When did Elo inflate?

....but isn't elo a zero sum game ?....I mean as they inflated ?....wouldn't there also be equal & opposite lower ratings elsewhere ?

So where did the Elo points come from, after 1985 or so?

That's a good point. If we want to talk about some kind of general rating inflation it would be better to look at averages, not just the top 10 players.

One way to inflate ratings would be to introduce many players whose skill is below average.

Another way would be if many above average players started retiring.

In this case I mean inflation as a person's skill stays the same but their rating number increases.

Elo probably inflated in his fifties. Too much goulash I suppose.

In the past, getting a GM title at age 18 was very good. Anand got his GM title at 18 or 19 for example. These days, of course, that's far too late.

Of course Anand himself is (or has recently been) in the top 10, but it would be "too late" for a modern player because these days the people who later become top 10 players start systematic training at a younger age, have more access to information, and play more often. So if we assume they're doing that they should get the title before Anand got his.

And because of all this learning and practice, even if a rating system didn't exist, it would be very reasonable for a person to conclude that today's professional player is simply stronger than the professional player from the past.

Sometimes I've heard older GM lament "oh, Botvinnik wouldn't have misplayed this position, or this endgame" while looking at a modern game. So they conclude today's players are weaker. One thing to remember is these days the time controls are shorter and there are no adjournments where they could study the position all night and then play in the morning.

And the better tools modern players have to improve (chessbase, strong engines, frequent tournaments) disproportionately affects the best.

You know, just because chessbase exists doesn't mean everyone buys it. Just because there's a tournament every month doesn't mean everyone is playing... I'm sure many people on chess.com have never even been to an OTB tournament

Since only a relatively small number of people make use of the tools, what happens to the bell curve is the right side stretches out. Here's one from chess.com

https://www.chess.com/leaderboard/live/bullet

Maybe the overall average is the same, and it's just the top 1000 players have taken something like 0.01 rating points (on average) from the rest of the field. If this were accurate, then today's 1500 player would be comparable to a 1500 player from years ago (even though historically ratings didn't go that low in FIDE so it's hard to know).

In the US, from scholastic chess, where low rated kids are adding ratings points and then leaving the pool by adulthood faster than old chess players with high ratings are dying and leaving the ratings pool...prior to '85 you also had the Fischer bump doing the same thing.

People never seem to understand that the ratings pool attenuates over time.  Adding a bunch of lower rated players who lose a bunch of games and then leave chess never to come back creates lower rated players at the bottom end whose rating will never change, but more importantly, through feeding all the way up the chain, create slightly higher rated champions at the top end of the ratings pool.  A kid comes in at 1000, loses to 50 other kids rated 1000, and quits chess in disgust rated 600...that 600 pool rating sits stagnant as a placeholder.  The key here is that the 600 ratings points points remain stagnant, but the 400 ratings points nibbled away by other players do not...all of those 50 kids that beat this player gain a few ratings points, which are then lost to, say 500 1200 rated players, who then lose them to 5000 1400 players, etc.  It's a slow process of diminishing returns as those points are spread more and more thinly across the entire pool, but in the end you have raftloads of 600 rated defunct players, and Magnus Carlsen and the top 100 gaining a tiny sprinkling of fractional ratings points...but those build up over time.

Elo is flawed in this way, because it tries to be a closed system, but it's not closed at all, and can never *be* effectively a closed system.  Players join and leave the pool, and as long as chess is not played seriously by more than a few percentage points of the world's population, the number of people joining the pool at the low end and leaving after losing a bunch of ratings points will always be greater than the relatively few high rated players that die off by comparison.

I think there was a movie about this.  They even made a sequel to it.  

It's called "Dead Pool."  Never saw it though.

Thanx Stiggy. Makes sense now  .

http://www.chessgames.com/perl/chessgame?gid=1028959

http://www.chessgames.com/perl/chessgame?gid=1001662

http://www.chessgames.com/perl/chessgame?gid=1001867