Re: ranking system modification

On Thu, Oct 09, 2003 at 05:40:56AM -0400, Jason Short wrote:
> Horn G=E1bor wrote:
> > Hi!
> >=20
> > W/ the current ranking if there's a draw at the end of the game (like i=
> n
> > #229419, u can check the log at
> > http://civserver.freeciv.org/viewgame.phtml?game=3D229419) the player i=
> n
> > with the higher original ranking (Iuz) loose some points, and the other
> > winner gets some (Donald D. Dumbsfeld). Is it only me who is surprised
> > about that? I (and i guess these 2 players too) thought if it will be
> > draw there's no ranking change relatively to each other).
> > If u agree i fix it, and leave the difference of the rankings of the
> > winners unchanged (their absolute value will increase a bit because of
> > the AI looser, but that is very small and is ok i think) in case of a
> > draw.
> 
> I think there is a distinction between a draw and a tie.
> 
> If two players agree to have the match a draw, there should be no=20
> adjustment of scores.
> 
> But if two players battle to a standstill (e.g., a tie), then the=20
> lower-ranked player has shown he can compete with the higher-ranked one=20
> and deserves some point reward.

This is the idea behind the ranking.  If we are going to be changing the
system, please everyone read up on Elo.  

If I have a lower score, my chances of beating you are much lower than
you beating me.  You SHOULD beat me.  Therefore a true draw means that
I have succeeded and you have failed, albeit in a smaller way than an
all out victory.

-- 
Paul Zastoupil