Chess.com Ratings are a JOKE

all cheating

Okay bud ...

His longest streak in Blitz is 19 games.
His longest streak in Bullet is 22 games.
His longest streak in Rapid is 32 games.

He does not have a single 50-game streak in any single time control.  Now, he might have had a streak across time controls, but each time control has its own rating (for a reason!).  His blitz rating seems to be the one you are most concerned about.  This is how his progression went:

Starting rating:  400
After game 1:  580 (+180)
After game 2:  706 (+126)
After game 3:  795 (+89)
After game 4:  832 (+37)

He was over 1000 by his 8th game (when he lost to a 900 and dropped 50 points).  By his 23rd game, he was over 1400.  Now, you can make the case that they tweaked his RD a bit too fast, but the pattern of accelerating to his strength is consistent with how the Glicko system works (his main problem was being stuck playing people in the 400-600 range when he was over 900 ... and then losing a few games to players in those lower rating bands).  Saying "there was nothing quick about him getting back to 2200+" is asinine.  That is what happens when you lose!

And saying "he only lost 4 in his first 100 games" ... where do I start.  If you are actually strong enough to beat a GM (which he did around game 190ish), you never lose to someone below 1500.  Losing not 1, but 4 games to players in the sub-1500 range indicates he is not as strong as his current rating.  To put it in perspective, almost any random NM can play a 20 board simul against players in the sub-1500 range and not even draw a single game.  So losing 4 games to players in that range when it was not even a simul but a single game breaks your assertion about his strength entirely.  The same issue comes up when he got to the 1800-2000 range where he lost a ton of games (players that are actually 2300 should win roughly 98% of their games against 1900s).

As you said before, his previous account was banned for fair play, so his previous rating is irrelevant and useless (and looking at his games, you can make a case his current rating is equally as meaningless - especially considering his OTB rating is 1427 USCF).

Let's Turn This Into An International War

No, his previous account was banned because of something inappropriate he said/did on the forums. He is not a cheater

I think it's pretty obvious that, when you consider that there's constantly reports of someone 800 points lower rated than daniel naroditsky bot defeating him soundly, that the bot ratings are completely unreliable.

And you can't really program a bot to perform at a certain level.

He was banned for not being very nice to another member. I don't really wish to say any more without his permission.

I agreed with everything about your post except the parts i didn't cut out lol.

As a 2300 in blitz, I cannot maintain a 98% win rate against people rated 1900 in blitz. I believe the first time I played an untitled tuesday, I immediately lost to a 1900 in the first round (I think I was 2200, but it's close enough (?) ). Although I do agree that I'd win most games, just not up to 98%. In otb, though, I think it would be completely reasonable.

Ive looked into kowarenai's otb games (he blogs). I think he plays otb like he plays bullet. Move the bishop three times. Threaten checkmate in one. Oh, he saw that? Now what? Hmm... 

Also, he wasn't banned for cheating.

My mistake - though it wouldn't surprise me.  1400 USCF in classical, and 1500 in blitz (though it will jump up to 1700 on July 1 as he had a good event recently) and 2200+ online and beating GMs is highly unusual to say the least.

he spends a lot of time on this site, not only for playing but even for chatting with people and so .. so I am not surprised at all that his rating is higher here than irl and also you can't compare classical with blitz online... thats totally different time control... 

Granted, there is more variability in blitz, but in general a 400 point rating gap means you would expect the higher rated player to win roughly 98% of the time (obviously, in practice this doesn't always work out - but the point is you would expect players 2300+ to have a very large score against 1900s ... and literally never lose to 1200s, 900s, or 800s).  When he was playing people in the 1800-1900 range, he was roughly equal for quite a while.  1800 online is more reasonable for someone who is 1400 OTB (as online ratings tend to be slightly higher than OTB ratings).  2300+ for someone that level is not expected.  For reference, there is a kid at the local club who is 2182 OTB (18 points shy of NM) and his ratings on this site are roughly 2400 (he can hit 2500 on a good day).  So, 200 points difference between players online when over 800 points separates them OTB?  Very odd.  So, getting "back to his level" was actually done very fast - he got to ~1800 in under 75 games.  You cannot really expect more when you lose to much weaker players along the way.

You can go look all you want.  He doesn't have anything close to a 50-game winning streak in any time control.  When chess.com gives you a second chance account, you do not get to pick the rating.  It starts at the lowest one.  Would have it been better if they had started him at the Advanced setting (1800)?  Maybe, but that is not the fault of the system.  LiChess and Chess.com literally use the exact same mathematical system.  The only difference is how they configure it.

And no, I was not claiming "he'll get to his rating in 20 games".  I was saying he would get there quickly as long as he keeps winning.  It take much more time when you lose (the same thing happens on LiChess, by the way).

Not violating TOS - simply stating it is highly unusual and losing to lower rated players is a big part of WHY it took him 2 weeks to get back to 2200+.  And it wasn't just beating a GM 1 time.  He has done it quite a few.  If you've never played a GM in blitz, you do not understand.  The likelihood of someone who is 1500-1700 in OTB blitz beating a GM in blitz is about as likely as a 2200 beating Magnus in blitz.

But again, to the main point, when you lose and your RD is high, you lose more.  When you win, you win more.  As you play more games, your RD comes down.  Thus, when you have a provisional account and you lose to a very low rated player, it is going to start trying to stabilize you around that level.  You do not expect someone who is actually 2300+ to ever lose to anyone below 1500.

