a 2700++ GM could crush ANY engine if you take away the opening book IMO. but who knows it might be a good match.
GM's vs best engines
Sort:
The point is computers have a very very low rate of inaccuracy compared to humans. So while they wouldn't always find the best moves (even after the opening) the human opponent will make enough small errors during the course of the game to fall into a worse position.
Not that the computer would win 100% of the games, but a very strong engine would beat a 2750+ player in a match every time. They already have matches where odds of a pawn and move were given and the computer won (Rybka3) and today's engines are even a bit stronger. Playing without opening book and Nalimov aren't nearly as bad a handicap.
Computers don't calculate to 60 ply anyway. If they did it would be too unreliable in all but the most forceing variations.
On unofficial rating lists, Deep Rybka is ranked @ ~3000, which is much stronger than any human player.
kevinjin wrote:
On unofficial rating lists, Deep Rybka is ranked @ ~3000, which is much stronger than any human player.
Yes, but I thought it was a "different" type of rating that applied to computers and doesn't translate into a "human" strength, though it would seem to (much like Tactics Trainer and Chess Mentor don't correspond to rating strength but they use a similar system that makes it look so). I'll have to find where I read that or maybe someone else can clear up the confusion.
Now why do computers get destroyed without an opening book? I am guessing it is because programmers have a hard time in evaluating the values of certain early moves and positions?
What if a computer is forced to play an "inferior" opening (like Blackmar-Diebert Gambit), will they still eventually turn the tables and crush the super-GM?
engine vs 2750+
nonsense
not worth comparison
I was just curious if anyone has any idea...
How would the strongest chess engines (like Deep Rybka or Stockfish) perform against super-GMs rated 2750+ if they were limited to lines "only" 25-ply ahead under standard time controls and no opening book or Nalimov table? How about 60-ply? How about each of the above but with an opening book and Nalimov table access? At what point are most of these matches in favor of the engine, the GM, or mostly draws? I have heard that rating an engine "2800" is different than for a human.