The idea is to give engines random positions from games(part of them easy win for one side) and decide to make a match of 2 games between the engines and decide that a game is basically a match of 2 games.
2 draws mean 0.5-0.5 result
2 wins in the same number of moves for the same color means also 0.5-0.5 result
In all other cases there iis a winner and if the same color won twice the winner is the engine who won faster.
Maybe programmers will start to care about winning faster in case of a list like that and even if not it will help me to decide which engine to use(I do not like the stupid analysis of stockfish in some winning positions and I am not sure if stockfish is the best in these conditions).
an idea to a different type of rating list
Moderator: Ras
-
Uri Blass
- Posts: 10800
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
-
jefk
- Posts: 955
- Joined: Sun Jul 25, 2010 10:07 pm
- Location: the Netherlands
- Full name: Jef Kaan
Re: an idea to a different type of rating list
could be done i guess, but instead of a 'random position from games' it probably
would work better with a carefully chosen set of positions, eg from ply 15-25 in
sharp or unbalanced lines.
Something else than TCEC (as alternative(, for a change...
After some test runs (and a lot of counting) special tournament software
for such a purpose (eg as extra in cutechess) would facilitate it all;
but then i wonder (cq it remains an open quesion) who would actually
would pursue with such an idea; you maybe
would work better with a carefully chosen set of positions, eg from ply 15-25 in
sharp or unbalanced lines.
Something else than TCEC (as alternative(, for a change...
After some test runs (and a lot of counting) special tournament software
for such a purpose (eg as extra in cutechess) would facilitate it all;
but then i wonder (cq it remains an open quesion) who would actually
would pursue with such an idea; you maybe