Re: Shudan Taikyoku

          Vain, idle speculations.  Obtain the program and hook it
     up to one or more of the Go Servers.  Produce a rank/ratings
     graph URL and send the results to the manufacturer (while
     also posting it here to the newsgroup).  There's no science
     in discussions only with presuppositions; without evidence.
     Though "server evaluation" is not a -definitive- means for
     estimating rank/ratings it appears to have sufficed for Michael
     Buro's computer Othello (Logistello) which would eventually
     defeat Japan's top "pro" players (at Othello).  The pro (who
     awarded a 1-dan blessing to a computer program) should
     be reminded of Official Computer Go Tournaments elsewhere,
     and their results.  Snivellers, who do not focus on truth and 
     justice, are not wasting merely their own time by posting here.
    
          Also:  verify that the (possible) "scam" is not a "man in
     the box" or "woman in the box" version of a Chess Machine.
     Once armed to the teeth with the relevant data, present this
     case to the Nihon Ki-In and ask them what they do for a living.




                      - regards
                      - jb


----------------------------------------------------------------------------
>> Michael Goetze <[EMAIL PROTECTED]> wrote:
>>> Well, if they mean Japanese amateur 1 dan, it might actually 
>>> perform close to that level when playing against people who 
>>> don't play against computers regularily. Note that Japanese 
>>> amateur 1 dan is, uhm, not very strong. ;)

> "Warp" <[EMAIL PROTECTED]> wrote:
>>  I think I have read somewhere that the strength "titles" given to 
>> these go programs are granted based on how well they perform 
>> against a pro  *on the very first game*, not on average. That is, 
>> the pro plays a handicap game against the program and solely 
>> by this one game classifies its strength.
>> Is this true?
>>
>> In my opinion it would be more fair if it was tested with several pros,
>> each one playing eg. 4 games against.   AFAIK it's a known fact that 
>> the strength of a go program magically drops
>> quite many stones the more you play against it.

"Steve Fawthrop" <[EMAIL PROTECTED]> wrote:
> Also, I do not know but I would speculate that if the program performs 
> well in the fuseki and early middle game but then resigns then it might 
> be rated too high based on a single game because it never got into the 
> complex late middle game.