Thoughts on Puzzles

I’m not sure what all has changed on Puzzles in recent months. Sometimes I do puzzles every day, and often I go weeks between sessions.

I see more glitz. “Coach” who offers information of dubious value in game analysis has invaded the puzzles sanctuary.

A big problem the past two days is that the puzzles I see average 2000 in rating while my puzzle rating is close to 3000. The past two mornings, I’ve done 55 puzzles. When I get one wrong—too often—my rating drops 18-29 points. When i get them right, I get a maximum of +5.

This puzzle, the 55th is the sole exception. Solving it in 7 seconds netted me +10.

https://www.chess.com/puzzles/problem/2699246

Please feed me better puzzles and eliminate the distracting graphics.

This algorithm change has made me swap websites for standard puzzles. It doesn't seem to apply to puzzle rush / battle so I'm still doing those, for now at least.

Hello,

I do agree that it is highly frustrating to loose 25-30 pts (i even lost 50 once) on a single puzzle while collecting only 5 pts on a successful one. A huge amount of efforts to offset a single mistake.

On top of this, some of these puzzle admit several solutions and one only is available. You then get punished even though your sequence is fine.

In the long run, you can only lower your score.

There's a chance the rest of the changes will be available next week. There will probably be some additional modifications after.

That said, you can't really look at the puzzle's ratings as the value showing is from the old system and not from the rerate and rating being used for the selection algorithm.

I agree that chess.com seems to have a history of trading in good and popular systems for ones that are less enjoyable and a lot buggier.

"On top of this, some of these puzzle admit several solutions and one only is available. You then get punished even though your sequence is fine."

That used to drive me crazy when I first began doing tactics training. I would play an obviously winning move, perhaps even leading to a forced mate, yet I would be marked wrong if there was another move that was one move faster.

However, after a while, I came to appreciate this training in ruthless accuracy.

That should never be the case and if you run across any like that they should be reported. While there are some in the database, the majority don't have secondary solutions that are also winning.

"That should never be the case and if you run across any like that they should be reported. While there are some in the database, the majority don't have secondary solutions that are also winning."

That may be the case now, but I remember from years ago, being marked wrong for choosing a move that led to a mate in 5 when a mate in 4 was available. I remember because it annoyed me at the time. Now I would be good with that.

Also, there was one where a clear win of the queen was wrong, because a mate was available.

Even then, those should have been reported, though I personally liked ones that happened to have longer mate possibilities as well, as that could hep with calculation skills.

Did some exercises on my computer this morning. Failed the fourth one, a diabolical puzzle: https://www.chess.com/puzzles/problem/1872680.

The graph had a green dot, instead of red. 
There are further errors in the math challenged "coach's" comments. I'd like to make him go away permanently from all areas of the site. He's worse than Kevin (an internet poser).

If you get it partially right it may do a green dot instead. I haven't done rated only mode for a while so don't recall.

Partial right is still wrong and red dot, but sometimes the rating goes up.

In this case, I got it wholly wrong and still a green dot. The dot shows as red in the iOS app.

The app is cleaner and I prefer it. No obnoxious “coach”.

BTW, after changing my setting to “hard”, I’m still getting puzzles that average 1000 below my rating. Both iPad and laptop.

I suspect this is due to chessdotcom tilting at windmills, i.e., attempting to lower my tactics rating to match rapid or bullet. I thought that @Erik had a MA in Stats, and yet his hired guns don’t know that all rating are pool specific and cannot be reliably compared.

There has been a change in the ratings of puzzles, as mentioned previously. The rating you see in the interface is the old rating band is not being used for puzzle selection and isn't used for rating changes.

That should be updated fully, likely next week, and the ratings being used will be updated. The intent is to have member puzzle ratings be more in line with game ratings, instead of massively inflated as they are now for most members.

I want the massively inflated ratings because they are stable. The only thing about ratings that matter are comparisons to prior performance and to my friends. I want to get above my previous peak (last June) of 3232, and get a higher peak than my friend Antonius.

If chessdotcom “fixes” the ratings, they will break another thing that is not broken.

”If it ain’t broke, don’t fix it.”

Imagine if @Erik adopted this slogan.

Inflated ratings are broken and that's one of the reasons for the changes. Unfortunately, trying to beat old peak ratings isn't going to be possible, for most members, after the change. However, the ratings will make much more sense

Ratings are pool-specific. Inflation is a myth.

Staff decided the puzzle ratings were not in alignment to where they should be. That includes member ratings and actual puzzle ratings.

I’ve disagreed with a substantial number of staff decisions. This site has been moving towards glitz over substance. You see it in the constant promotion of playing bots. You see it in the “friendly” game review that now leaves out a substantial amount of data. In addition to be of far more personal value to me than the “coach” interventions, this data was invaluable when I did fair play analysis of my state’s scholastic championship when it was online on chessdotcom during the pandemic. I guess glitz works better as the business model for a site that killed online broadcasts of grandmaster events.

When will @Erik sell to Cerberus Capital?

Follow the money. A few decades ago only the strong players were making money in OTB tournaments. Then, someone got the bright idea that most players were in the C class range. So money was siphoned off the top prizes and went to the D, C, B, and A classes.

Soon, many of the C players that I knew were making more money at chess than I ever had.

I think it's similar in chess.com. I suspect that most of chess.com's revenue comes from kids and their parents. That would also account for their prudish censorship.

I have been doing the puzzles for about 15 years. I remember when getting past 2600 was a real accomplishment. I know this from knowing several strong masters and experts on this site. A few years later, 3000 was doable with a little persistence. Yesterday, I blew past 4000, and now I'm shooting for 5000.

Once all the changes are live, that shouldn't be possible anymore.