Why am I not in the rankings in the bullet?

How can you not get this into your head? It's the same as Magnus Carlsen playing me. It's the same as the team of Real Madrid or Barcelona, playing AFC Wimbledon. It's the same as Roger Federer exclusively playing outside of the top 500. It's like Usain Bolt competing against secondary school athletes.

You can tell in all examples that they are superior, but because they never play anyone stronger, you can't tell what their true rating is. The system can automatically give them a rating, because that's a mathematical algorithm, but it denies giving them a rating because their +- deviation is so large.

It's not a punishment, it's just impossible with any certainty to discern where that player lies until they play someone a bit closer to their rating. The rating system is fine at ranking top players. He's the number 1: http://en.lichess.org/@/Elda64 notice he plays blitz and bullet with opponents from 2100 - 2300. He doesn't just play people 1,000 points lower than him all day, so the system is more certain as to his rating.

So if Marta wants to get a more accurate rating, and supposedly has a 2800 rating, he should play this IM to battle it out.

The player is artificially boosting his rating because he is using selection bias to only play people he is 99% certain he will win against. In a real life setting, it does not work like this. You do not choose who you play. As your rating improves, you get tougher opponents more at your level. For example, my rating is about 1700 in real life. When I play a team, I can't "choose" to play against their 800 rated player, it'd be absurd. I have to play an opponent usually around 1500 - 1900.

So in an online setting when you can "choose" your opponent, you can definitely artificially boost your rating.

Cynosure is correct.

No. in that case not boost rating .Human create mistakes.not engine play human

#31 I get you or this in my head, this is not the problem. But I would appreciate if you accept other opinions as well!
Or at least accept, that there is something like a small uncertainty, that leads to this discussion. ;-)

If a player has played a lot of games he gets the rating he deserves, regardless of the opponents he plays. You have to live with the fact that you can chose, whom you want to play.
Take the ELO calculator (http://chess-db.com/public/winprob.jsp)... to get 2000 it makes no difference, if you score 96/100 against an average of 1500 or 4/100 against 2500.
So likelihoods are build in the algorithms to balance rating differences. glicko-2 might be more sophisticated than ELO, don´t know.
As far as I remember (playing on chess servers since 1996) other implementations of rating systems are not taking the opponents rating into account calculating RD. Can´t judge if this is good or bad.
I do not want to defend Marta (or others doing the same), but my personal feeling is, that the handling is not entirely fair.
Once again: These guys do not violate any rule, they are simply playing chess. Nevertheless they are not ranked. This gets not in my head. ;-)
Unfortunately I will never have this "problem" because of my limited chess skills. ;-)

For those who do not know him; here is an extreme (and funny) story of someone who was, at a point, the second best player in the US by rating.

Forgot the link: en.wikipedia.org/wiki/Claude_Bloodgood

Very interesting read, thanks.

"the USCF changed its ratings system rules to attempt to prevent "closed pool" ratings inflation."

For rated games, I either propose the same (you can only play people in rated games +-500 of your own rating), or we just leave it as it is and people who play in closed pools don't get a ranking with their rating.

I do not agree. Everyone should be able to play whomever they want. I have a +-1000 scissor and, even if I am probably (for sure, I'd say) overinflated due to the fact that I play lots of people under 1900/2000, I enjoy having an almost instant reply to my seeks. And I sure would enjoy playing someone much stronger than me from time to time (That only happen once with a 2400 here).

I believe that tracking the resulting rating as "unreliable" is enough, if that is what is done here

#36 yes, interesting, thank you.

#38 in my opinion the possibility to select your opponents freely is not comparable to playing in a closed pool (prison), where the "master" teaches most of his "victims" the game. ;-)

#39 I agree.