Re: [computer-go] Good-citation

[Geza Bohus <gbohus@xxxxxxxxx>]

chrilly wrote:

> The best I have found in my library is:
>
> Cho Chikun: Go A complete introduction to the game, Kiseido Publishing
> Comp., Tokyo, Second print 2002,
> In this standard-introduction book by one of the strongest Go-players there
> is an own chapter "Go and Computers". p.116-119.
> It is too long to type all here. I just write the end.
>
> "In conclusion, it will be a long time before computers are able to play go
> at the same level that has been achieved for chess and checkers. Major
Hi,

One aspect hasn't been mentioned yet in this thread AFAICS: it is not the case 
that go programs are not playing well but that people are. By a fairly natural 
measurement, the best go programs are as much better than a beginner (human) 
then the best the best chess programs are better than a beginner.

This measurement is as follows: pick your favorite number p between 0.5 and 1, 
say p = 0.75 . We say that player A is better than player B by one rank if A 
beats B with probability p. With this definition, we define ranks starting with 
the beginner level up to the best players' level (human or computer). So the 
distance of two players is their diffence in rank. The level of a player is 
his/her/its distance from a beginner. Picking a different p will give a 
different scale of course but that doesn't change the rest of the argument.

The distance between a 30 kyu and a pro dan player is much larger than the 
distance between a beginner in chess and the world champion. But the best go and 
chess programs are of the same level.

Not my idea unfortunately 8-).

Best,

Geza

--
Géza Bohus
gbohus@xxxxxxxxxxxxxxxxx

  Entropy requires no maintenance.
                -- Markoff Chaney
(Send me a message if you get the second joke in this sig.)

_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/