Puzzle algorithm change?

That's correct! And it's in direct collision with the fundamental logic of Glicko/Elo systems, and it must be eliminated. Literally hundreds of users have been calling for that!

@Martin_StahlAs far as I understand, the main complaint is not that it's only +5 (in contrast to +6, +7, +3). Rather, it is that Chess.com attempted to gamify the Glicko system, thereby sacrificing one of the system's most fundamental principles: the finite rating pool.

"In a pure Elo system, each game ends in an equal transaction of rating points. If the winner gains N rating points, the loser will drop by N rating points. This prevents points from entering or leaving the system." (Wikipedia)

Once all related work to the puzzle system is complete, staff could potentially revisit the minimum +5 on puzzles, but it's possible it's not going to be a huge issue once the work is completed.

My experience is similar to others. So apologies for the rehash. I turned the puzzles up to extra hard - but still far off what I am used too. Are the actual puzzles re-rated far lower than they were before or, as it feels, completely broken? I did really like the chesscom puzzle feature.

The good news is that puzzle rush survival is not broken and works as it should (for now). Hopefully it stays that way.

Even the very hard setting is useless now. God knows what the point of the change is, but as others have said, it makes the puzzles effectively useless, and the ratings are definitely no good. Has anyone seen anything from chess.com about this thread, with any answers?

I changed it to extra hard, but it didn't make any difference. Still getting the same puzzles 600 points below my rating or sometimes even less.

Same here...

Summarizing as best I can: after the latest unexplained puzzle changes

• some users' puzzle ratings have shot up after a previous recent round of unexplained changes
• other users are now suddenly suffering from stagnant ratings due to large & demoralizing rating hits
• setting "Extra Hard" difficulty in fact makes little-to-no useful difference to the difficulty of the puzzles selected, which are now always several hundred points below the users' puzzle ratings.
• This is a recipe either for boredom (if the user solves the puzzle) or rage (if the user muffs it: for instance, because of boredom).
• Some funseeking Chess.com users find the alternatives of boredom and rage to be unattractive.
• The new rating system, whatever it is, is suspected to be incoherent and unsustainable.
• the puzzle ratings displayed to the users aren't the puzzle ratings recorded in the system
• yet more changes are in the pipeline, but nobody's very sure what they are, when they will be rolled out, whether they will please users all that much, and whether they will in fact stabilise the situation
• an unfortunate but commendably patient forum mod is doing his best to explain all this, but he doesn't really know what's going on either.

I expect I've misunderstood some things and missed others, but none of it sounds particularly wonderful. With better forethought, all of it looks completely avoidable.

I was thrilled to receive a membership that allowed me to improve my puzzle rating. However, my rating then spikes by 1000 points, and the puzzles are boring and not useful to any chess player. I hope that chess.com chooses to address this issue instead of ignoring the clear dislike that this idea has received.

I've been on a diamond subscription since 2017, and really loved working the higher rated puzzles. Some could take hours/even days. Now everything is destroyed in a few weeks, and the fun gone away.

I'll quit my subscription in september :-(

Anybody knows any good substitute for chess.com regarding puzzles?

This is something I got from staff to share.

We are currently working on a significant overhaul of our puzzle difficulty system and puzzle ratings. Over time, the current puzzle rating algorithm created highly inflated puzzle ratings for our members — and consequent inaccurate estimates of a puzzle difficulty. As we roll out our new system, you may see some inconsistencies in your puzzle stats, and rating changes from puzzles solved and missed may be greater than in the past.

We ask you to wait with us just a bit longer! Since in the coming weeks we will be publishing an entirely new rating system for puzzles for all members. We’re confident it will be a major steps towards our ultimate goals:

 - Provide a more meaningful rating which correlates more closely to player’s Blitz and Rapid ratings.

 - Serve puzzles better suited for each player’s level.

 - Offer options to member who are looking for different puzzle-solving experiences.

Does this regard to users setting their own difficulty level as they remain in the same rating pool as everyone else?

I assume that means the difficulty setting and once all changes are complete, my guess is that different difficulty settings probably won't have a huge influence on rating. Though until everything is in place, that's hard to say for sure.

The developper that has committed an update totally broken is in holidays, no one can revert it ^^

@Ottobp1st and foremost Lichess, which also has puzzles according to themes, openings, and even mixing an opening for white against a specific defence for black, e.g.
puzzles for the Indian Defence against the London System

Puzzles are given a rating (Glicko) exactly like players. They exchange rating points with the users, like players. So that they self-balance into correct ratings, which also have a much better correlation with the players' playing strength.
Also, the Internet Chess Club has been resurrected and now features a very cool interface.

everything is still broken. takes away the fun of it all.

hii

i recommend ch*sst*mpo - it's a boutique site with only about 10,000 users, but they take this seriously:

* almost all the problems are taken from titled player games

* you get a complete history of all your puzzles. You can dowload it in excel. I recently repeated the last 3,000 problems I did to make sure i actually learn from my mistakes. You can't do that on this site

* different modes: one without timing, and two with timing. There is also a mode where not all the problems are winning. So you have to figure out that you're not playing for a win, but that that draw with a perpetual is your best option. There are a lot of problems where your opponent is coming at you and there is only one move to hang on - it's actually somewhat painful to do these, but then it prepares you for real games

* meaningful rating algorithm (Glicko) that actually measures your skill level over time, not how many problems you do. If you solve a problem that's way below your rating you get very few points. And if you fail it it deducts a lot of points. The way it should be.

* none of the rating inflation we have here - they actually make an effort to map your puzzle rating to a FIDE rating

* there is a lively comment section for each problem. People discuss the problem, what kind of tactical motives it uses, what kind of mistakes people typically make and how to avoid them. I learn a lot from these discussions.

Very good, thanks. I'll try it