Creating new game - how to avoid opponents with provisional rating?

If you create a new game there should be an option to exclude players with provisional rating. Since all provisional rating are 1.500 in the beginning you will get a lot of opponents with a provisional rating if oyu are looking for opponents with a strength of 1.400-1.600.
In fact you don't know how strong your opponent will be and it's almost like a random selection.
So when creating a new game there should be a checkbox like "no player with provisional rating".

It seems like some players get a kick out of winning games by coming with a new name and playing as a new player. It sucks.
Now we are threatened with some kind of sanction for abor
ting another game with a 1500 question mark player

#1 Thanks for the feedback, but no, Lichess won't discourage new players from playing.

Also, new players have an RD of 350 so winning or losing a game has minimal effect on your rating: http://www.glicko.net/glicko/glicko2.pdf

@Toadofsky RD is not implemented in the way you think it is.
You lose the same amount of points against a 1500? player as you would against a 1600 player. That is far from a minimal change.
I do think that the best solution to the underlying problem is to lower that rating change.

The problem is a bit exacerbated, because lichess assumes that the bottom cap for the rating change per even game should be the one recommended for several hour long otb games, even if it is a hyperbullet game.

I still play unrated players on servers that allow those to be turned off, but it always feel like a sacrifice I make for the community, and not something natural.

I wish Lichess had the "assess" command (that tells you the potential rating and RD changes if players A and B played a game of rating category X) like on FICS so I could just go "ass randomNewbie" and compare it to "ass someEstablished1500" to see the magnitude of the impact of the opponent's RD without having to read a paper or any source code. Anecdotally, your own RD seems to have a far greater impact than your opponent's RD (which is to say: okay, I can see that the code does use the latter as well, but it's never made a large enough difference for me to notice).

github.com/ornicar/lila/blob/8dcddaa1048c347cde6d0a6773515764ee71877e/modules/rating/src/main/java/glicko2/RatingCalculator.java

The world‘s spelled: n-i-m-b-y

People who don't have a rating need to adapt to the lichess rating system, and that is why they need to play players with unprovisional rating. Aborting the games simply because that they don't have rating is a very bad habit and you should defenetly get rid of it.

It is selfish to first get a rating thanks to rated players being willing to play against you and then refusing to return the favour and also play against unrated players so that they can get a rating as well.

#6 Agreed.

#4 Is that so? I practically wrote the code and have been commenting on it (and how to fix its bugs) for the past few years, but whatever:
github.com/ornicar/lila/pull/3210 (PR for puzzle ratings; I had intended to do similar for player ratings next)

See also: github.com/ornicar/lila/pull/4034 (I don't know whether we're waiting on something here)

#9 My numbers come from looking at my games list, and seeing the rating changes. If the code is intended to produce different results, then it is probably wrong. Might be worth it to look at a more representative sample.
I don't see how those commits affect the topic. They seem to only handle RD change over time.

The problem is what happens in the normal case, where a player with the practical minimum RD of 60 plays against a player with the maximum RD of 350.
Actually that commit might make the problem worse. If my RD decays correctly when I don't play classical for a month, then getting paired with some 3000 Bullet player because of his 1500? classical rank will cost me 15 instead of 7 points. Not really making me feel better about playing those unranked players.

The cost of playing misrated players could be much reduced by making the rating change more dependent on the opponents RD. I still say the root problem is, that Glicko's recommendations for OTB games should not be blindly applied to online games.
On the other hand there are people complaining about unrated players even on KGS where you get rating back if the unrated player rises in rank, so complaining is unsolvable.