Highest possible standard deviation?

I don't see a rating deviation in puzzles. With puzzles my rating dropped because I was practicing a weakness of mine.
https://lichess.org/training/dashboard/30/dashboard
The dashboard points out the lowest of ratings. Then you go to that theme and practice.
https://lichess.org/training/themes
At first it might drop a rating, but with practice it will rise again to your depth vision.

A chess player will always be limited by their depth vision. It takes time for an engine or a human to reach a depth. That's where cheats stand out and cannot guess how long it takes to have that vision. A vision depth deviation profile can probably be built on that assumption.

@Toscani said in #11:

I don't see a rating deviation in puzzles.
You might not see it, but it has to exist. Glicko-2 requires it.

Anyway, I'm not talking about a user's puzzle RD. I'm talking about the puzzle's. Each puzzle has not only a rating, but an RD and volatility, just like the user. But the motivations for Glickman to add RD to his algorithm (old players get worse, idle players get worse, etc.) don't apply to the strength of a puzzle; it's just a puzzle. Short of inventing a new rating algorithm that allows for an RD on one side but not the other, why not just allow a puzzle RD to drop very low?

I see that in humans a rating gets old and the human could have bettered their rating or with age worsened their rating.
A puzzle does not age, so once it reached an RD of 45, then maybe it could become a permanent rating. But what if the rating was built from players that use engines to solve puzzles. So the RD needs to remain to compensate for that abuse of assistance.

I guess over years of it getting solved a median permanent rating could be given to the puzzle.

Mine is terrible if anyone wants to see my profile for proof.

Deviation difference of 967 (Between Ultrabullet lowest and Blitz highest peak)
I believe the larger the rating difference the more it proves we play too fast. If I don't count the ultrabullet, then the rating difference will be narrower.
@LordSupremeChess

@Toscani said in #15:

Deviation difference of 967 (Between Ultrabullet lowest and Blitz highest peak)
I believe the larger the rating difference the more it proves we play too fast. If I don't count the ultra bullet, then the rating difference will be narrower.
@LordSupremeChess
Yeah true.

Gotta admit, that title does have me snickering a bit.

What title??

Since lowest possible is 45, I think maximum possible is around 500.

@LordSupremeChess said in #18:

What title??

The title of the thread! Wake up.