Well, I don't know what to tell you. Many titled players start playing here before they get official recognition, so they don't get their rating set to their OTB one.
I could go out and grab a bunch of players with OTB ratings and compare them to their chess.com ratings. While some will end up with lower ratings here, some will be closer in rating and some will be lower OTB. I've seen it first hand across a number of accounts (ones where I know what their OTB ratings are). Most certainly are not anywhere near 500 higher rated OTB.
At the higher levels, there very likely are larger discrepancies, but those are not systemic of a problem with the rating system, or of the fact the site starts with a default rating.
But you keep claiming the ratings are false. The only real ratings are those you get by playing rated games against other players. Ratings measure past performances within a pool. While they provide a statistical measure of winning expectations against players, that is all they do. Again, they are valid within pools and there may be some meaningful correlations between pools, but for most people, they are not 500 points stronger than their chess.com rating.
But if it makes you feel better, keep on thinking that. 
All pacts are undeniable and proven. I do not explain them here again. Read my previous posts again and ask yourself why you do not want to acknowledge undoubtedly facts?
One thought is enough to finally prove that the Glicko System and all ELO online system are extremely wrongly normalized if such starting values are used.
Here is another evidence that my chess skills are 2000 Elo not 1500 ELO - as I told you:
If you would do the test on this or other sites your would, again and again, find that I am right.
You are strategically denying all facts in a stubborn childish way that is not discursive. You cannot simply say that points are deniable without giving a reason for each point, keep this in mind.
If you compare online rating of some players you might find only marginal difference - but that does not mean anything, as the online rating of many players is false on many servers, even on FIDE.
But the OTB ratings distribution tells the story compared to your distribution not individual players.
For example, if a grandmaster starts at an online site they do not start with 1200 ELO points, right? They start with their official ratings or something like that. The will never play in the same continuum with the normal chess players that start at 1200. They get 2800 ELO for free and the only players that can beat them are players that also got 2800 ELO points for free. They are not in the same batch, get the point. Comparing two online ratings of normal players is the same effect if both ELOs are downgraded by artificial ELO scarcity for that batch of players. That's why it appears to you that it would function there. Also, it is the steepness of the ELO curve that is altered by the UNDENIABLE ARTIFICIAL SCARCITY OF ELOPOINTS. Thank you.