can magnus carlson cross the 3000 elo barrier

Elo states that inflation is possible, but hasn't happened. 

...in 1988.  The curve upward began in 1986, so was not detectable yet.  The point is the second paragraph of his letter.  The Elo ratings system is completely relative, does not reflect actual playing strength, and can be influenced any number of ways.

I read Regan's paper, but his addressing of the effects the system being open instead of closed are superficially addressed and discarded.  I liked the "now that we've proven there's no inflation, here's my conclusions" when he did nothing of the sort.  Like a lot of statisticians, he seems to want to pretend that chess ratings are like some clean room experimental results data.

Nobody who has claimed inflation has happened has been able to state what that rate of inflation is. 

That's funny, I could have sworn I had just posted a link to a study showing the estimated rates of inflation down to the individual ratings point, from 1986 to 2006...

Well, maybe that's considered "nobody" becuase it isn't Regan, or published on Chessbase?

I decided to quote this in its entirety since you have just retroedited two different posts. You added the link to the study and then added the comment in the post I quoted saying "That's funny"...

No...I added the study link immediately after I posted it the first time, before your post.  I always do that, get the link right afterwards and go back and put it in.  Feel free to stalk me and find this out over time if you are worried about it ;)...

The "that's funny" addition was just a legitimate afterthought, and yes, I edited my post and added it.  Also not an uncommon occurence; it prevents me from posting one-liners and "and another thing..." posts, which makes the thread as a whole easier to read.

Rod Edwards argues that there was rating inflation of 24 points from 1986-1988 (28 points by January 1st, 1989). Arpad Elo states that there is no discernible drift or inflation from 1970-1988 (he wrote his letter in November, 1988). 

Rod Edwards does not use any traditional method to determine inflation, rather he cherry-picks the ratings of the players rated 11-50 on the rating list then extrapolates that out to the entire population.

You stated that the rating inflation was not large enough for Elo to have noticed in 1988. A rating change of ~25 points in 2 years would be a huge anomaly that any statistician would notice. Elo was tracking the entire rating population, not just 40 players.

There is an explanation to why ratings of the top 50 players increased after 1986: Glasnost. After 1986, Soviet players were allowed to compete internationally. There was an influx of really strong players. In January 1985, there were 11 players from the USSR ranked 11-50. In January 1988, there were 15, including Boris Gulko who was playing for the US.

The quality of tournament chess players exploded thanks to Glasnost!

Elo's "the ratings are stable" assertion refers to a FIDE memo he wrote in 1986, not 1988, so...he would not have seen anything.  He also refers to losing access to the "entire rating population" data, and it's fairly clear that he was not even monitoring this.

Elo's ratings system was more a fire-and-forget thing.  What FIDE has done with it since was out of his influence.  What he did know from outside the situation, he stated:  the ratings should not fluctuate much unless allowed to drift, and that "drift" was entirely possible, since the ratings system is completely relative.

Yes, this is one of the "inflationary one-offs" I referred to.  It explains the immediate jump from '86 to '88, which is much steeper than the rest of the inflationary climb, but not anything else.

Glasnost was a process, not a one-off. Glasnost lasted for years. Chess saw a steady influx of new players through the early 1990s. For evidence of this, just take a look at what happened to the Olympiad after the collapse of the USSR.

...a process that would have started with a bang and tailed off over time...not a "steady influx".  

By January 1990, there were 21 players from the Soviet Union (again, including Boris Gulko), ranked 11-50.

Edited to add my source:

http://fidelists.blogspot.com/2008/03/january-1990-fide-rating-list.html

You're just helping explain the data in that study ;).  So, 24 points increase from '86 to '88, 16 point increase from '88 to '90, then about 2-3 ratings points/year, very slowly decreasing in recent years to 1-2 points/year.

Looks like a tailing off duck to me...by the early 90s the Soviet "boom" effect on chess ratings was over.

Btickler, Rod Edwards cherry picked his data by only analysing the players ranked 11-50. The general population did not see a 25 point rating increase from 1986-1988. 

There is an explanation as to why the players ranked 11-50 increased in rating, there was an influx of strong players. 

That is not rating inflation, that is accurately measuring an influx of new, strong players.

Smyslovfan, I just stated that it was a one-off effect, so I am not saying that a 25 point increase was caused by some overall ratings inflation process.  If there were better published numbers for the overall pool that clearly factored in all events and adjustments, that would be great for proving the point beyond a doubt, but there aren't, and I don't expect it to be in FIDE's interests to lay out such a timeline/history.  Indirect data, approximations, and examining subsets are the only way to scrutinize this...circumstantial evidence.

