RE: [Bug-gnubg] Re: Strange FIBS ratings

[kvandoel]

> > 1862 +- 23 (actual 1700).

> > I have no definite explanation for why it is so far off your 1700 rating,

> At least you understand the origin of my concern now.

Yes, maybe you're weird that you are not modelled correctly or maybe I'm
weird that I am.

> > What is needed  now is more analysis of games  by players with known
> > ratings for validation/refutation. I am  willing to do the work if I
> > get the data to work on.

> I'll play more and send you my matches as they come. In any case, thanks
> for looking into this.

That's  good, but what's  really needed  is a  largish set  of players's
matches with  known ratings to  be tested. My  informal tests on  my own
opponents came out quite good as  I showed before.  My working theory is
you're an anomaly :-)

If such human  studies show my current formula is flawed  I am sure such
(human)  data can also  be fitted  in a  similar fashion  as I  wrote on
http://www.cs.ubc.ca/~kvdoel/tmp/ratings,   which  would   lead   to  an
adjustment of the current numbers in GNUBG's rating estimator.


Kees
--
PS I don't  know what Mr. Zare has been arguing  about my studies behind
my  back on that  closed private  forum without  the common  courtesy of
informing me of  such arguments, or what led him to  label my studies as
"quite  flawed".  Mr.   Zare's style  of "publication"  makes  a factual
argument impossible.   IMO this  whole argument can  only be  settled by
experiment anyways,  not by a-priory mathematical theory,  no matter how
stylishly presented.