Re: [gnugo-devel] Three stone games

[Morten Gulbrandsen]

I do appreciate this reply. 
If we programmers cannot get an exact 
definition of the average strength of a go program, 
then it is very difficult to improve. 

The real strength of Handtalk 97-6 should be 25 Kyu, according to the
game 
against Didier Garcia 1d 1999-07-20. It is ranked as 4 Kyu japanese. 

a tremendous discrepancy.

Against professionals we go programmers cannot survive. 
So perhaps it should be fair to state that gnugo has no human comparable 
strength. How many handicap stones do we need in order to survive ?
If 25 stones is not enough, then we have to increase the handicap level.

address@hidden:~$ gnugo  --boardsize  19  --handicap 13  --komi  0.5  
--color  white  --clock  600   --byo-time  600  --byo-period  20 
Illegal handicap: 13.
Try `gnugo --help' for more information.

I suggest that we increase the maximum available handicap number to 
more than 25, so that we can pay efforts on how to beat a shodan go
player.

What happens if a professional  1 dan pro starts gnugo 9x9 with 9
handicap stones.
I think gnugo can solve that. If not, we have to confess that it is
impossible.

When I started to play go, a korean 3-dan ranked player gave me 9
handicap stones,
on a 9 x 9  board. Guess who won ? I had beaten a korean nearly
professional.
And I was proud. Later he gave me 27 Handicap stones on a 19 x 19 board.

If we cannot on any board size with infinitely many handicap stones,
like this:

    A B C D E F G H J K L M N O P Q R S T
 19 . . . . . . . . . . . . . . . . . . . 19
 18 . . . . . . . . . . . . . . . . . . . 18
 17 . . X . X . X . X . X . X . X . X . . 17
 16 . . . X . . . . . X . . . . . X . . . 16
 15 . . X . X . X . X . X . X . X . X . . 15
 14 . . . . . . . . . . . . . . . . . . . 14
 13 . . X . X . X . X . X . X . X . X . . 13
 12 . . . . . . . . . . . . . . . . . . . 12
 11 . . X . X . X . X . X . X . X . X . . 11
 10 . . . X . . . . . X . . . . . X . . . 10
  9 . . X . X . X . X . X . X . X . X . .  9
  8 . . . . . . . . . . . . . . . . . . .  8
  7 . . X . X . X . X . X . X . X . X . .  7
  6 . . . . . . . . . . . . . . . . . . .  6
  5 . . X . X . X . X . X . X . X . X . .  5
  4 . . . X . . . . . X . . . . . X . . .  4
  3 . . X . X . X . X . X . X . X . X . .  3
  2 . . . . . . . . . . . . . . . . . . .  2
  1 . . . . . . . . . . . . . . . . . . .  1
    A B C D E F G H J K L M N O P Q R S T

I count 64 + 9 = 73  Handicap stones. Which could mean 73 kyu, at 50 %
winning chance 
reproducable against opponents with 1 dan performance.


please let a 9 dan professional try to play against this Handicap level
?
I believe that even only 64 Handicap stones, would firmly beat most
drunk shodans,
even in nearly total darkness, within 10 minutes, but that could give us
the reliable 
feedback, that if we could achieve 50 % with 64 Handicap stones 
against 100  different  1-dan amateur players,  each playing 10 games, 
then gnugo has finally reached the level of 64 Kyu. Which is my
assumption.

If this is the true, we have to confess it. 
I believe it is a reliable way to find out the real-life strength.

And then we could try to increase the strength with decreasing 1
Handicap stone,
every 3-6 month or so. 

So I suggest we increase the Handicap level to at least 29.

Then the board looks like this:

   A B C D E F G H J K L M N O P Q R S T
 19 . . . . . . . . . . . . . . . . . . . 19
 18 . . . . . . . . . . . . . . . . . . . 18
 17 . . X . . . X . . . . . X . . . X . . 17
 16 . . . X . . . . . X . . . . . X . . . 16
 15 . . . . . . . . . . . . . . . . . . . 15
 14 . . . . . . . . . . . . . . . . . . . 14
 13 . . X . . . X . . X . . X . . . X . . 13
 12 . . . . . . . . . . . . . . . . . . . 12
 11 . . . . . . . . . . . . . . . . . . . 11
 10 . . . X . . X . . X . . X . . X . . . 10
  9 . . . . . . . . . . . . . . . . . . .  9
  8 . . . . . . . . . . . . . . . . . . .  8
  7 . . X . . . X . . X . . X . . . X . .  7
  6 . . . . . . . . . . . . . . . . . . .  6
  5 . . . . . . . . . . . . . . . . . . .  5
  4 . . . X . . . . . X . . . . . X . . .  4
  3 . . X . . . X . . . . . X . . . X . .  3
  2 . . . . . . . . . . . . . . . . . . .  2
  1 . . . . . . . . . . . . . . . . . . .  1
    A B C D E F G H J K L M N O P Q R S T


What about letting some 1-dan players try this gnugo version ?

If this is no success, then gnugo is no 29 kyu. Even after 1000 games,
on the internet, against any dan ranked player, we have to win at least
50 %.

In order to establish a reliable kyu - rank.
It is no real achievement to play against ManyFaces or Handtalk, 
when they perform moves lightyears apart from go moves played by
experienced dan ranked go players.

We do want to beat a professional human, not ManyFaces or Handtalk ?

And we want to beat him or her ManyTimes ? Firmly ?

Yours Sincerely


Morten Gulbrandsen


Daniel Bump wrote:

> Ratings of Go programs are problematic. When a human plays a go
> program many times, the human learns about the opponent's weaknesses but
> the computer does not learn about the human's. So after some practice
> the human can beat the computer much more easily.
>