Average Chess.com Rating

Chess.com provides some statistics about the average rating and number of active players which you can find here: https://www.chess.com/leaderboard/live

A few years ago I was interested in how the number of active players and average blitz rating would change over time, so I started taking screenshots like these:

This became more interesting with the double chess boom (COVID and the show "Queen's Gambit") which the site owner wrote about here:
https://www.chess.com/blog/erik/incredible-second-wave-of-interest-in-chess

I also wrote a program to simulate new players joining an established rating ecosystem. I wanted to understand why ratings might inflate, deflate, or why the average might change. I suppose that could be a separate blog, but the short story is the creators of rating systems have done their best to minimize any inflation / deflation effects. As a consequence, the average rating changes in just the way you might expect i.e. if beginners join then the average rating goes down, and if grandmasters join the average rating goes up.

---

Here is a graph of the data I collected at that time (below). You'll see that as the number of active players rose by 4.7 million, the average blitz rating dropped 120 points.

-

-

What's interesting to me now is to notice there are fewer active players today (10 million vs 12.7 million 20 months ago) but the average rating has continued to drop (down 60 points)... I suppose this means that in the time period since the chess booms the players who have quit have tended to be above average, but is that really the case? If it's true, why would below average players tend to stick with it longer? I don't know, but it's interesting.

While I didn't save data about rapid and bullet ratings, from memory I know they have also dropped since then and are currently also below 800.