Brilliant Moves in New Game Analysis Report

I have been wondering what defines a "Brilliant Move" rather than just a "Best Move" in the new game analysis report ( https://www.chess.com/news/view/improvers-rejoice-new-analysis-feature-is-now-in-beta ).
From my own games,

18. f4 is marked as a "brilliant move" in https://www.chess.com/a/qtjk2dMg


44. Kc6 in https://www.chess.com/analysis/game/live/3467664379


2 back-to-back "brilliant moves" 19. exf6 and 20. Rfe1 in https://www.chess.com/analysis/game/live/3476510047

I would not have anticipated marking any of these moves as particularly brilliant or surprising (though 20. Rfe1 comes closest), so why are they singled out? It cannot be simply a large evaluation difference as capturing a blundered queen does not get marked as a brilliant move. Any ideas? What "brilliant moves" have you found?

I’m curious about the same thing myself.  This move seemed rather obvious (perhaps it’s looking at the leave and forward career of the piece used (sorry, don’t have the vernacular down)?  

https://www.chess.com/analysis/game/live/3534963234

I also don't know what's the difference between the best and brilliant move.  I also can't think what's better between them.

Brilliant move, a move that only is good if you play the only one correct line a few moves in the future. As with the queen move is only good cause you can sacrifice the bishop, if he push the pawn, and win the rock.

Ok

Brilliant move is the best move which the engine can find only after certain depth is reached. So it is difficult to find for the engine. 

Thank you so much for clarifying to me the difference

Brilliant moves outranks best moves.

Best moves are the moves found and assessed as best by the engine (whether you played it or not)

Hence brilliant moves are only made by the human player.

The engine assesses a move to be brilliant only after it is played (can never be proposed as mentioned above).

The engine was capable to assess a move to be brilliant in comparison to its best move when it is played by finding the outcome to be better than the engine proposed (its best move).

Hence a brilliant move is one that is proven to be best only if the engine had gone into a depth that is further than the depth it traverses, but it does not. This is why it can only figure it out after it is played by the human player.

Lol. Wrong. 

Why?

Thank you for your response BTW.

Brilliant move is the best move which the engine can find only after certain depth is reached. So it is difficult to find for the engine. 

You are exactly right.

But if the engine finds it, it is assessed as best. It cannot be assessed as "Brilliant".

Correct me if I am wrong. Have you ever seen the engine proposes a "Brilliant" move as "Brilliant" or as "Best"? I believe NEVER.

If the engine says "Brilliant", it means "I could not find it myself"

Read my statement again and try to understand. 

Read it when you first wrote it and agree with it word for word.

Thank you.

• Jalaal, you’ve been very gracious in your responses to Hikarunaku (which I don’t think is deserved).  Yours was the best answer on this thread IMO, and his responses don’t make sense.   At minimum, he’s confused and discourteous, my guess is that he’s just garden variety troll.

He might have been gracious but given that I clearly explained what a brilliant move is and he still could not interpret it.I was more than correct to point to him that he should read my statement again. 

Both of you are actually both pointing out the most important thing which makes a move "brilliant" - a move that engines would not find as quickly* as other solutions. Be it due to search depth (as hikarunaku said) or being completely unseen until played (What Jalaal was talking about) because of some "horizon effect" (wrong evaluation due to evaluating partial trees?), I think that this is not something to get too heated about!

Ya! @snowcrashed is unnecessarily trolling on this thread without contributing anything meaningful. 

Did not previously feel the need to correct this but let me point out why you are wrong since few people still do not understand:

• Brilliant move is the best move in the position. 
• Engine classifies it as brilliant and not as best move since it evaluated it to be the best move after certain depth was reached before that it was not the first choice of engine. 
• Engine does find the brilliant move by itself  but only above a certain higher depth. 

It could work like that, but there is an alternative way to implement 'brilliant' move:

A move is brilliant if the engine doesn't think the move is best before the move is played at a certain depth. But after it is played when the engine analyzes the next move at the same depth, it'll figure out in hindsight that the previous move was better. The evaluation of the position changes significantly in that player's favor.

And yes this is a useless discussion.