Have players been given rating floors?

I just won 2 games against this member by way of 'timeout' so i checked his profile to see what had happened to him and his games.....http://www.chess.com/echess/profile/cyo_07. If you look at his rating it hasnt dropped below 1984 even after many rated losses? im a little confused?

I think that there are rating floors implemented but that they only apply when the losses have been the result of mass time-outs.  There was a case some time back where a member timed out almost 1000 games and I believe the floors were ultimately the outcome of it.

It's absolutely a source of rating inflation.

Rating inflation is already a serious problem on the site without this innovation.

There won't be any comparisons of the players on the site today with the players on the site in the past so inflation is not an issue. One "problem" is players thinking they are getting better when they are not.

It depends a lot on two things:

The K-factor they use for higher rated players and how long and erratic the play is (a fault of the Glicko system).  He has lost some points.  They may have been small and the rounding still rounds to the same number.

I have written my blog on the Glicko system.  This is one of many reasons I personally wish Chess.com would match the USCF or FIDE (the USCF would be better for a site with many lower ranked players).  The Glicko system could have wild swings.  Therefore the multiplying factor had to be floored.

I think Chess.com (it is only a guess since I do not work for them) choose the Glicko system and choose not to use provisional ratings (you could lose with provisional ratings and your rating would go up.  Lose to a much higher rated player but are close and your rating goes up) is to have "instant" ratings and to keep interest right away.  I do know several players rated in the 600s who are very discouraged.

A provisional system would have been fair and kept them from being quite so discouraged.

You have some of the problems of the Glicko system in action.

Flooring is not an innovation.  It was done to TRY to keep an accurate rating of a player.  The Glicko system, admittedly by Dr. Glickman himself, can be erratic without the control of the multiplier and flooring.

Jim Fallet, Ph.D. (in Applied Statistics)

P.S.  A high rated player can play a low rated play (more that 400 points difference) and not lose any points.  Why.  The multiplier is essentially zero.  This is so higher rated players will play lower rated players without have to worry about them getting "lucky", etc.

If the K-value is 24 and the multiplier (in the Elo system it is a probability) is 1.3 a low ranked player can beat a high rated player and gain 30 or more points depending on the number of days since playing occurred last and how erratic the play was.   The higher rated player will lose essential nothing.

Now, during the first 20 games, your rating can jump hundreds of points because you start at 1200 and that may not be accurate.  Hence provisional rating by most chess associations.

This is the opposite of what occurs. Higher rated players lose many points for a loss while gaining none for a win.

Here in a series of games mingmercy gains zero points for two wins and goes on to lose considerable points to players rated more than 400 points lower.

Starting rating: 2519

Win vs 1190: 2519 no gain

Win vs 1094: 2519 no gain

Loss vs 2330: 2489

Loss vs 2199: 2456

Loss vs 1768: 2418

Loss vs 1953: 2383

Loss vs 1587: 2346