[Computer-go] May KGS bot tournament: 19x19, fast

Don Dailey dailey.don at gmail.com
Mon May 9 05:03:09 PDT 2011


Different tournaments have different characteristics.   Swiss is excellent
at resolving the stronger and weaker players but it's not so good as a
general ranking procedure.

The reason I built a simulator long ago was that I wanted to find a way to
determine who the best player is from a pool of players,  in order to
improve my chess and my go program using learning algorithms.    The problem
is how to do this with the least amount of effort.

If you have many players and want to find the best and their ratings are
unknown,  there are probably statistical methods to do this most efficiently
based on finding the pairing with the most information return.   But a
SIMPLE way I found was to play a knockout tournament where the number of
games (per pairing) increased with the round.     In my simulations I could
find the best player a high percentage of the time with relatively few
games.    Round robin is orders of magnitude more expensive for this if the
number of rounds is very high.

The knockout tournament is a "divide and conquer" algorithm, so it should
not be surprising that ti's efficient.

If the goal for a tournament is to (relatively) reliably find the best
player with the fewest number of games,  it should be a knockout tournament
where you play more and more games as you advance.      But this is a
horrible choice for most tournaments, especially for human tournaments,
 because if you lose you go home.   Not much fun :-)

Don



On Mon, May 9, 2011 at 7:44 AM, Petri Pitkanen
<petri.t.pitkanen at gmail.com>wrote:

>
>
> 2011/5/9 Nick Wedd <nick at maproom.co.uk>
>
>> You are proposing that the tournament should start by pairing strong
>> players with weak players, and claiming that this is more likely to result
>> in the strongest player winning the tournament.  I don't see it.
>>
>>
> Maybe  itis  easier if you think Swiss system as a cup. Which it for
> winners  (assuming number of rounds being like 5 rounds for 32 players). If
> you pair strong players with strong ones, on second round you end having
> players still contending of winning that are both weak and strong. And some
> strong players dropped to competing for third place effectively.
>
> Swiss system is a cup with kinda consolation being played by the losers.
>
> And you have reliable a priori information then swiss is not not the best
> choice. MacMahon gets better results. Not fair for quickly advancinf players
> but nothing is perfect.
>
> You could simulate this easily by assuming that win probabilities follow
> exactly ELO estimates for example.
>
> Petri
>
> _______________________________________________
> Computer-go mailing list
> Computer-go at dvandva.org
> http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://computer-go.org/pipermail/computer-go/attachments/20110509/95b7cb40/attachment.html>


More information about the Computer-go mailing list