For simplicity, let's focus on just 1 time control.
In chess.com: I've previously reached a peak rating of 1360 under a starting rating of 1200. Then I stopped playing chess on chess.com for awhile (e.g., I went to lichess or chesscube or chess24 only). During this time, chess.com changed their rating policy to allow people to choose their starting rating between I think 400, 800, 1000, 1200, and 1500.
Question: Despite the supposed benefits, is it actually overall harmful to the rating system or confusing for players who started in the old policy?
Here's why I think this is harmful to the rating system:
For players who started in the old policy:
If most people choose, say, 1000, then...
Let's say I was around 1240-1290. Then when I get back, I would've been, under the old policy, 1370 - 1420 and so I would've gotten a peak rating of maybe 1490.
But now if everyone is choosing 1000, then it seems my rating is going to be like 1170 - 1220 with a peak of 1290.
This 1290 is not going to be in my statistics. It will still say my peak is 1360. Now, 1360 is a lot higher than the 1360 before. 1360 now is like 1560 before.
This 1360 gives people, including myself, the wrong impression, but this can be avoided if I just make a new account under the new starting rating policy.
For all players:
- If an overwhelming majority of people choose 1000, then it's the same thing as changing the starting rating to 1000.
Actually even for players who start in the new policy:
- If 2 people are about the same strength but then select different starting ratings... like let's say 10,000 players about 1300 on lichess join chess.com. But 5,000 pick 800 and another 5,000 pick 1000. If they play against people only of their own chess.com strength say (-25,25), in particular if these 10,000 actually make up majority of the entire chess.com pool (or you can imagine maybe there's a new chess website and then 10,000 lichess players join) then it seems we may have situations where there are people of ratings 200 points apart like 800 and 1000 but their actual strength is the same.
5.1. And then there's possibly no correction for it because these 1000-rated players never really play against these 800-rated players because they always pick (-25,25) opponents.
5.2. If I were such a 1000 and were to play against such an 800 rated player say in a private challenge then they are underrated and there is a huge asymmetry. Even if this is fixed in the long run, this is a huge hit for me in the short run.
I think it's harder to convert chesscom to lichess ratings now this way. Ignoring the glicko 1 vs glicko 2 thing, just pretend it's the same pool and it's 1200 vs 1500. Then I think it's simply a matter of doing +/- 300.
I think guess the rating eg eloguessr or Levy's guess the elo series will be much more difficult esp if players don't have a variety of opponents.
- 7.1. If they had played under the old policy then their peak rating might be nonsensical.