Comment by Mr_Minderbinder
10 days ago
> In terms of strength he's the weakest player to win in half a century even in absolute terms.
Gukesh is arguably stronger than either of Khalifman, Kasimdzhanov and Ponomariov, who won the FIDE title before it was re-unified. Also his current rating is higher than either Karpov’s or Kasparov’s were when they first won the title. His rating when he first won was about the same as Fischer’s when Fischer first won. Neither Kramnik or Anand were clearly the best player throughout the entirety of their reigns and both of their ranks fluctuated amongst the top ten positions.
> Also his current rating is higher than either Karpov’s or Kasparov’s were when they first won the title. His rating when he first won was about the same as Fischer’s when Fischer first won.
This doesn't really mean anything. Rating is a purely relative system, as in the other thing that matters when performing Elo calculations is the difference in Elo between the two players. The absolute value of an Elo rating carries no real meaning and drifts over time based on the volume, skill level, and initial rating of lower level players. Since these change frequently, it's pretty much useless to compare ratings separated in time by more than a decade or so, maybe less. 50+ years is certainly far too long.
My views on this, which are mature and have been held for many years now, are mostly informed by the results obtained by Kenneth Regan and Guy Haworth in their paper “Intrinsic Chess Ratings” which, unless you have intelligence to the contrary, is the only rigorous treatment of this issue that has yet been performed and is yet the only argument that has any persuasive hold over me.
You say that ratings drift over time to such an extent that to use them in comparisons across long time spans is meaningless yet their analysis determined that chess ratings as a measure of intrinsic quality of move choice (which must be highly correlated with playing strength) is stable over several decades with only some indications that a small amount of deflation has occurred.
Your argument in comparison amounts to informal speculation. If I were to share my own I would say that those potentially error-inducting considerations, are statistically insignificant compared to the sheer number of games, that is to say corrective and informative exchanges of points, that occur. Further, I would add that the absolute values of ratings were defined by the playing strengths of the original players and that this definition has been well preserved even as the player pool has evolved.
I have heard many such arguments in my time yet not a single proponent cares to demonstrate them. What I find amusing is that those same proponents will often readily accept a comparison across time of a single player (often themselves) across similar time spans without controversy, as evidence of their progress as a player, for instance using Carlsen’s rating today and comparing it with one from early in his career, say from 2003 or 2004, which at this point was more than 20 years ago.