[computer-go] use for Monte Carlo on 19X19?

[Gunnar Farnebäck]

Stefan Mertin wrote:
> on 07.11.2007 07:35 Gunnar Farnebäck wrote:
>> Stefan Mertin wrote:
>>> I am using GnuGo scoring in my tournaments.
>>> But GnuGo 3.7.10 mostly doesn´t score seki correctly,
>>> has this been revised for v3.7.11 ...?!
>> What scoring mode are you using?
>>
>> /Gunnar
> 
> SORRY - I was completely wrong here,
> I just realize that I was still using GnuGo 3.6
> for scoring instead of 3.7.10!
> 
> (for chinese :
> --score estimate --chinese-rules 
> for japanese:
> --score estimate)
> 
> In my actual computer Go tournament 
> I just set up SmartGo2.7 to play GoIntellect10,
> and from 150 9x9games there are 7 with a seki in the final position.
> 
> 5 of these were scored wrongly because I mistakingly had
> scored with GnuGo3.6 - I now have changed it to 3.7.11 and
> everything seams to be perfect! 

With GNU Go 3.6 you can try "--score aftermath" instead. It plays out 
the game until all stones except ones in seki are unconditionally 
settled, which usually gives a reliable seki detection. Also with 
current development versions "--score aftermath" is expected to be the 
most reliable scoring option, but the difference is much smaller nowadays.

/Gunnar