diogenysos: Well, it *should* be a 'disadvantage'. That's the entire point of a rating system; if a higher rated player plays a lower rated player, his expected score is more than 50%. So, if the match ends 1:1, the higher rated player is expected to lose.

Your suggestion could lead to a 2000 rated player playing a 1000 rated player in a 2-game match, where the 2000 rated player winning the first game 26:22, losing the second game 25:23, and still getting his rating adjusted as a win.

That's not to say that I don't think the rating adjustment could not be improved. Currently, the potential of ratings isn't fully taken advantage of. Regardless of the length of a match, for rating adjustment, it's considered a win, a draw, or a loss. But ratings (at least, ELO ratings and their derivatives) predict an outcome. For instance, that the higher rated player ought to win a 10 game match by 7:3. If the higher rated player wins by 6:4, the higher rated player would actually lose points, while the lower rated player wins rating points, despite losing the match (because he did better than he should do according to his ratings).