Because some people here feel that grandmasters still might have a chance against Rybka without an explicit handicap if only Rybka is denied various advantages, I am publicly making the following offer.
I will donate $1,000 of my own money to any grandmaster who can win a six game match from Rybka (my choice of version and pc) under the following conditions:
1. Opening book: Rybka is limited to a 3 move deep opening book I would prepare (to provide variety and avoid prepared games).
2. Tablebases: None
3. Hash table size: 512 MB.
4. Color: Human gets White every game!
5. Time limit: FIDE time control (90’+30″ increment) for the grandmaster, 45’+15″ for Rybka. External clock governs human, computer clock governs computer.
6. breaks: five minute break twice each game on request by the grandmaster. Computer may reboot if frozen.
7. Schedule: two games per day for three days, intermediate break as requested by the grandmaster.
8. In the event of a drawn match, I will donate $500 to the grandmaster.
9. Playing location: my home in Potomac, MD (free accommodations to the grandmaster if desired).
Here is the link to the Rybka forum where the challenge was made. Special thanks to Scott K. for pointing it out to me.
Susan,
This issue is over. The computer programmers won. I have followed this “struggle” almost from the beginning of computer chess, going back as far as Sargon I on the Apple II computer, later Sargon II which defeated a mainframe at the time. I heard the overconfident statements from the real chess players, that “yeah, it is very nice, but the computers will never play a decent chess”. Later “never will play as well as the masters”. Later “never will play as the grandmasters”. Later “never will defeat the world champion”.
In 1997, despite all the complaints from Gary, Deep Blue has defeated the world champion in a match. Not much later it didn’t even take a special computer, an up to date laptop………you know the rest of the story.
Now we have computer programs which are playing above the level of the human chess players. Of course you can dumb the programs down, take away some of the features and a grandmaster maybe able to defeat the computer again, but this would defeat the purpose (and the concept). The world must accept that the early guesses about the ability of computers playing chess were wrong. Computers became excellent “opponent when human is not available”, excellent training tool, excellent practice tool, excellent teaching tool, but humans are not forced to play against computers. In fact, the faster humans disassociate themselves from trying to defeat computers in chess, the easier it will be to erase the memory of the “computers will never……” claims.
Gabor
omg $1,000 !!! lol …
A grandmaster cannot refuse this gold mine challenge
😉
BTW : I vote for this april joke.
This read as a challenge from SP: but apparently it’s not.
The constraints on opening-books and tablebases are ludicrous. The GM would come to the table with a richer opening book: the 5-man tablebases can be created on the side in no time flat as part of a valid minimaxing search-algorithm.
The GM gets White in every move – and the author of Rybka gets nothing if Rybka beats the GM. Let’s have some respect for the authors of chess engines.
A better challenge : 500$ for every game won by human.
It would be interesting to see whether two grandmasters, cooperating and consulting, could defeat regular Rybka (Rybka not lobotomized of its opening phase and endgame phase knowledge).
Having a pair of humans work as a team would reduce the human mistake or blunder rate.
Human blunders, minor if not major, are a significant source of chess engine victories Man v. Machine matches. Let us see how the engines do when avoidable human mistakes are eliminated.
People say this Man v. Machine issue is all settled. Yet the debate rages as to whether computers play better moves, or computers just fewer avoidable mistakes.
Kramnik lost to Fritz_10. But I suspect Kramnik + Susan Polgar would have won that match.
GeneM
CastleLong.com
The most fair way to judge if computers are really better at chess would be in the following ways:
1. Give the human 3 times more thinking time.
2. Remove all table-bases from the computer’s memory.
3. play Fischer-Random chess.
4. Allow the human to consult a computer with limited capability (3 ply deep).
Does anyone know who did the chalange ( his name ) ? In the Rybka Forum he uses the nickname
‘lkaufman”, but I didnt get to know who he is.His profile doesnt help much.
I would think a top grandmaster would be an overwhelming favorite and a mid-level grandmaster maybe a 2-1 favorite in this match. That means the mid-level GM’s expected winnings are about $700, for a gruelling three day event. Compare that with their expected winnings in a large tournament, or what they could get from 3 days of simul fees or lessons, and from a purely financial point of view this challenge is not so attractive, given the publicity at stake. Put a few million bucks on it like IBM did with Kasparov and it gets more interesting ;-).
I didn’t see the forum post Lkaufman is almost certainly IM Larry Kaufman, a computer chess whiz from way back.
lol
I think he is some sort of a perv. He will lure in to his home a GM – to kill him and suck the brains from him lol
can’t stop laughing
rastamann
Looks like the kitty is up to $6K now from the comments on the referenced site.
Very good initiative Susan. Congratulations!
Maybe raising of 10.000 USD would make this offer more attractive.
Keep up the good work.