[computer-go] A new pairing system idea for CGOS

[Łukasz Lew]

On 10/5/06, Christoph Birk <birk at ociw.edu> wrote:
> On Thu, 5 Oct 2006, Don Dailey wrote:
>
> > While thinking about the problem of a fair pairing system for CGOS, I
> > think I have discovered an interesting alternative system for fair
> > pairings ...   turn CGOS into a tournament server!
> > ...
>
> I think this is way too complicated. Also the idea of scoring
> games that got a 'bye' is really bad. There is no need on a server
> like CGOS to use a "shortcut" or beeing "fair".
>
> I much prefer you last suggestion: pair from the top down.
> KISS :-)

Top down is good, but with a fixed length tournament and fixed player pool
there would be some anomalies with a deterministic algorithm.

I vote for a pairing top down with a bit of randomness for opponent choice.
Don's proposition - Choose best form random N is one option.
(N should be a constant fraction of unpaired bots)

The another is to compute explicit probabilities for each opponent by
Boltzman distribution.

I vote for best from N.
KISS :)


>
> Christoph
> _______________________________________________
> computer-go mailing list
> computer-go at computer-go.org
> http://www.computer-go.org/mailman/listinfo/computer-go/
>