It's relative to the pool of individuals, and there are tons of caveats (as other posters mentioned).

Despite this, due to the size of the pools of individuals and relative seriousness of players who belong to official tournament organisations, the ratings have been fairly stable. Yes, there's been some inflation, but I'd say it's somewhat been in line with the increase in knowledge and training methods. While the system has all kinds of minor issues, it's still "quite good", especially when we compare it to other sports or activities with a more variable pool and quality.

In this sense, most people know what 1600, 1800, 2000, etc, mean to fairly precise degree.

Lots of new players that join up, drop their ELO points into the "bag" and then leave (or if online, sign up new accounts). ELO "points" aren't a finite resource.

That would quickly lead to hyperinflation of ratings. The rating lost by a new player is not won by an old player. If you join and is assigned a rating of 1900, but loose your 5 first games, you probably loose a 100 rating points each game. However, the player beating you at 1000 rating doesn't get 100 points, just a few points.

It's generally understood that online chess ratings are inflated over what the FIDE/table rating is.

The first "1900" was meant to be "1000" as well. Cannot longer edit.

there is inflation (deflation too) and all kinds of fraudulent schemes.

https://en.wikipedia.org/wiki/Elo_rating_system#Ratings_infl...