Author: Graham Laight
Date: 07:19:34 05/15/01
Go up one level in this thread
On May 15, 2001 at 09:45:24, Robert Hyatt wrote: >On May 15, 2001 at 06:03:11, Graham Laight wrote: > >>I still say it's quicker and easier to draw a graph of strength of players >>plotted against proportion of draws. >> >>Since I don't have a copy of chessbase, the graph below is based on guesswork >>rather than actual study - but here's a quick example of what it would probably >>look like: >> >>Percentage Of Draws >> >> >>100 | * >> | * >> | * >> | * >> | * >>75 | * >> | * >> | * >> | * >> | * >>50 | * >> | * >> | * >> | * >> | * >>25 | * >> |* >> | >> | >> | >>0 | >> ------------------------------------------------------ >> 1000 1500 2000 2500 3000 3500 >> >> Elo Rating >> >> >>If this were accurate, the maximum possible Elo rating would be 3500, because >>above this level, almost all games would end in draws. >> >>From this, one can calculate when computers will be able to play at the maximum >>level. If normal (ie not Deeper Blue!) computers advance at about 100 Elo every >>5 years, and they are currently at 2600, then they need another 900 Elo to play >>perfect chess - which they will be doing in 45 years. >> >>And what ply depth will be needed to achieve this? If they advance 1 ply every >>3 years, then they will advance 15 ply in 45 years. >> >>So - in conclusion - to solve chess (or at least to play "perfect" chess), >>computers need to search 15 ply deeper than they do now (and probably increase >>their knowledge at the same rate as they're currently doing). >> > > > >I don't like that definition of "perfect". Just because they are nearly >unbeatable, they won't be playing "perfect" at all... You won't play >perfect until you can prove a move from the starting position is a forced >win, or else prove that all are forced draws. In which case, you cannot prove that smoking causes lung cancer - because although there's an excellent statistical correlation, nobody yet has actually seen cigarette smoke cause lung tissue to become cancerous in real time in a live human. But of course, Bob is right - so the expression "perfect" should really be changed to "beyond reasonable doubt". So when such a computer + program plays a game, it will win almost all its winning positions, and almost never lose a position which can be drawn. -g >>Again - all the figures in the previous 3 paragraphs are my guesses rather than >>rigorous studies. Anybody who knows the numbers more accurately is welcome to >>correct the figures. >> >>-g >>
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.