Depth 30 Analysis broken?

I use Game Review at depth 18 by default. Sometimes, when I find the result interesting, I then do a depth 30 analysis. Recently I realized that the result is always exactly the same as the depth 18 analysis. I compared with the results of the self-analysis at depth 30 and found totally different results sometimes. In one game I played a book line and one move was marked inaccurate by depth 18 and depth 30 analysis. I know that it's the best move and self-analysis agrees. Something is going wrong there.

Well, please post the game and let us know the move in question.

It happens apparently with every game, but here you go:

https://www.chess.com/analysis/game/live/33232194383

Move 8.Qh5

The issue with 8. Qh5 isn't related to depth 30 game review. Our book move classification system doesn't have every move played by masters in it, so sometimes we have to rely on engine evals. 

But I believe you that something weird sometimes happens with depth 30 game review. Do you have a specific move in a specific PGN that is showing something weird (like the result being exactly the same)? I want to investigate further, but am having trouble reproducing the issue.

I understand that 8.Qh5 is not classified as a book move. That was not my point at all. The point is that the review gives 8.Qf3 as best at both depth 18 and 30 with an evaluation of like 3.31. The self-analysis never shows values anywhere close to that and also gives Qh5 as best at all depth >=18. Also, this entire game is an example of the result not changing at all after running depth 30 analysis. Even the accuracy values are exactly the same as at depth 18. Other games I tried had the same issue.

Ah gotcha. I see how the eval on Qf3 was unchanged. Will investigate further

Thx a lot!

I recalled another example that might be helpful: https://www.chess.com/analysis/game/live/32474950853,

Here, move 13 is classified as a mistake at depth 18, but brilliant at depth 30. As with the first example, the evaluation doesn't change after the depth 30 analysis finished, but when running the depth 30 analysis without running a lower depth analysis first, the correct result shows up. So it looks like the result of the previous analysis is just not overridden when it should. Note that when the game was played on 2021-12-05, the issue didn't occur, so either the change causing the problem must have been made after that date, or it has to do with me joining beta (I'm not sure when I did that).

FYI: I left beta and the problem is gone.

I think it might have been fixed because the fix we pushed went live. I don't think this issue should be affected by Beta

Then it was a coincidence. Thx!

The problem returned with https://www.chess.com/analysis/game/live/33682864623, where the result of the depth 30 analysis again was just thrown away. Find the correct result of the depth 30 analysis here: https://www.chess.com/analysis/game/pgn/5V1GB93SXQ.

Maybe I’m wrong and the fix didn’t go live yet. I’ll be able to confirm on Monday. Thanks for the update and I appreciate your help with this issue!

Okay so this is somewhat related. When I change depth after having previously run the review, it doesn't actually change the review, just the analysis. How do I get the review to have a higher depth after having already run it at default? Is this some weird glitch related to my browser or does everyone have this happen?

You can't run it at lower depth once it's been completed, but as far as I know, if you change the default to a higher level it can be run.

So basically, DON'T immediately click on review, but first go to analysis, change it, and THEN click review?

Never mind I think I figured it out. Thanks anyway.