Rybka vs. Top Human Players

It is generally accepted that with standard time controls for both players, humans would take a huge number of games to win against a computer like Rybka. So, how often would humans get a draw against Rybka? For instance, Bobby Fischer in his prime might manage 4 draws in a 14-game match (I personally think even he would only get 2 or 3); Kasparov might manage 3 draws in a 14-game match (I think around 2 would be more likely) in his prime. When I say "prime", I mean the best tournament performance of that players' career and their life.

To repeat the question: how often do you think these humans (on their best weeks ever) could get draws against an extremely top-level computer (take Rybka as the example)?

0

This would be fun to watch, but you'd have to pony up $$$$$ to get a top GM to take on this challenge.  BUT, Steinwitz has it exactly right......the human would score 0 points no matter how many games were played.  It might be interesting to see how "non-top-level" a computer it would take for the human to score some wins.  Maybe an iphone running its best program??

Really, "If you now how to play against computers"? Wish I knew that trick.

As I understand it, even 'anti-computer' strategies are quite ineffective these days. Computer engines have simply become too complex and diverse for any single set of strategies to work (such as, getting the thing out of its book, or surviving to an endgame).

But then, I don't understand it very far.

That's how I understand it, cberman.

don't know where to find the rybka - ehlvest matches but even at odds the best humans do not well at all.  transwarp [rybka] (c) on icc ate gms also for instance and of course once in a while the computer would lose to a human, but this was very infrequent. the hardware has become so strong and the opening books so deep in computer chess that it would be very difficult for a super GM to fare well in a match against a supercomputer.  in practical chess, there is no interest in opening lines where black has to find precise moves thru move 43 whereas in computer chess this is quite possible or practical to play to a theoretical very difficult but drawn endgame.  i think under the right conditions the human can still win, but i am optimistic and there is no interest commercially in such a competition human vs machine, i mean, it is known that the computers are so strong so what else there is to gain.  besides the precedent and reality suggest that the computers are stronger, 32 or 3300 or something.  even the best noncommercial engines play at such a level.  adams was trounced by hydra 5.5-0.5, this was the last rather major match that i can think of, this was as of 2006 maybe, i don't know. and before that kramnik lost to deep fritz, but this was a narrow loss.

the anti-computer strategy for draw i have used is to play hippo, wall it up, and premove, but this is not chess, and works better with white and not all the time. or play king's indian trades down the d-file and hang on for dear life for a draw.

[edit: jjeffrey, interesting links thank you!]

Read about the Rybka Ehlvest 2007 match at ...

http://www.rybkachess.com/docs/RYBKA_EHLVEST_2007/Rybka_versus_GM_Ehlvest.htm

See the results here...

http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=1614

More than three years ago, 6 games where Rybka always had black, and began each game a (different) pawn down.  Today there is no hope for humanity...

I don't know what you are talking about, didn't you see the thead where the 1300 with Black easily played Rybka to a draw using the French Defense. I suggested this was less than credible but I was put in my place by the next poster who assured me Black had played very naturally.

rybka didn't play the position out.  if she did, her superiority in strength in all likelihood would manifest itself quickly.  If not, I would refer you to Littlewood's Law. :-)

Quite. I've also read quite a few posts on here by someone called "sloughterchess" who seems to have had some tremendous results against top level computers.

I was thinking more along the lines of how strong computers are, not actual matches with GMs (hence my asking the result of theoretical matches between Fischer or Kasparov during their best tournament periods ever and Rybka, instead of, for instance, Anand or Magnus Carlsen now against Rybka).

An iPhone running its best computer program might be an interesting discussion, but I'd rather keep this discussion to full-fledged Rybka on a good computer. To make it more competitive, how about lowering time controls? Surely Rybka wouldn't win every game (against theoretical players of the quality I'm talking) if the time controls were as low as 1 minute for each player?

If Rybka could even win every one of those games in your opinion, how about seperate time controls? What kind of seperate time controls would humans require to beat Rybka?

Actually computers are more likely to win in Blitz or Bullet Chess. Longer games are more helpful for humans to force draws.