Not comparing to compare the raw numbers, but an indication of actual playing strength.  And his OTB blitz rating is much lower than his online blitz rating as well.  That was kind of the point.  CooloutAC was complaining that it took him too long to reach his "actual strength" - looking at his games (both online and OTB), his online ratings appear to be highly inflated, so he likely reached his actual level pretty quickly (in my estimation, roughly 75 games in the blitz time controls).

You are saying that his actual level is his OTB rating but I don't think so because from what I know about him he plays much more online than irl

Again, that is what happens when you lose.  That isn't the fault of the system.  The RD will continue to come down as you play more games.

For example, suppose you were ~2000 strength.  You create a new account and select "beginner" because you want to do a speedrun.  You are not expected to come close to losing a game until you start playing people over 1500.  So, as long as you win, your rating will continue going up quickly.  Once you lose, it will start to "wiggle".  If you look at his rating graph, you see just that:  it basically shoots straight up and then relatively flat.

It isn't just the number of losses, but the rating of the opponents you are losing to.  If his first opponent was rated 2000 and he won, his rating would have jumped several hundred points (just like it does on LiChess).  His first opponent was rated 500, so it didn't jump quite as much.  His first loss was to an 800.  Again, you do not expect someone whose actual strength is 2200+ to ever lose to an 800.  Statistically, you have a better shot at winning the MegaMillions while holding a winning Powerball ticket with a Sharknado swirling around you.

I'm saying it is highly unusual to have such a large gap between the two.  He literally just played in an U1600 G/25+5 event a couple weeks ago and went 3/5.  Not a bad score (though he lost rating points in it), but not one you would expect for someone with a 2200+ online rating either.  I expect online ratings to be higher than OTB (you can play more games, makes sense).  I don't expect to see a 1400 player OTB beating NMs, FMs, and GMs in blitz online.  The point being that complaining that it took him too long to reach his "actual" rating is a bit silly when he reached it pretty quickly if you expect him to really be around 1800-2000 online.

That is not at all what I said.  I said that they way Chess.com configured their rating system, they expect the mean to be 1200.  LiChess has theirs set such that they expect the mean to be 1500 (that is why LiChess starts you out with a 1500 provisional rating).  You can make the case that Chess.com should do that (they used to actually - but that was before they released version 2 of the interface).  But the underlying system is the same - you are literally just crying over how someone gets sorted at the beginning.

And not to be mean, but I don't expect someone rated 600 to really grasp what it is like to play a master.  They got there because they know, see, and understand things well beyond what a 1400 player can keep up with.  Sure, they are not robots.  But when you have built up the habits, knowledge, and skills to achieve master status, it is several levels higher than an average tournament player (which is what 1400 is).

Frankly, if you are looking to improve, I wouldn't recommend playing blitz or bullet.  But that is another topic entirely.  Once you have an established rating (i.e. your RD is low), you are only going to get about 8-10 points per win against people similarly rated.  So if you suddenly have a breakthrough (maybe you were practicing tactics and things finally started to click), it might take you 10-12 games to break into 900 or 1000.  If you stopped playing for a while, your RD would go up, and when you came back (with the same epiphany) it may only take you 1-2 games to make the same rating jump.

I don't know anything about how long it took than he did reach something..
I am just saying that he is a type of person who
1) plays chess fast, with focus on time, trying to trap his opponent in some tactic what would be too cheap too be useful in long games but is quite strong in blitz..
2) he plays chess much more online than irl
these two things are enough for me to explain why when he played 25+5 he had worse score than when he plays here blitz and also why he is only 1400 USCF, his blitz game style is just terrible in long time control and also he doesn't play enough irl games to his irl rating would be accurate
If you are not satisfied with this explanation and do you think that he cheated somehow its your thing.. but its only a pure speculation and nothing more and this website doesn't much like speculations about cheating

Not at all.  I just know what it is like to play against masters.  You can be a "blitz specialist" all you want - there is still a massive knowledge gap there.

The rating system measures past performance in an attempt to get you into competitive matches as quickly as possible.  Chess.com lets you pick where to start (except for second chance accounts) where LiChess just assigns it at 1500.  You seem discouraged that someone could be really strong and pick 400 only to crush the lower rated players for a while before reaching their actual level.  How is that much different than having those same lower rated players start at 1500 and get crushed by actual 1500s for several games before they finally get down to their actual level?  The system cannot look into your mind and figure out where you belong - all it can do is measure your performance.

You seem to think this is a major problem, but the irony is the system you recommend (the one LiChess uses) is exactly the same.  The only differences are where they configured the expected mean and where they start your provisional rating.  The end result is the same:  you play games and your rating goes up when you win, down when you lose.  So, to borrow your closing:  get over it.

I will make it easy for you since you seem to have a hard time here:

And no, BOTH chess.com and LiChess use the Glicko system (Glicko 2 to be exact).

I wasn't bringing up your rating to insult you, nor to make myself feel better.  I was doing it as you were asserting "people aren't robots" as if a 1400-rated player would ever beat a GM.  To put it in a another sports analogy:  it would be like your local high school football team beating an NFL playoff team - or hell, even the worst NFL team.

And feel free to continue believing that you can improve that way.  Good luck!