Author: Dann Corbit
Date: 16:19:27 09/07/01
Go up one level in this thread
On September 07, 2001 at 19:11:50, Mike S. wrote: >On September 07, 2001 at 18:07:16, Dann Corbit wrote: > >>I don't think you can use the plies in a comparison. >>(...) >>What a ply means from program to program is (unfortunately) incredibly >>non-standard. You can make meaningful comparisons of ply depth only for a given >>program against itself (and even then, only if the search tree retains its shape >>for both tests!). > >I know, but I think the ply depth comparison would at least be a bit more >meaningful than the node rate comparison between these two very different >systems. The plies comparison won't tell us everything, but I tend to come to >the conclusion that the node rate comparison tells us close to nothing. > >In short terms, ply depth tells always much more about strength than node rate. > >(And if this last sentence if proven wrong, I have wasted a lot of time during >the last 20 years with computer chess :o) I think the bottom line is this: "The proof of the pudding is in the eating." IOW: If it continually kicks bootie playing chess, then it's strong. If it gets its bootie continually kicked then it is less strong. Deep Junior plies are not at all comensurate with Deep Fritz (for example). There was a test a while back where someone did a "material only" evaluation. 20 ply searches were less effective than 2 ply searches for an eval function with a tiny bit of brains. There is no measure (other than playing strength) that can tell us how strong a program is. I do agree that plies searched is a very important measure *within* the same program. However (as an example) if you turn on some of the pruning algorithms in Beowulf, the depth will reduce by about two plies (for some positions) but the strength is changed very little. So even within a single program, we must ensure that the tree is formed by a constant algorithm.
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.