Author: Dann Corbit
Date: 12:34:32 01/06/00
Go up one level in this thread
On January 06, 2000 at 14:57:47, Chris Carson wrote: >Ref: Monty Newborn > >I think his book (How computers play chess) has this table: > >ELO PLY >1600 5 >1800 6 >2000 7 >2200 8 >2400 9 >2600 10 >2800 11 > >Thus: 200 points per ply. I think he also says the average >number of moves per ply is 36-38, so machine speed must increase >by a factor of 6 to reach the next ply. If someone has the book >handy, check my table. I will double check tonight. I think plies is probably the most important measure of strength. However, it is a serious mistake to imagine that n-ply == n-ply for two different programs. One program uses null-move, and the other does not. Big difference at the same ply depth (but it probably takes much longer to get to a given ply for the non null-mover). Suppose that two programs both reach 11 ply. But the second program has extrapolated to 30 plies using some check extensions and things like that for some special lines. Same ply depth, but one has a clear advantage. Even what "one ply" means varies from program to program in nomenclature. Finally, some plies are easy and some are hard. Suppose, in a difficult pawn endgame with each side having two pawns and a king, you have a long pawn race on your hands. A human can easily see what will happen in ten moves at a glance. That's 20 plies! It will take the computer much longer to see that far.
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.