I think it was Kasparov who said a few years ago that soon, the test of a computer v. human match would simply be whether the human wins any games. By that measure, I think that the humans are still in pretty good shape.
Also keep in mind the huge role that opening theory and opening selection plays in human-computer matches. It's hard to completely separate the strength of the engine from the strength of the team that optimizes its opening book.
Quote:
Originally Posted by Discipline
If Rybka and Carlsen played a match today with a time control of 60 minutes per move, five moves per day, Carlsen would win convincingly.
Sure, but I don't think anyone has tried to develop an engine with correspondence chess in mind. I imagine that there are ways to optimize play for those sorts of time controls that just aren't implemented, but I'm not very familiar with the computer chess scene so I could be wrong.
But sure, humans are still better at very slow time controls, without a doubt. Especially if they have computer assistance.
Quote:
Originally Posted by DaMaGor
Marion Tinsley got pretty close.
Quote:
Originally Posted by Fleebrog
discipline is right, it will be just like when humans achieved the ability to play perfect checkers alongside the machines
wait that didn't happen
Yeah, he didn't just get close, he achieved that goal, in my opinion. The only thing that stopped him was cancer.
Quote:
Originally Posted by Dire
In this position, if any human couldn't almost immediately evaluate the position you'd think he's a weak player. But give it to any computer and it wants to say white has a crushing advantage. Computer chess is such a paradox. Yes, they are incredibly effective at winning games, but they are also terrible.
Well, it's a question of what you want. If you're building an engine, the important part is that it generate the best moves. Whether or not its evaluation is very accurate is sort of immaterial. And really, who needs to know "how many pawns" you're winning by...that computers might stumble into blockades is a bigger deal, in my opinion, but I imagine that this is something that programmers try to avoid. Getting the numerical score correct in positions where it doesn't matter isn't that useful.