I don’t think that’s as true as we want it to be.
If some random person comes along and thinks “I’m 7d on fox I’ll pick 7d” and we give them max deviation of 350, and they lose to an established 3d
the 3d gains like 12 rating points, but that’s actually comparable to beating an established player at their own level
This probably will happen regularly even if we could provide an accurate rank comparison chart, because people probably won’t go to that effort to play games. It doesn’t affect them anyway, but it probably will cause some instability at the entry points.
@Vsotvep was playing around with glicko calculations and I believe they suggested it might not matter if we let people set ranks
I’d probably rather see it done with the actual OGS games database, which should be doable on python from here
At least then you might get some overall pictures of how it affects certain entry ranks or overall rating drifts and so on specifically for OGS with all the strange things real people do with their ranks and games.
Various quotes about letting users choose their own rank.
It’s not just an OGS thing that people can’t reliably choose their own ranks, I played a player at a recent tournament where they said they were strong like first rank (it wasn’t said to me directly so I don’t exactly know) in China or something so they were supposed to give me four stone handicap as a two kyu - I won by a good bit, then we played even and I won by a lot, and then I could estimate there were more like 8-10kyu than X dan, which we could test by having them play a 11 kyu where they won.
Only that we discounted the first round pairing (so it’s like I sat out), and played some additional games to figure out their rank did we give them a better chance at having some more even games in the tournament.
I see this as somewhat analogous to what glicko is doing anyway, someone isn’t sure how strong they are, they play some games lose some games and they quickly within like three games (if something silly doesn’t happen) start to pair with people their own strength approximately as in with this real life case.
For total beginners, who know they are total beginners, there is a point about having them play a 12kyu to whichever, which I think can be solved by fixing the pairing algorithm, rather than messing with the ratings entry points