Humans might be much better now than when Kasparov drew Deep Blue as well. That was over ten years ago. Also, you have no way of knowing if computers are "about 200 elo points stronger than any human who has ever existed", because computers don't play in tournaments against humans. All you know is that they are stronger than they were.

And yes, I do have an idea of what I'm talking about.

EvilSteve:

I think that if another match takes place it should be at a super-long time control, possibly with an adjournment. (Of course, the adjournment would have to be fairly short, and the human and his second would have to be monitored, because if they left they might use the engine they're playing against).

If the best human player in the world played Rybka at something like 40/4 SD/3, I think that Rybka would lose the match. I don't understand why all the man vs machine matches have been at normal time controls. The humans should demand longer time controls.

btw for what it's worth, I just asked Jan Gustaffson....rated well above 2600 FIDE, recently played in tourneys with Ivanchuk, Kramnik etc. works with lots of the best players in the world and computers all the time.

He said the idea that Carlsen could have a chance against Rybka in a few years is completely insane.

Anyway listen, come to me with one of the top 30-50 players in the world who agrees with you and then maybe I'll take you seriously. I have no problem getting more opinions from GM's almost all of whom I'm near certain will tell you that you have no understanding of how strong the computers are.

Wait, how could I forget - Jan plays poker himself doesn't he? At least he used to.. Do you know if he posts here?

As far as thread goes, I stand with my opinion - no chance for humans. And Discipline, I think your logic is wrong - if there was a chance for humans it would be now, rather than 2 years from now..

Why would it be now instead of two years from now? The new generation of GMs hasn't fully matured yet.

Curtains, do your GM friends tell you why they think that the idea is completely insane? Have they played practice games against strong engines with long time controls? Appeals to authority don't really tell me anything. Sure, some GM thinks that the idea is insane. But why? The fact that they are above 2600 ELO doesn't mean that they know for a fact how a 2850 strength human would do against Rybka in a game with a long time control. Why would it?

You are right, for whatever reason you, rated like 2000 or something, understand everything. In the meantime every top player in the world, who has a much firmer understanding of chess strength, how good Rybka is and how rapidly it's improving, may disagree with you, but what would they know.

Blabbity blah, appeal to authority, blabbity blah. Just tell me what your GM friend bases his opinion on. My guess is, I don't know, nothing?

You know someone rated around 2650 can probably get a sense for just how incredibly strong a chessplayer is when they spend all day every day analyzing with this player (or engine), and have also spent huge amounts of time analyzing with and playing against the top players in the world.

I don't know why I'm bothering with this thread, I mean really there's no way I can prove that you have no idea what you are talking about, but you don't. I'm done with this thread, good luck convincing everyone of your theories.

Carlsen will not be better than rybka ever. Computers are getting much better much faster than people.

I respect that your 2650 friend has a much better feel for exactly how strong Rybka is than I do, but:

a) this forum is not going to function if you are right about everything because of your GM friends

b) you still haven't said what your friend bases his opinion on other than his "feel" for how strong Rybka is

To keep things amicable, here are some sincere questions: has your friend played training games against Rybka? Has he analyzed training games that stronger players have played against Rybka? If so, what % does he score against Rybka in these training games? What % do the stronger players score? What time control do they play? Does he think that their % would go up with a longer time control? Why does he think that it's insane to say that Carlsen could ever become better than Rybka? Is it just sheer tactical ability? What would he estimate Rybka's ELO at? What ELO does he think that Carlsen and/or one of the other wunderkinds can reach?

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

Looking forward to listening to this rant tomorrow morning curtains I like hearing you talk about stuff like this.

False analogy. Checkers is solved. Chess is a century or more away from being solved, if it's even solvable.

Marion Tinsley got pretty close.

1) I have no idea who Carlsen is
2) It is a good sign for the future of this forum that we already have a forum fight!

As an interested observer and a terrible chess player, I'm rooting for nakumara and whoever this carlsen fellow is to beat the machines. As a random side question, how does the newest chessmaster engine compare to rybka?

If you played one and then played the other, I doubt you'd notice a difference. Me neither though, along with almost everyone in this forum. Sorry that might not be the most helpful answer, but any commercial chess engine on a modern computer will be really, really good (along with some you can download for free). Rybka's better than chessmaster but I don't know by how much.

Can I redirect this thread slightly? Sorry if this is an unwelcome hijack or should be just a different thread, but I've been distracted thinking about discipline's statement that each generation is significantly stronger than the one that came before. Let's say that up until now this statement has been correct. How much longer can this continue? Let's assume that nutrition, environment, etc have plateaued so our brains can develop maximally (I realize this isn't true everywhere) and that each generation will be approximately equally mentally endowed (possibly not true, but just imagine). At some point we simply reach the threshold of human performance, right? At some point, even if additional training methods or theoretical ideas are found, the change in strength would be negligible. If this seems reasonable, how do we know that we haven't achieved that point already? Computers give us a great way to achieve mastery much, much faster but why must that necessarily elevate the best players higher?

Same reason people are still able to shave time off the world's best records at the olympics each 4 years, with no end in sight. And physical limitations are presumably much tighter than mental limitations with humankind utilizing still just a tiny fraction of their brain's potential.

Swingdoc interesting issue but I think you should put that in its own thread.

Also, am I the only person in this thread who thinks of Neo and The Matrix with all this Man (Super GM) vs. Machine (Super-PC) talk?

Kasparov in his prime would almost certainly dominate the world of chess today.

Maybe not if the world contains Kramnik in his prime?

But I guess every1 has to have a Geller.

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.

LOL, I just had Fritz play this game out vs. itself, and 48. f8Q+ to avoid 50-move repetition in what it thinks should be a won game is a hilarious move

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.

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.

Yeah, he didn't just get close, he achieved that goal, in my opinion. The only thing that stopped him was cancer.

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.