is progress being made but there is a long way to go.