With the rating floor dropping (originally 2200, then 2000 for a while, now 1000) there will naturally be rating deflation as the weaker players improve, taking rating points from the stronger players. This was not a problem when the floor was 2000 as far fewer players were improving at that rate. The deflation started at lower levels (below 2000), but now it seems to have spread to the top level. You can view the past top players lists here- http://ratings.fide.com/toplist.phtml

I made a graph of the rating of the #100 player for the last 3 years. Here it is.

This seems to indicate rating deflation around the 2650 mark, and at the 2700 mark inflation has stopped. I expect we will see deflation at the 2700 mark soon.

There wasn't a problem with inflation (in fact it was good, leading to things like Carlsen's record rating getting attention) But we will only be getting more lower-rated players, and thus deflation. Now this will be very bad for chess when, for example, Carlsen keeps improving but stops going up, and chess players appear to be getting weaker.

So do you think FIDE should do something about this before it's too late? If nothing is done I expect there to be less than ten 2700+ players in ten years time. Which would not be good.

It's only a problem because people simply cannot comprehend the simple fact that elo rating does not directly correspond to playing strength.

Indeed. But it's the best indicator there is. I think the problem is comparing ratings from different time peroids, like comparing Bobby Fischer's 2785 to 2785 today.

It's not possible to compare objectively, only speculate.  Although as engines get even stronger more objective analysis of playing strength will be possible, but even now the strongest engines can't really accurately evaluate alot of say, Carlsen's moves, which proves that you can't accurately measure the playing strength of very high level players purely through engine analysis.  It's hard to say whether or not Fischer's play when he reached 2785 would have earned him a higher rating today, I'm inclined to think it would have based on my assessment of his games vs the games of current superGMs of similar rating, but obviously that is still only a subjective judgement.

How fide know my rating

Actually, ELO rating tracks playing strength extremely well. In fact, it is the best predictor of the result. Why would you think otherwise?

That is not the point I think. It's more like ''you get a black belt if you can perform action X.'' In chess, there is no objective way to test playing strength. For example, if you took tactical puzzles, then someone like John Nunn could be world champion or something. 

It only measures strength relative to the pool of players within the same elo group.  Yes it is a good predictor of a result, but there is no way of measuring, for example, whether or not Bobby Fischer's 2785 is equivalent to the same elo rating today, because the pool of players is different.

I think it is kind of silly to have lowered the FIDE rating to a 1000 floor. In my opinion FIDE rating floor should be 2000 when players who achieve a FIDE rating are most likely serious chess players who are trying to become chess professionals and international tournament players. National ratings can take care of the lower ranked players and those who might be fairly strong players but have no desire to become professionals and play internationally.