hgm wrote:
Why would that only apply to humans? It seems a cheap way for engines to gain a few rating points too. When you are losing, miss the time control by 0.1 sec, and hope for a replay...
Well, THAT is an interesting point which might gain more elo to the Littos than years of compiling different Tab sizes in the source code ... !
time lost stat from IPON is interesting for myself because I have around the same rate with around the same games (161.000 IPON games, 160.000 SWCR games).
Not directly to compare because I tested Ktulu and Ktulu produced a lot of such games. I had 389 games lost on time games since I started with SWCR. 311 games are replayed, other time lost games are in clear end position and not replayed but with the right results in SWCR database.
From my view, I never have problems with Critter. No lost on time games here. Perhaps it have to to with the time control I used. I support what HGM wrote here, lost on time games should be replayed for many reasons.
SWCR database is available for download. Furthermore, it make not sense to have bad games in the database for people which like to replay games. Furthermore, lost on time games are bad for statistics.
More interesting as lost on time games are games ended with 1:0 or 0:1 in remis position with resign = on. Over 260 games here, most of the games in bishops endgames. Such games I don't have in SWCR database because I am playing without resign and this games ended with the 50-moves rules.
Well, THAT is an interesting point which might gain more elo to the Littos than years of compiling different Tab sizes in the source code ... !