I get your point, I think, but haven't a lot of people here used rank and straightforward measures as a way of estimating the competition that a player has faced? By an objective algorithm, the Guardian one isn't wrong, then. And we here have used the same measure.
I have doublechecked. Their method of calculating the difficulty is definitely relative. Even when you give the same set of seven players,
their method will assess different difficulty level for that GS depending on whether Roger or Rafa is facing them. Here is a
to the wiki page of Elo rating system . As you can see the prediction (the chances) of
who will win depends on the
difference between the Elo ratings of the two players involved. These chances play a crucial role in
determining the difficulty level of a GS win in their method. It is very clearly and definitely not an objective scheme.
To give you analogy in terms of rankings, it is like calculating the average difference in ranking between you and your opponent, instead of merely calculating the average ranking of the opponent. If your own ranking is bad, the same seven opponents will then yield a higher difficulty level.
Let me know whether you agree or need further explanation.
I don't have a problem with them using Elo ratings. You could just think of it as another alternative to ranking scheme and see how it looks under that system. However, they should base the difficulty level on the average Elo rating of the opponents faced. The moment they get into the difference in the Elo ratings, that involves your own Elo rating and hence is not an objective measure of the difficulty any more, but a relative measure of difficulty with respect to you.
