Mean rating deflation

Over on a reddit thread, plixmax calculated the following comparison of mean ratings on lichess in Dec 2021 vs today.

Here are his calculations:

"Means:

Bullet (which I don't play): 1475.82 => 1450.31

Blitz: 1534.03 => 1463.17

Rapid: 1525.63 => 1446.56

Classical: 1529.90 => 1495.14

They all deflated, especially blitz (-71) and rapid (-79)."

Does that suggest that of all the players on the lichess database, only lower rated ones are currently active?

Any idea what's going on here?

Source thread: www.reddit.com/r/chess/comments/w46kht/my_unscientific_case_for_rating_deflation_in/

this checks out with my experience. people seem to have gotten much better at lower ratings

@cameodemon said in #2:
> people seem to have gotten much better at lower ratings

But is that what it's saying?

Let's say lots of lower rated players drop out. You'd be left with the average of active players that was higher than the base of 1500.

But perhaps that doesn't work over the long term as higher rated players play other higher rated players. Would the mean then tend to fall?

it may mean that a lot of people started playing chess in the last few years and have now gotten better

Yes, I assume that's the case.
Due to covid, many people got into chess.
With time, they probably have gotten better at the game - how much varies from person to person, of course.
Many also left in 2021, 2022 and will leave - naturally, as interests change over time. So they affect the pool too.
In either case, we now have active players, who are a little stronger, as there are more "stronger" players than before.
So, I guess the cheating-incident did not get people to log in, rather to log out?

lichess.org/forum/general-chess-discussion/the-overall-level-has-risen#1

Same effect in other words.

Does that mean that I may add +79 to my epic 2300 rapid peak?

is it possible there is an influx of old players making new accounts?

Sure. There a zillions of double accounts, see the spikes in the cumulative distribution at full hundreds. They play until they reach their peak, making a new account and so on.

There are many effects simultaneously but how to adjust the mathematics properly?

@Sarg0n said in #9:
> There a zillions of double accounts, see the spikes in the cumulative distribution at full hundreds.

I don't think the spikes are because of double accounts. I think that's just human behaviour. I play everyday, but when I hit a nice round number I often stop for the day on that form of chess just so I don't drop back down again. I suspect many people do that.