Evaluating Opening Lines using computers

[Airhog]

I think computers have reached a point in chess when they can play as good as the best humans in the world. I watched the games last month on ESPN, featuring Kasparov and Deep Fritz. I was thinking the other night about doing large evaluations of opening position using a computer. Even the fastest off the shelf computer could not easily evaluate a given opening position. there are just way too many branches. And the only way to truly evaluate an opening position is to start from the given you wish to analyze, and play 1000's of games, in hopes of finding a real weakness with a position that makes it unsuitable for opening.

But, I thought of a great way around this limitation of processing power. What if you used a distributed computer model. Similar to the Seti project, where each program analyzes a small amount of data when it is not being used. I think with enough computers helping, you could actually evaluate positions. Each computer would be asked to analyze a very narrow number of moves. It would look like a pyramid, with the first move at the top. Each succesive layer would get larger. Of course I still think it would take a large number of computers, say 10,000 to really make an impact.

[Shkspr]

You know, at some point Marmel's going to do a forum search for "Model", "Opening Line", and "Deep Fritz".

Hi, Marmel.