Re: [gnugo-devel] arend_1_25.3: less (??) experimental influence

[Arend Bayer]

> Currently W has won 8 out of 9 games:
Hmmm, looking good. However, if you want to see the weakness of
--experimental-influence, try playing a 9 stone handicap game with
GNU Go as white. During fuseki, it will tenuki at any moment it possibly
can, ignoring peeps etc. 

> I suppose I'll run it about another day then collect the
> games. We can get some information from individual games
> by replaying them to find out when the two versions
> differ on particular moves.
I have tried that with two NNGS games so far. The experimental version
had a clear plus; somewhat to my surprise, the biggest gain came from 
improvements in early endgame.

Arend