Anyway, I doubt we are going to reach any agreement on this given that we are arguing about the contents of a black box that will not be opened, but I will say that I respect and appreciate you discussing this with some research "vigor" rather than saying "this is abstract and arbitrary and who can say one way or the other?" ;)

P.S. If you ever want to come at the ratings/leaderboard stuff from a different angle than just chess, try Wharton Business School's Gamification class...it's available online free through Coursera.

Btickler, the FIDE ratings lists are published and available for anyone who has the time, desire and mathematical background to peruse it.

That work has been performed by Elo who stated that he had not seen a drift in 18 years, not 16 years (1970-1988, not 1970-1986), and by Kenneth Regan. 

People who take a small section of players, such as the #1 player, #100 player, or those ranked 11-50, are not measuring general inflation. There are many explanations to why ratings have increased at the top. The most obvious, and simplest (Hi, Occam!) is that chess players have improved over the years. 

Chess historians and players, from Reti and Euwe to Kasparov and Nakamura, have all acknowledged that chess skill has progressed over time. They have shown how players today are standing on the shoulders of giants. Ratings improve as players improve.

While the FIDE ratings list is not set against absolutes, it is stable. 

The argument in favor of ratings inflation always feels like an attempt to denigrate the accomplishments of Kasparov, Kramnik and Carlsen. The person who benefits from these arguments? Bobby Fischer. It always seems to come back to arguing that he was better than any player since. 

Fischer was amazing, and it has taken decades for the rest of the chess world to catch up. Even now, his best rating (2785) would tie him for fourth overall on the FIDE lists. 

That statistic alone should be proof enough that ratings have not been inflated. From 1972-2014, Fischer's rating has been good enough to be among the top players in the world.

I am not arguing that overall distributed chess skill  among the player pool does not also increase over time.  I am no stranger to the razor inspired by William of Ockham.  However, that does not mean that all of the ratings climb can necessarily be explained by it.  

The razor cuts both ways.  As I said, it's in FIDE's and the chess world's interests to see ratings go up over time, and while I am not saying this means there is a concrete conspiracy behind a controlled ratings increase, it is highly likely that evolutionary adjustments/policies have been individually created over the decades with a kinder hand towards ratings going up for players rather than down.  It's human nature.  The right hand does not know what the left is doing, but everyone is generally happier.  Ratings slowly go up, and press releases go out...it would be harder and more complex to keep the ratings pool pristine and level than it would to not really pay attention and let it drift slightly.  Downward drift would cause frowny faces and be corrected, reasonable upward drift would be left as-is even if identified.

Occam's Razor.

"this is abstract and arbitrary and who can say one way or the other?" ;)

You don't like it, but I think it was a pretty decent point to make, honestly. It's not that the general ideas you brought up (I referred to them earlier) don't work, but that you are jumping to conclusions with them more quickly and confidently than is probably justified.

Even with the Regan stuff, I still have some scepticism in me (see, I'm applying my own annoying comments you don't like to my position as well!). I tend to believe him, but I am aware that things can be misleading -- just as an example, there are so many different ways to choose engine data -- what move do you start from, when do you end, how important is the margin between the given move and the engine move, etc. I lean towards inflation being false, but I am still fairly far from certain.

There are other reasons for my belief too such as the general feeling that the quality of play is stronger (for example the kind of precision in the endgame it takes to win games now feels almost machine like) -- but note how I hadn't brought it up, because I know the fallibility of these subjective feelings.

It's definetly doable but he has to work really hard if he wants to achieve a 3000 elo rating. In Shamkir Chess Tournament he lost to Fabiano Caruana. Those small losses will work against his achievements. But I'm with him all the way. 

Engines are good, don't get me wrong. But they aren't the giant beasts that everyone think's they are. They can beat most players, but most players over 2700 I think would match and beat alot of engines out there because of the difference in how humans and computers function. Plus you can't play against computers and get FIDE points, only can play against human players.

if he plays for 10 more years he will be 

Magnus won today's Gashimov Memorial with +3. He  won the tournament by a full point. He gained 1/10th of a rating point. 

3000 may be mathematically possible, but it's not humanly possible. 2900 is incredible enough, and I do think he will break that mark.