Where? Do you mean http://www.chessbase.com/newsdetail.asp?newsid=7621 ?

Sorry, but not convincing at all. They were only looking at evaluations obtained by crippling the engines even more (from depth 2 to 12). Good for blunder checking, but not to compare the strength of super-GMs (unless one is ready to assume that super-GMs are only characterized by their rate of blunders, which is quite absurd). The very fact that they reach the same conclusion with a depth as low as depth 2 shows that their study is irremediably shallow and flawed.

Well said. I should also point out that there's some quite questionable cut-off in the methodology used by Ken Regan towards the end of the games. The analysis of each game is simply stopped when the evaluation given by the crippled Rybka reaches +3 or -3, and what is worst, the result of the game (1-0, 0-1, 1/2-1/2) is entirely ignored in the study (?!). How many games would be incorrectly evaluated with such a method? For instance a game ending in a fortress would be likely evaluated as a failure for the defender, which is quite absurd.

Regarding the challenger in 1972: maybe the entire concept of inflation stems from the idea that nobody could possibly top Fischer, so any number that puts someone else higher must be subjected to some correction that would get rid of this flaw.

I want real proof instead of somebody who heard somebody else say something.

yes, i see what you mean except that besides pure chess ability you forgot to factor in, opening theory of today compared with that almost half a century ago

"Caruana and Karjakin ... their careers have only begun."
Yep, that's my point - but the system or rather the way it's used to compare ratings from different periods suggests that they are _already_ as strong or stronger than Fischer and Karpov at their career peaks.

"There certainly is no mechanism that would cause the ratings to increase in time."
True (though see below), but - my other point - for this we have to look at the average rating of ALL players, not just at #1, top10 or top100.
There are two mechanisms, though, which might cause average ratings to increase in time (again considering the whole list):
- K-factors: You start with a high K-factor when your playing strength might (rapidly) improve, and get and keep a lower K-factor when you are eventually declining. For example, if a 'fresh' rising star beats a GM, he gains more points (K-factor 30 or 15) than the GM loses (K-factor 10).
- Players' habits: A young improving player might play a lot (and will be underrated while he keeps improving due to inertia of the system). But an old declining player, at least nowadays, plays less frequently, hence he might be overrated for many years (his Elo partly reflects his former but not his current strength).

And such effects indirectly penetrate through the entire system: I might play a 2200ish player who faced an IM who faced a weak GM who faced a strong GM who played against Carlsen.

Are Caruana and Karjakin´s opponents really stronger?? Untill the 80s, chess was one of the most popular activities in former Soviet Union, its top players were celebrities.It was like it is today in Armenia, with the small diference that Soviet Union (if I am not wrong ) had 250-300 million people and Armenia 3million... . Imagine how many strong players a 250-300 million people Armenia could produce !!! Keres, Petrossian, Smyslov, Stein, Bronstein, Korchnoi, Tahl, Geller, Spassky etc, were not inferior. For me, only Kramnik, Anand ( the young Vishy, not this one), Ivanchuk, Aronian and Carsen could be elite players in Soviet Union. Imagine in the entire world....

If you still read this ... that part of my comment may have been misleading, here I had professional top players in mind. For them, the situation may have become more equal: all have the same (engine) resources, all can spend lots of time on chess including opening theory. In the days of Karpov and Kasparov, strong seconds mattered more than they do nowadays!?

I guess for you it's impossible to follow opening theory not because you are (just) a FM, but presumably because you also have a job, maybe a family, maybe other hobbies etc. . For young and ambitious FMs (say, 14 or even 22 years old), it may be possible to compete in that respect with most GMs?

Anyway, I think it won't affect _average_ ratings and doesn't cause rating inflation. A win against an equally strong player is worth 5 points (and a loss -5 points) no matter if you play a sharp Najdorf, a non-theoretical setup or something dodgy such as the Blackmar-Diemer gambit!
That's a common misperception: the Elo system doesn't measure 'absolute' quality of the game in any phase (opening, middle- or endgame); it does measure results i.e. relative quality compared to your opponents.

Also, Anand's current 2771 is after playing little and losing points every time lately. He has performed considerably below 2750 the last years, but it takes a while for the rating to drop if you are playing as little as he does.

Anand reached 2797 in 2001 (higher than when he won Linares and World Championship in 2007) but slightly lower the year before after a couple of bad events.