There are no outside influences on a game of chess so the winner is ALWAYS
the
player that played the best during the game in question. The repective sums
of
differences for 'perfect' play will always be in the favor of the player that
won the game. If the losing player outplays the winning player and then
blunders into a mate in one, the negative value of blundering into the mate
in
one must outweigh all of the errors that were made earlier by the winner.


and based on that game, what's your estimate
of the outcome of the next games of these
players ?

Even if I would agree to what you wrote,
there is still a use for rating the moves:

suppose two games with the same result,
once both play pretty well, once both play
very bad. There is reason to assume that
in future games the bad players will
score worse, and I assume, that this can be
verified statistically.

Guenter.