ELO distribution spikes at values N mod 100 == 0

Just now I've stumbled upon an elo distribution graph at lichess (by following "profile" -> "bullet" -> "better than x% bullet players") and the first thing that struck me were those notable spikes at values 1200, 1300, 1400, ..., 1900, 2000, 2100 etc. Why do they exist? This cannot be a player-based bias, right?

Players simply stop playing for a while, if they reach their goal rating.

Fascinating. So you are saying that this actually IS a player-based bias. And here I thought lichess devs tweaked the elo formula a bit so that players reach these custom milestones easier.

Lichess doesn't use the Elo formula but the Glicko-2 formula instead.

krasnaya