The funny thing about MC scoring is that if white DOES win,  then it
doesn't matter what black plays, everything loses!  

That would mean that in a scalability study,   where colors are
alternated,  really strong versions of Mogo would score poorly against
the weaker programs. 

In a serious competition you would want to throttle down the playing
strength (when playing black) so that it could win more and not just
quit (resign) out of frustration!     

I wonder if we are seeing just enough of this that it's causing mogo to
get worse results as it gets stronger?     It seems unlikely - but when
a position is in reality losing, and the score drops,  mogo tries lines
of play that in theory maximize it's chances of still winning
anyway.     I wonder if these "desperate" moves are actually hurting
it's chances of winning?

You will see something like this happen with really simple MC programs
that do not have tree's.  You will see the score drop and it will play a
really weak move that wins if the opponent completely overlooks a simply
threat.     What MC doesn't understand is what good players know,  that 
risk has to be managed carefully - you can't just lash out like a
wounded animal.    Often "normal" play will give the opponent some
chance to make an error you can take advantage of.  


- Don




David Fotland wrote:
> If Don's data shows that after E5 C5, white will win, then if mogo is good
> enough, it will see that it can't win, so it will play it's second move to
> create complications, which C6 does.  A "stronger" move may just lead to a
> certain, small, loss.
>
> David
>
>   
>> -----Original Message-----
>> From: [EMAIL PROTECTED] [mailto:computer-go-
>> [EMAIL PROTECTED] On Behalf Of Olivier Teytaud
>> Sent: Tuesday, February 12, 2008 10:09 AM
>> To: computer-go
>> Subject: Re: [computer-go] Should 9x9 komi be 8.0 ?
>>
>>     
>>>     46 E5 C5
>>>       
>> By the way, mogoRelease with moderate computation times often plays
>> E5-C5-C6, and this third move C6 is very weak (at least I've
>> been told so :-) ).
>>
>> _______________________________________________
>> computer-go mailing list
>> [email protected]
>> http://www.computer-go.org/mailman/listinfo/computer-go/
>>     
>
> _______________________________________________
> computer-go mailing list
> [email protected]
> http://www.computer-go.org/mailman/listinfo/computer-go/
>
>   
_______________________________________________
computer-go mailing list
[email protected]
http://www.computer-go.org/mailman/listinfo/computer-go/

Reply via email to