Re: [computer-go] Pattern matching - example play

[Darren Cook <darren@xxxxxxxxx>]

> BTW my pattern system does not work for 9x9.
> 9x9 and 19x19 are so vastly different at the opening stage that it simply
> doesn't scale down well.
> I have never even tried it but I can't imagine that it will perform well.
> Let alone that I trust the very sparse 9x9 "Tengen" statistics for 19x19.
One of my favourite subjects, and my opinion is that 9x9 has all that is 
hard about 19x19 go. Meaning life and death. I'm sure this pin won't burst 
your bubble Frank, but you've (IMHO) been working on the easy stuff, as 
computer programs have been playing the early stages of the game competently 
for years (that was Mark Boon's point a few days ago).

Reasons:
 1. Strong 19x19 players beat weaker players at 9x9, consistently, at all 
ranks. Ditto for computer players. (There is a bit more noise, as one L&D 
mistake in the first 15 moves can screw up a game for you, whereas in 19x19 
you get more "lives").

 2. High-dan pro players beat low-dan pro players at 9x9. I don't have 
enough data to satisfy Frank, but it surprised me as I thought a 1p player 
basically had mastery of shape, life and death, tesuji, semeai and the 
endgame and that the difference between 1p and 9p was more about subtle 
positional plays.

 3. Computer programs lose games due to life and death more than anything else.

Darren

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