Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Go programming

Author: Roy Eassa

Date: 08:48:05 02/13/04

Go up one level in this thread


On February 13, 2004 at 11:21:51, Alberto Rezza wrote:

>>How many players in the world are at least good level amateurs in go?
>
>It depends. If "good level amateur" is something like 1-dan (on the european
>rating) then there must be at least one million players in the world who are
>this strong or stronger.
>
>However, go and chess are too different, and it's very difficult to compare
>their rankings. I think some people have tried to get equivalent ramkings by
>comparing the probabilities of winning for a given rank difference in the two
>games, but the results are not very meaningful: it turns out that there are,
>from beginner to world champion, many more "steps" or levels to progress in go
>than in chess; so if you fix an equivalence like, for instance, 100 Elo points
>for every go stone, you get a much lower probability to win with a "one level"
>advantage in go than in chess.
>
>Another approach is to compare population sizes: there are many more chess
>players than go players - the go players are only about 30 million, mostly in
>China, Korea and Japan. From a much bigger pool you should also get better top
>players, so the chess GM's should really be compared to the very top go masters,
>the 9-dan professionals. This might put a 1-dan pro at about 2350 Elo, and a
>1-dan amateur near 1700 Elo.
>
>The problem is that this is not consistent with the "feeling" reported by people
>who are strong in both games: for example, a friend of mine who was about 2000
>Elo in chess, and also a 1-kyu (euro) amateur go player, told me he believed he
>was definitely stronger in go than in chess.
>
>Personally, I think 2000 Elo in chess would be slightly better than 1-kyu in go,
>but then again I am not a chess player... (beginner at chess, 1-dan in go).
>
>Alberto


Awhile back I started using this rule of thumb: take your AMATEUR (does not work
for pro) kyu ranking in Go, multiply by 33 1/3, and subtract that from 1800
(perhaps it should be 1900?), and you'll get an approxmiate equivalent of a USCF
rating.

It's far from perfect, but in general it works better for me than not having any
approxmiate formula to use.



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.