Why have ratings gotten inflated?

Every time I read a thread about the awesome performance of Magnus Carlsen, I read a reply that we must take into account the ratings inflation over the last 30+ years. Fischer topped out at 2700+, but there are many at that level today.

The question I have is: WHY? What has happened over the decades to inflate ratings to the point that it's meaningless to compare Fischer to Carlsen, Anand to Petrosian, etc.?

Back in the days of Fischer there were a few issues that come into play...

a) Not all of the strong Soviet players were able to travel.  There were a large number of extremely strong players who didn't even have FIDE ratings because they couldn't leave the Soviet Union.

b) Prior to the last handful of years top flight players typically played in tournaments in which they would have to play against several opponents who were a few hundred points lower rated.  Today the top players play only in closed round robins so that they tend to keep their rating points circulating amongst themselves.   When was the last time Magnus played someone rated 400 points lower than himself? 

I hear that argument a lot that his rating is protected by playing only Super GMs. We see what happens when he plays someone even 200 points lower. 400 points lower and the only question is when he wins.

Well in my tiny opinion, those grandmasters from the past didn't have super chess playing computers, so today's grandmasters are probably way better than those of the past.

When an unrated player enters a tournament, calculations are based on an assumed entry level, which I believe is 1200. One way inflation can happen is if someone begins his career, loses a few games, and then quits forever. Since someone with an initial bad experience is more likely to drop out than someone with a good experience, this has a net effect on total number of points in play throughout FIDE.

Imagine 1,000 new players begin in January. half of them are pretty good, and half of them are pretty bad. They play a number of games for a few months, and 500 of them get discouraged by their performance and drop out of play. But before they drop out, they have given thousands of rating points to the other players. Since they leave with fewer points than where they started, there are now more points within the system, so the average score per player goes up. The effect is tiny, but it is accumulative over time.

The opposite effect happens when a very high level player retires. They take a large number of points out of the system, which slightly deflates ratings. But since people are more likely to stop playing after doing poorly, it is not enough to keep things in balance.

The other factor is the total size of the ratings pool. The larger the pool, the higher some ratings must be. Imagine you were the best player in the world, but that there was only one other player. You could never get higher than a 1400 rating, no matter how good you were. There would be no one you could take any more points from. With three total players, you could potentially get to 1600, and with four 1800. -- All assuming you never lost a game, of course.

it could be that advances in the sophistication and availability of chess engines have improved play at the highest levels. it is not inflation. it is reality. 

Computers cannot inflate ratings overall. Though they might increase any given player's rating, any gains he makes is at the expense of another player, so average ratings do not change. 

The only way for ratings to inflate or deflate is by changing the average rating per player. And the only ways to do that is to change the total number of players or the total number of ratings points in the collective pool. 

Magnus Carlsen is dominating the chess world beating super GM's on a regular basis.  If he's beating 2700 players regularly then what do you think he would do to players 400 pts lower than him?  I don't think this has anything to do with inflated ratings.  Personally I think trysts made a good point.  With the age of super chess computers I think players are more advanced and better prepared.  I think an even bigger reason is the larger amount of stronger players out there.  There are so many more master level players and grandmasters that the ratings have inflated to seperate the upper echelon of GM's.

Actually what has probably become inflated is the ego of people not even remotely being able to understand grandmaster moves in the past, now with the chess engine by their side they proudly claim to know when a master makes a mistake immediately after they move

Yeah I've seen that. People were saying that Nakamura made a patzer error when he simplified into a winning endgame against Hou, because the machine said the evaluation went down by 2.0 or whatever.

This is just statistics I think.

If you're in bad shape, you give away points to people who are much lower rated.

Also talented players can gobble up rating points from people rated much higher, creating an artificial rating jump (and loss for the opponent.)

So just as long as you have a massive field in which people spill points, the rating is going to become inflated over time. People become GM's by beating relatively weak GM's, and these are a way for even more people to beat GM's.

JamieKowalski is correct. Computers have nothing to do with it, since everyone has access to them.

How neat it could have been to have the Internet, but without the chess engine. The game analysis section of the forums would have had much more lively debates as to whether or not a player took the right path in their game. But now a master could put their game in the analysis section, and an 1100 rated player can point out the mistakes

The more players the more points in the system - therefore higher ratings. 
Anyways ratings are still only good or bad when compared to another rating. 

I definitely agree with this!

The statistician, Kenneth Regan, argues that ratings are not inflated at all. In fact, there's been a very slight rating deflation over time!  He has shown that playing levels have risen considerably in the last 40 years. This makes perfect sense to anyone who isn't thinking of ways to make "Bobby  Fischer was the all-time greatest player ever, no excpetions possible cos he was great when I learned how to play" sorts of arguments. 

It's really strange to say this, but Nakamura is, move for move, about as good as Fischer was in his prime. Nakamura overtook Fischer's rating at the Olympiad in 2012. I'd take Fischer in his prime over Nakamura in a match only because his will to win was so great. But Nakamura is, move for move, the greatest American chess player ever.

Kinda depressing, isn't it?

Nakamura greater than Fischer ?!   What a joke .   LOL 

Yes, but take their engines away and they are still just hacks .  Nothing irritates me more than to see hacks arguing with IMs and GMs on ICC when famous games are being broadcast and analyzed simply because they have a strong engine running .  To avoid this I now set my kibitz to 2 so that I only see the comments of titled players . 

I totally agree.  Fischer dominated chess during his time.  Nakamura is nowhere near as dominant.  To say he's better than Fischer, a former world champion at this point in Nakamura's career is just silly.

There are a couple of problems with this way of thinking.  The first is assuming that Magnus would win every game.  He'd give up draws and even the *gasp* occassional loss.  Each of those draws would take far more points than he loses for drawing a 2750.

The second is that his wins would give him fewer points.

Fischer and Spassky played plenty of games against 2300 players.  Most of which they won, some of which they drew.  Imagine what Fischer's rating would have been had he been able to play against only the top 20 or so players in the world 95% of the time...

Carlsen's rating would only be negatively impacted if you think he would get less than the expected score against those players. There's no saying that this would be the case.

Infact on one hand he gets criticised for his tournament wins coming against the lower rated players, on the other he gets criticised for only playing Super GMs.

He's the best in the world at putting away minute advantages, I don't see why there's any reason to think that if he played more lower rated players that his rating would decrease. Infact his rating is probably impacted most negatively playing the strong players e.g. Anand, Kramnik, where he's almost guaranteed to draw.