Not sure I should trust the Basic Analyzer!

Mate in three called Blunder

Yes. They have been working for several months to fix the disconnect.

Can you give me a link to that game? It should have been able to find that. Crazy... 

https://www.chess.com/live/game/3895406216?username=djbasson

If you look at his screenshot, the yellow highlight is on a line stemming from black's 18th move.

Perhaps this arose after clicking around within the analysis. That line after black's 18th does show the smothered mate.

@notmtwain I see what you are pointing too. The move 21.Ng6+ is in a grey bar meaning it either is a self analysed variation or an engine variation. Only @djbasson can tell us his actual manual actions at the keyboard, but regardless...the moment one clicks on a move the engine begins to work till it gets to d=20 in this case, as he has engine suggestion mode enabled (that shows the graphical move quality icons, arrows etc). Only when it reaches d=20 does a result get returned. The system should have not shown that move as a blunder.

I passed this to a dev to look at it. 

@jdcannon great. The issue is also now part of the thread "fixing new analysis" (@notmtwain posted a copy to that forum topic). It would be good to receive a simple explanation in that and this thread as to why this output arose. Dev have been struggling with this new engine now for months! I really thought they had stabilized it after many reports and incidents like this have been reported and investigated.

Basic analysis is garbage, everybody knows

@congrandolor wow that's quite a broad sweeping statement to make. Now don't get me wrong...I have had many issues with the new analysis system and contribute regularly to the Forum Topic Thread "fixing new analysis" where we post problems and @erik and @dallin see these and see to it that the dev guys get to fix things. And I believe they are trying. What I would want to know from you is...do you base this statement on your personal experience with the new analysis engine (and what have you experienced? Have you reported this to anyone?) or have you had feedback from other people as well about it?  

Because the analysis system should not be "garbage". It should work.

I can't reproduce this faulty analysis. See analysis of game, https://www.chess.com/a/2vESPex2SgQPk and click  on Ng6+ and it instantly indicates best move.

I challenge anyone else to reproduce this.  A screen shot of an isolated, non-reproducible situation does not qualify as an indictment. Click bait maybe,

Said Ng6+ was best for me aswell....

@jas0501 @SorryNotToday thank you for taking the trouble to retest. @jdcannon said the dev team were investigating so it appears this is "fixed" (but @jdcannon has not reported on this so I am surmising). I have seen a few cases where a fault is reported and the link to the game is given and then nobody can reproduce the fault. 

The entire reason for that result is the d= indicator top right showing d=10. That's the cause of it all. It should not be showing that depth no. The move quality icons, arrows and engine feedback is only supposed to show up at d=20! So the engine appears to have prematurely returned a result at d=10, which will certainly give an inaccurate result. Subsequent testing returns the correct value when the engine correctly analyses to a depth of d=20. 

I don't think the issue was fixed per se. The fact that the depth in the image was 10 seems to indicate something was amiss. You can't set a depth of 10. So given freakish circumstances of getting a bogus depth setting the analysis will vary from the depth = 20 results.

Click bait posting trolling for chime in dialogue is my spin on this. If I'm wrong then I expect the poster can present more examples of "garbage" analysis that led them to their conclusion.