Lichess Rating Floor

@Abigail-III you're correct my bad. either way it isn't very competitive

@rickrenegade exactly this is the way floors create inflation and it goes upwards. Originally Fide had high rating floor and it quite accepted analysis that this caused inflation. Not as bad as absolute floor as back then people who got less 2000 were dropped from the list untill they back again when fluctuations in rating raised them back.

Rating floor of 600-800 does not really affect that much as the are very few players that weak and who still play

en.chessbase.com/post/rating-inflation-its-causes-and-poible-cures
search for words "very high rating floor" to get that point and see the graph that shows the effect

In Elo system other source of inflation is the way initial rating is done. No estimate onstability of rating is in use

@petri999 That's a very interesting and detailed article.
The original 2200 rating floor is outrageous. What were they thinking.

Next debate;

Why are Lichess ratings so inflated and everyone is so overrated?
+200 - 300. Maybe even 500 points higher than their FIDE.

Not sure: if you constantly pump very weak players in the system, (over-)rate them with a too high value. Like guys who barely know how to move the pieces, start with 1500 or 1800 on some servers. Everyday lots of new members, double-accounts when reaching a „cap“ or milestone, accounts rise and instead of decrease they are kept high and passive. It‘s a mess.

The effects are multi-layered, no clear idea what causes what.

Llichess rating are not supposed to match FIDE and hence the difference is not significant. On a closed pool a median player should end up with starting rating used in the pool. And then the pool quality dictates where you end up. Like Lichess classical ratings are hundreds of points above blitz - that is not a problem it just happens that people playin online on longer times are weaker than those playing blitz so it is a weaker player that gets 1500 there.

Once the pool is started and algorithm is something meaningfull (i.e. no Elo) then ratings shoudl not be subject too much inflation even if starting rating is off. Since the fact that rating is provisional is taken into account while calculationg ratings changes to others. and dropping 300 points takes only few games.

Funny thinng that FIDE knows that their rating system is not great. They even organized competition for better one but thats about it. The ancient algorith which meant to be simple enough that you can compute the values with paper and pen is gonna be in use forever.
www.kaggle.com/c/ChessRatings2

Alos there are even on public better systems. These dudes have proper maths behind their computations
http://universalrating.com/

@jg2017
My rating has the '?' because I lost most of my games, and the rating is therefore still provisional. You get out of the provisional range quickly if you win and lose against a range of opponents, but I mostly lost against just one opponent. I know the floor is 600 because my rating just wouldn't go any lower after that.

@petri999 Rating inflation will happen if, on average, people retire from the group with a lower rating that they had when joining the group. Their net effect will be that more points have been brought in, so on average, everyone else gains a little. The opposite will happen if people retire with a high rating.

I thought the rating floor was 800. I've seen the edit on github. When did it change to 600? And can I test it?

I have been banned one time already for sandbagging. Lichess doesn't believe someone can be bad at chess. But if they changed the floor from 800 to 600, maybe they changed their mind?