Rating Inflation

I am curious to see if anyone else has noticed rating inflation in the overall rating of chess.com. 

When I started playing here, 8 or so months ago, the 50% level was in the low to mid 1200s.  Now its closer to the low 1300s...

My point is two fold: (1) Players need/like an objective measure of performance, and that (2) that few players want to see there rating go down, so this problem may not have wide spread concern (at least not at the moment).

Because have our own little macro economic universe at chess.com, which could be ground for some cool analysis, and I think we should take a look to see how ratings are changing. 

Erik, if you have a way to query members ratings in real terms (percentile) please let me know.

Ratings are going to rise in the aggregate, and the average of those remaining - will also rise.

  there are a couple of reasons for this:

a. new players come here, start at a nominal 1200, and lose some games (donating some portion of their rating points) into the aggregate of all ratings if added together for all remaining players.

b. players depart abruptly and permanently, and abandon or resign all games - dropping many points by that.

c. players come here, lose a few games, depart and return with a new name and start over in rating.  the games they lost, and rating points remain part of the aggregate of total rating points for all players.

d. these increases (if you will) to the aggregate of all players' points - are offset to some extent by those who cheat - are detected cheating - and are summarily removed with the high rating and points they had 'stolen' from other players.

Overall, there will be an increase over time - as the numbers in a, b, and c exceed the points removed from the aggregate by d above.

my thoughts, not a management answer.

A comparative metric is only good if it provides a consistent evaluation.  My question regarding inflation addresses just this.  I ask: Does the chess.com rating consistently give a measure of a player's ability over time?

I argue (without data, though I would like some) that the average player is it trending upward because of two factors.  First, there is an ever increasing supply of points to donate to a winning opponent as new players join chess.com. Second, the math used to calculate points does not conserve points in a transaction, and favors an increase of net points for any game especially when the loosing opponents RD is high. 

I dont mean to define "inflation" as relative to some other non-chess.com system. I think that such a correlation, while evidence of an accurate rating metric, does not address if the metric is in fact constant over time.  Such a comparison is looking at one point, rather than looking at a players rating over time. 

My question could be tested by looking at if the 50% level is the same raw point score last year this time. If its a little higher or lower that means nothing, but if its much higher or lower one of two situations is true: (1) as a group chess.com has gotten better or worse or (2) there is inflation in the raking.   

All this matters much less two things were true, (1) ratings werent looked at as similar to  other chess metrics outside of chess.com, (2) chess.com used percentile rather than rating as a gauge of skill. 

Issue (2) is the larger problem. If the Gliko system is a good rating system, then it can generate a raw score. To order by skill using the percentile measure gives an un-inflated result.  

The problem now is that the scoring metric is also the skill metric and they look at two different questions. 

But this implies that the ranking system is susceptible to inflation.