Why am I not in the rankings in the bullet?

P.S. if your entire point this time has simply been "don't call people cowards" I don't think anyone here has argued against that. Just saying...

I don't argue against that people lose their rank because of high RD.
My main point is that it's impolite to say "coward". And here you are agree with me.
And your third statement, about rating being unstable. Yes, I also don't argue against it. It's part of Glicko-2, and I absolutely agree with the system.
The point why I argued for 1st and 2nd, is that people needed it maybe to determine: if it's true and people who play lower rated players boost their rating, we have the right to call them cowards. But I argued for that it's impossible to boost your rating playing with lower rated players.
But I didn't argue against RD being high.
So, it seems, we agreed about my main point. That it's impolite to cal people cowards.
This is what I argued for. And here we are agree. So, I don't have any objections now.

jimj12, I gave a lot of evidences for the point #3.
And nobody yet refuted it. There are arguments for and arguments against.

Also, for the record, when I brought up ladders and pools I was agreeing with someone else's point based on the fact that ratings aren't meant to determine rankings but rather to match players of relatively equal skill for matched play. It was supplemental to my argument, not directly arguing what you were discussing.

And yes, we agree, calling top rated players "cowards" or "boosters" because they find it difficult to match with other top rated players is silly. I'm really not sure who was arguing that it was "right" to do so, but either way I'm not sure how it warranted 10 pages of tangential discussion of irrelevant statistical calculation. But then again, maybe that's why your name is "Chesstroll" lol

Fenris1066, I can say, why I took so long.
Because initially here people said that Singer__Marta, by playing lower rated players (and I don't refute that he plays them intentionally), boosts his rating. And was said, that if he would play opponents who have higher rating, he would drop to 2300-2400. And from this point, it might be (but it's still controversial, and I don't agree) possible to call him a coward.
And that is why I argued for that he is not 2300-2400 rated. Because this would prove, that it's impossible to boost one's rating such way, and that's why if he would play players of higher rating, he would stay his 2800.
So, if it's true and it's not boosting, and he can't boost his rating intentionally, he is not a coward.
But I didn't argue against high RD.
But his rating is approximately 2800, no matter how much rated are his opponents, if their rating is stable.

#153 you can't refute what doesn't have enough significant statistical evidence to support one way or another, and there is just no way to gather that. So basically you have a hypothesis that allows for a pretty wide margin of error, but when allowing for that wide margin produces results that support your hypothesis.

My main point of contention with said hypothesis, aside from the lack of substantive evidence one way or the other, is that with Glicko-2 your hypothesis becomes fairly irrelevant because of rating variance. You may have the same rating after all those games as you would if you'd played players at your own level, but it will end up with a high variance (and thereby now yield a ? after the rating on lichess). Likewise, I'm fairly certain that if an 1800 player does nothing but play games against a 2800 all the time, their variance will also go up. The ratings will only remain stable and thereby reasonably accurate if you are playing regularly and playing people relatively near your own rating.

And to that end, if a 2800 had a rating variance of +/- 300 then it could be argued that they might be boosting in the sense that they could, in theory, be a 2500. Of course, they could in theory also be a 3100 but it's unlikely. When someone's variance gets significantly high from consistently and intentionally playing low rated opponents, it calls into question the validity of their rating. That's why Glicko-2 uses this rating variance in making calculations, and why when the variance becomes significantly high, lichess now adds a (?) after the rating and trophies are removed. So if someone wants to continue playing nothing but low rated players, then they will eventually have to prove their validity again (I'm not sure how long it takes for their variance to rise, it has to do with total games played, time in between games, strength of the pool they're playing, etc....it's a lot of calculation that goes into Glicko-2, which is why it's such a good system).

So while I don't agree with calling anyone names or demeaning their character, it is justified to call into question the validity of someone's rating if their variance is always very high because their strength of pool is weak. I think that may be the point people were making.

Fenris1066, but isn't this a part of Glicko system? RD becomes higher with time, yes? So, if you wouldn't play for a year, you would get higher RD. And we also mentioned, that we need, if rating difference is very high, many games to get more or less accurate results. So, Glicko system consider it. And if you play an opponent of your level, it's like a game of, say, period = 1.
But if you play with an opponent who is much higher or much lower ("I'm fairly certain that if an 1800 player does nothing but play games against a 2800 all the time, their variance will also go up."), this period becomes 0,1. Because you need ten times more games, to get more or less accurate results.
So, it is the same, as if a player, who plays with opponents of his own level, would play ten times rarer. So, his RD will increase because of time. And, of course, despite high RD, it can't boost one's rating. Because, as we now, if RD is higher, you would get more points for a victory and lose more points for a loss. So even if he gets more points from much lower rated player, he will lose more from a loss to such player. So, high RD is because of time. And because big rating difference needs more games to be compensated, RD doesn't becomes low fast. And it's the same, as if one who plays with his level opponents would play much rarer. Same effect.

The point you're missing, and where we have contention, is that if someone 2800 plays someone 1800 a million times, sure there's a liklihood as you say he'll lose 555 games. But that isn't the reality.

No one plays that same 1800 player a million times. They play a large pool of low rated players, each and every game having a very low chance that they will lose. If I have a 1 in 1000 chance of winning against Singer, and I play him 10 games, I don't suddenly get a 10 in 1000 chance by multiplying my single chance over 10 games. It's a 1 in 1000 chance each of those 10 games. So the odds I'll ever win a game even over 1000 games is still relatively slim.

Now rinse and repeat those 10 games with 500 other players, all with equally poor chances of ever winning. There is very little liklihood barring outside forces (connection issue, too much gin, etc.) that any of those 500 people will win a single game in those 10 games. So that's 5000 wins. Even with a 0.1 per game, that's a +500 rating boost.

So yes, even though Glicko-2 takes that into account, it can never correct itself if someone doesn't win, and by spreading the games across multiple opponents the odds of someone getting a lucky win is even less likely. So again, that's why the (?) was implemented when the rating variance gets extremely high.

Fenris1066.
Let's postulate, that there is one killer on every 1000 people.
So, if you meet one person, the chance that he killed is 1/1000.
So, according to your logic, if this rather will not happen, that one single person turns out a killer, then if you meet 5000 people, you rather will not meet a killer. But no.
If you meet 5000 people, and you know that there is 1/1000 killers, there will be approximately 5 killers on 5000 people. I would agree only if it was showed another way:
If we have 1/1000 people killed, then among 5000 people there are 2...8 killers in average. So, yes, variation can be.
But if you say, that if chance to win is 1800, and it will never happen if we will change opponents, it's the same as if you would say, that there are no killers in the world, just because the chance of any random person to turn out a killer is 1/1000.
I don't understand this.