Based on experience of doing 10,000 old puzzles and 2000 new puzzles, new puzzles rating, that started in last year, is about +300 elo inflated.
In routine mode, 2500 new puzzles are around 2200 old puzzles difficulty.
Do you agreed?
I don't know if I agree or not especially when you're comparing old to new format and what I assume to be a different rating system?
I think you have quite a bit more experience with puzzles but to me it seems somewhat logical.
That is, the difference between my elo and my puzzle rating.
300 points difference actually makes sense to me as it demonstrates the difference between your ability to play and your ability to analyse. The latter should be significantly higher.
<Comment deleted by user>
I have a higher rating under the new system, but I think it is partially because it also consistently giving me difficult problems now. Under the old system I would get spammed with problems 500 points below my rating, so if I missed one it would cost a ton, but getting one right would give me very little.
I am between 2700-2750 puzzle rating at the moment and most of the strong GMs (much stronger players than me) seem to be similar. My suspicion is that it is due to a combination of most players solving problems more casually than me (answering quicker, before being certain) and perhaps above that range there are not enough puzzles?
What I am trying to say is that relation between puzzle rating and playing strength is not that clear, both previously and now.
The purpose of ratings is to create equal matches. If you get puzzles equal to you rating you should get 50% correct.
A particular value such as 2000 or 2500 only means what you decide it means.
LiChess could subtract 750 from everyone's blitz rating, add 500 to their standard rating and add 1800 to their puzzle rating (and the same number to the puzzle ratings) and everything would still work perfectly.
Someone might be 250 blitz, 1700 standard, and 3500 puzzles. Those ratings would be 100% accurate just like today's ratings are.
@StingerPuzzles I think you are missing the OP point here.
This isn't a question of math and/or disconnected ratios.
His point was that a 1400 elo player shouldn't have a puzzle rating that is inflated to a point of absurdity or else it doesn't indicate his level at all.
@Korniliousim Ratings have nothing to do with level.
Ratings are to create equal matches.
If you're doing 50% against a puzzle or person with the same rating your rating is accurate.
Cool, so I got 2400?
I'm sorry but this is getting rather weird.
Ratings have everything to do with levels.
That is the very point of ratings, to tell you who's a beginner, who's intermediate, who's an expert, who's a master, etc.
An expert level player (i.e - a rated untitled ~2200 player) should be good enough to solve puzzles of equal rating.
That your analytical prowess can exceed your practical level is the very issue here but in general a 1200 rated player isn't supposed to be able to solve a problem that is positioned at a grandmaster level. Hence, the OP's point that the ratings are inflated. It's all based on its relative connection to a theoretical elo.
Your comments are disregarding it entirely and addressing it as a stand alone system. It isn't!
If you're scoring 50% against 2400s, then that's your accurate rating.
They could subtract 2300 from all ratings, and then any number above zero would indicate a great player like you.
