Author: David Blackman
Date: 05:19:21 08/26/01
Go up one level in this thread
Keep in mind there was a series of machines. ChipTest -> Deep Thought -> Deep Thought 2 -> Deep Blue -> Deeper Blue. They were similar in some ways, very different in others. It was Deeper Blue that finally beat Kasparov. Deeper Blue was significantly faster than Fritz running on any platform they are likely to have this year. Deeper Blue also had a significantly more complex evaluation function. Whether the eval in Deeper Blue was actually better is a difficult question, but certainly it evaluated more positional things for most positions. The quality of the search algorithms is another question. Deeper Blue's search is not very well understood by anyone who has chosen to talk about it. It was probably a lot better than Fritz at finding long forceing lines. It is quite possible that it wasn't as good as Fritz at finding lines that are not obviously forcing but still important. But that is not completely clear. Fritz probably still makes use of null-move pruning, which improves depth, but occasionally makes you miss important stuff. Deeper Blue probably didn't use null-move, at least not in the parts of the search near the root. Opening preparation is another question. Both would be quite good in this respect, i think, but i don't know how much special preparation Fritz will do for Kramnik. It will have to do quite a lot to be as well prepared as Deeper Blue was for Kasparov, but then opening preparation might not be quite as important against Kramnik. Overall, i think Deeper Blue was probably stronger than any Fritz you will see this year. But the gap is narrowing and the uncertainty is getting greater every year.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.