Should lichess forbid this kind of behavior?

Is beating 2700s 100% of the time not impressive enough to get a 3200 rating?
The rating system is designed so that even a single loss/draw to a much weaker player leads to much more rating loss than gain. the rating gain is minimal, and the higher rated the player becomes, the less rating gained. Even if the player does this infinitely, the max rating he can achive using this method is capped at around 500 points from the opponent rating, and that is about the strength you would need to beat the opponent 100% of the time.
Not to mention it takes a long time to get paired with a 2900+ outside of titled arena ( obviously untitled players cant play titled arena ) you dont have much of a choice other than playing 2600s.

Have you ever tried beating much lower opponents consistently? I've played in many arenas where the opponents' average rating was several hundred points below mine. The favorites are likely to lose rating there, at least in the long run. I've played several 2900 and higher rated players, and I have won quite some rating in these games.
Last but not least, in variants the highest-ranked players find it hard to get similarly rated opponents, given that many high-rated users only play bullet or hyper. And who wants to want to wait several minutes for getting paired?

@likeyacutg said in #1:
> I have started to see more and more high-rated bullet players farming people 500-600 points lower than them for +0 to +1 per game. On the surface there is nothing wrong with playing weaker opponents, but if somebody deliberately avoids people anywhere close to their rating I think it becomes a problem because they achieve ratings far beyond what they should be under normal circumstances. I have edited out the names of such users, but let’s just say there is a certain untitled 3200 currently on the leaderboard who perfectly exemplifies this behavior.
>
> For everybody who does this, you can search each user's rated bullet games and see that opponents are all an absurdly lower rated, and you'd be lucky to find anyone above 2700 in these games. While I don't think a user should have their account closed or labeled with the "violated tos" banner, I think lichess should enforce rankbans for this kind of behavior, because it's clear that people are inflating their ratings far beyond what they should be. Thoughts?

@Cedur216 said in #7:
> Excessive farming can work in variants, but it only gets you rankbanned, so there's absolutely no gain from it.
Many people are getting high ratings in variants through playing mostly hyperbullet, not zerking and not playing in prize events. All of this is legal.
There're much fewer players earning rating primarily by playing much weaker opponents. I suspect that many people do not really realize how difficult it is to score 19/20 against opponents rated 500 points lower.
The main problem with variants ratings is that they take all the time controls together.

Just a concrete example, my finest arena result on this acc so far: lichess.org/tournament/SUSRKQgx
I scored 14.5/16. The opponents' average rating was 2388, many of them are titled. I reached one of my finest results, gaining 3 rating points there. It could've ended differently, in one game I delivered mate with half second left. From some comments it looks as if I were a coward who's farming rating. I hope that my point of view is more understandable now.

@ILikeBlitz I see what you are saying, but the difference is that is a tournament where you don't choose who you play, but you can outside of tournaments. As you said it does happen in hyperbullet, but is there not an issue if the person is intentionally avoiding anyone who has a chance of winning?

The main argument against banning such excessive farming is that it is not effective.

If anyone does not believe this happens and people do gain large amounts of points from farming low-raters, DM me and I will tell you the usernames of some of the most egregious cases so you can see for yourself. I'd be happy to share these examples because once you see it there is no denying that farming very low rated people and not playing anyone near your rating is an effective rating-gaining method.

Also as I am writing this message the untitled 3200 farmer who I hinted at in my original post has been rankbanned, actually.

@likeyacutg said in #15:
> @ILikeBlitz I see what you are saying, but the difference is that is a tournament where you don't choose who you play, but you can outside of tournaments. As you said it does happen in hyperbullet, but is there not an issue if the person is intentionally avoiding anyone who has a chance of winning?
>
> The main argument against banning such excessive farming is that it is not effective.
>
> If anyone does not believe this happens and people do gain large amounts of points from farming low-raters, DM me and I will tell you the usernames of some of the most egregious cases so you can see for yourself. I'd be happy to share these examples because once you see it there is no denying that farming very low rated people and not playing anyone near your rating is an effective rating-gaining method.
>
> Also as I am writing this message the untitled 3200 farmer who I hinted at in my original post has been rankbanned, actually.
OK, now I understand better what you mean, I've removed thumb down from your initial comment and from several others. That said, in my opinion the problem (and reason for rank-bans) is mostly in games between concrete pairs of people, there might be some agreement. There're cases when some players got surprisingly high FIDE ratings through playing low rated opponents, but those people were often excelling in their own tournaments, so the main problem might have laid elsewhere than in the rating difference. In general, when one keeps playing random opponents rated 400-600 points lower, the stronger player is more likely to lose rating than to win it. In bullet, blitz or rapid, a player rated 500 points lower is not without chances, at least in some games. And scoring consistently 90 per cent or more is very difficult. (I'm not sure about some variants like horde.)

@likeyacutg said in #6:
> @petri999 in theory this is true, but in practice it is largely effective when you get a weak enough opponent.

If this is really the case, it is a failure in the whole rating system. It is designed in a way that your lose or gain is independent on the opponent's strength (i. e. if you are supposed to win 80% of games against -300 players, every defeat counts as four wins).

@OctoPinky I agree, and I think the rating system may break down when the rating deficit is too high, as often the lower rated player plays worse than they normally would because they are nervous to play such a strong opponent, and makes farming easier because most don't play at their true strength when faced against someone 600 points higher.

@likeyacutg said in #18:
> when the rating deficit is too high, as often the lower rated player plays worse than they normally would because they are nervous to play such a strong opponent,

There is that possibility, right... or they could be more relaxed or focused and perform at their maximum... I've experienced both.

It would make an interesting experiment to check the rating system accuracy for high rating differences.

one can test how glicko works faster than in game by playing with the drop down difficulty menu..

try the easiest.. and let your rating go down by doing very fast play.. do the puzzles as if they were bullet, I guess..

the higher one is with respect to the puzzle the less the gains, well I have had many examples where success was no increment at all because a lot lower than my rating.. on the other hand the risk of getting some rating flesh ripped out by failing a lower puzzle is higher. something like that.. I think there is a bottom gain to framing and it might be zero.. (maybe puzzles have an added threshold as it does not give non-integer increments).