There's another chess tournament between a computer (Deep Junior) and a human (Garry Kasparov - the guy who lost to IBM's Deep Blue). More info at http://www.chessbase.com/events/events.asp?pid=183 and http://www.x3dworld.com/Entertainment/CI_X...g_Frameset.html. Today they played the first game, and Kasparov won.

So does Deep Jr. Have the ability to beat Deep Blue?

will never be known. deep blue has been dismantled.
seems like kasparov vs deep blue was just an ibm marketing campaign "see? our machine is so powerful it can beat the best chess player in the world" and ibm had no interest in furthering the cause of chess progression.

The following paper is nice because it's from the people that built Deep Blue and it dispells a lot of myths about what it could and couldn't do:

http://sjeng.sourceforge.net/ftp/deepblue.pdf

I know current top professionals are far above this level.

But as tangent said, we're never going to see DB play again. Nothing to win for them, but a lot to lose...

If one takes Kasparov at face value, he claims that reviewing his Deep Blue games using an earlier version of Deep Junior showed that Deep Junior was generally better, except for a couple of moves which he has claimed Deep Blue must have had help with. But my guess is that those couple of moves were the result of Deep Blue's hundredfold speed advantage over Deep Junior.

ff123

Not quite - the speed advantage is worth almost nothing. The tree of a chess game is exponential, and any improvement that allows you to cut down on the branching factor (number of moves you have to investigate in each position), gives an exponential return.

Compared to nowadays programs, Deep Blue was utterly primitive in that regard. It may have had an apparent speed advantage, that doesn't help when it has to look at *a lot* more moves. The end result is that current programs are searching deeper than Deep Blue despite raw speed disadvantages.

Likewise for the evaluation. It was touted as being 'obviously' better than what the PC programs do since they could afford to evaluate anything at no cost due to the hardware. The paper I quoted above doesn't leave much whole of that impression - PC programs do all that stuff too now, so there was nothing exceptional about DB there.

Kasparov's problem was that in the previous match he had prepared by playing against a PC program (Fritz 3) of that time ('96 program on '96 hardware) as a preparation for DB. But in '96-'97, DB *was* better than the PC's by quite a bit. Ironically, he was advised to prepare this way by the people that are now promoting Deep Fritz and Deep Junior. The net result was that when facing Deep Blue he was getting something totally different than he expected. He interpreted that as human interference and accused the DB team of cheating. They later debunked that claim by publishing all logfiles, but I believe that Kasparov to this day won't admit he lost to a machine.

I personally believe it's very hard to look at 6 games and judge who is better. GM's are very bad in recognizing what a computer can and cannot do on top of that. So I believe that DJ is better than DB, but judging that from 6 games is very hazy.