why is depth 30 glitched so bad?

I am running an analysis at depth 30 and the bar which shows percentage completed is going crazy. It was 20 and then it went up to 97 and then it went down to 23 and then it went to 99 and then it went to 27 and so on. Also, it shows 54 best moves for me yet the game lasted 38 moves.

I guess the chess.com engine has a harder time analyzing precisely after a certain depth? Maybe because it's a browser engine? Just a guess, but I know that other engines also have this problem. However, I haven't heard of many engines going crazy after 30 depth. 

Why are you running stuff at depth 30 is the real question

Cuz its meant to give the best results. Why not?

Why would you need find a move at 30th depth. you gonna join demigod's open?

People cant go that far. Even half-human half-machine magnus doesn't go that far.

u mad bro?

I just ran one without issues. Maybe a temporary problem?

Best results for what? Brute forcing lines?

Sometimes moves which are inaccuracies in depth 26 are "good" in depth 30

Possibly, however it has happened multiple times over a few months.

yes because you totally need to know if a move is an inaccuracy deeper than 26 lines 

If I'm interested in a position I always let it go to at least depth 30.

I use my PC though, not websites.

There are plenty of times I know a move is bad, just by how it looks, but the engine wont figure it out until around depth 30. Or I'll think a move is pretty good, and the engine suggests an artificial looking move, until around depth 30 it will agree with me.

(Of course other times I have no idea what's going on, I'm just saying, it's not about calculation, it's about engines are dumb sometimes).

It's very possible that some positions can result in a shift in evaluation past move 26. It may not be extremely common, but can happen.

The evaluation shifts all the time past depth 26...

Although I will say with this NNUE stuff (e.g. stockfish12) I've noticed it a lot less lately (that I think the engine is dumb or artificial at lower depths).

And actually the evals go so far sometimes that I wonder how confusing it is for new players.

Here's an example

-

-

Most people who are familiar with the Ruy would assume the position after 8.Kxe2 is somewhere between completely equal and a small plus for black.

And an engine like Stockfish9 would back them up, giving an eval of about -0.50 at depth 31 which is even consistent with Kauffman's idea that the bishop pair is worth a 1/2 pawn bonus.

But SF12 bounces between giving this -0.8 to -1.00

Sometimes I even see a position and think "ok black is slightly ahead, maybe a win between GMs but it will be a grind" and SF12 is giving crazy numbers like -4.00

For new players trying to learn I have to imagine evals like this are confusing.

Maybe in the future we'll see recommendations for lower rated players (maybe anyone below professional level) to use weaker engines so that they can see evals that make more sense. Or maybe we'll have guides on how to interpret it when stockfish15 is showing -8.00 in a position that most GMs would consider equal

SF13 is less enthusiastic, bouncing between -0.6 and -0.8

wow engines are weird