Rating distribution excluding uncalibrated players?

I am interested in the percentile figure on chess.com. The problem I see though is that new players are often given a rating of 1200, that’s very high for a beginner. What does the distribution look like on chess.com if people who have played less than 50 games are excluded?
I don’t think those self-assigned ratings count until they actually play
Even if they play a few games, it won’t be representative though. A new player just learning should be in the 400s, maybe 500s. Seems like you need 25+ games to converge to towards your “real” rating

I think that you have to play more than a few games in order to be included. Like 10 or something like that, I can't really remember. 


1200 for beginners is not what it is, though.  Its 1200 for new to the site players.  There is a big difference; I would bet the majority of new to site players have the rules down and have passed the beginner phase of 'how does the horsey thingy move again'.   Not all of them, but most.  Whatever points you give the newcomers, that rating band becomes a free for all with the 2500s pitted against the 500s....   and I am not sure the math works out reasonably if you start them all at zero.   I think Magnus has an account on here and unless they have exceptions he probably started at 1200... wonder what lucky player got that match...