Puzzle algorithm change?

The Difficulty setting was just something added but is only part of the overall changes to the system. There are back-end rating changes made as well and the UI is currently showing the old data (rating of puzzles for example). There is work being done to show the back-end ratings it UI and that should happen in the near future. My understanding is there will also be some type of communication around those changes when they happen.

I didn't use a hint or reload the page. The only thing different is by default all of these settings are turned on. Note the setting

Show chat hints
Get advice from Coach when you ask for a hint.
I absolutely did not ask for a hint. Is there a glitch with this setting then?

As I wrote above, this is not new. I got exactly similar cases 3-4 times in the past 6 months. I have no idea what causes this.. .it is a bug I think... but definitely not new.

I can't say for sure. I don't think I've received negative points on a successful puzzle,, that I've noticed, in a long time. When I do puzzles I usually don't pay close attention to the individual rating changes.

there has been alot of threads about this, has anyone heard back from Chess.com customer service after asking them? has Chess.com made an announcement about this?

After reading about the difficulty setting in this thread, I tried upping the level. But, as others later mentioned, it appears that it's still serving up easy kills.

As a former software developer myself, seeing phrases like "they're going to fix this" strikes fear into my heart. What that means is they're determined to continue forward, trying to fix something that is clearly quite broken, and that we will all get to live with multiple iterations of wrong as they stumble through it, probably inflicting every attempt on everyone with no warning again.

I'd like to make a different suggestion: rather than being determined to battle through something that many customers agree is worse in pretty much every way, just go back to what was in place before this set of changes, at least for now. As many have said - and I agree - it wasn't perfect, but it worked pretty well. As you got better, the difficulty of problems increased automatically, just as almost everyone would want. And then, if you want to make changes to it, do it the way a professional software company would - figure out what people don't like, how to fix it, and recruit people to beta test what you do and give feedback,, rather than dumping it on everyone with no warning and no way back. The very best is to implement it as a choice - let people choose between the "new" and "classic" version. When people start VOLUNTARILY moving to your new version, then you know you've actually made improvements!

Just a thought.

(I also miss the list of recent puzzles under Stats. I often went back to re-analyze a puzzle hours or days later.)

Steve

That's still there, for the last 25 solvef

I absolutely agree with you, @samarak. Chesscom even has the feature to opt in as a beta tester, which you’re asking for. However, the recent changes were deployed to all users, although being clearly broken.

It’s actually worse than you described. In early May, Chesscom randomly assigned users to one of three groups, each group consisting of 5% of the total user base. Each group was assigned one of three puzzle selection algorithms, which are now linked to the difficulty setting.

I, for example, was assigned the algorithm that has now become the "extra hard" difficulty. I immediately noticed the change, voiced my concerns in this forum, opened a support ticket, and so on. Unfortunately, I never received any meaningful feedback. During this time, I saw many other users of the different groups describe their own problems in the forum. Yet, all of this was essentially ignored, with @Martin_Stahl responding very selectively to a few points in the issues described, as he did in his previous post. Since only 15% of the users were affected in this way, many users with the normal selection algorithm showed up in the threads, claiming everything was fine for them, meaning the problems never received the attention they deserved.

In the end, all users assigned to these groups were forced to use the algorithms they were given. This meant that 5% of all users were stuck with the "standard" difficulty (which I would honestly call trivial) for the past two months with no option to change it. As a fellow developer and a (until recently subscribed) user, this behavior is far beyond acceptable. It’s completely absurd.

I have. I sent an email to the co founder with my concerns with regard to the ratings system for puzzles. He agreed and said it will all be fixed soon.

The stats do say that you got 0/3 moves... If you don't find a move on your first try, that doesn't count... So let's say you try a move that's not correct, but you find the right one on the second attempt, then same for the second and then the third one, the trainer bot will still cheer you and congratulate, but you get negative points...

Same reason why a puzzle solved like this is not counted in your solved puzzles stats

That is not the case, as I've explained. My first move WAS correct. I've been playing puzzles for a long time, and know how it works. I posted here because I wanted to alert the team that they clearly have an issue as others have described. Whatever the cause is, I'm not sure but it is absolutely an issue

Regarding puzzle stats, we used to see much more historical puzzle data than just the last 25 puzzles solved. I used to be able to go back in time much more than I can now. I'm guessing the change was necessary due to server resources, still I miss this feature

At one point I believe one of the apps may have. Pretty sure the website only ever had 25.

in 4 years I solved more than 12000 puzzles and get a rank >2500. I was closed to finish the legend level when the algorithme changed. I finished the legendary level... I have to say it completely broked my effort and satisfaction. It leaves a bitter taste. I compared with lichess. first, the estimated level is not the same. I'm around 1800 (I have the feeling it is closer to the truth, do you have feedback about it?). And the pleasure of the raise is clearly better!

I see the option now to change the level, have not touched mine and everything seems the same for me. The only difference is some from my friend group have exceeded my level by making the puzzles easier. I am thinking I just need to focus on me and ignore the comparisions to others for the puzzle rating currently.

Good!

I don't understand why chess-com doesn't just use the Elo system with puzzles. I believe this is what Lichess does. If you play only 1000 rated opponents, you're going to have a very tough time getting your rating much above 1400 because the points gained per success will be so low. Sure, go ahead and do that, but your rating gain will become proportional to the logarithm of time (ie increase very slowly).
Just treat puzzles as opponents and everything will work.

This is the beta forum, right? Did Chess-com ask anyone's opinion here about this scoring scheme before springing it? Isn't it plainly obvious you can't offer the same point bonus for easy puzzles as for hard puzzles?

With the exception of the minimum+5, the site is using Glicko as far as I'm aware.