Brilliant Moves in New Game Analysis Report

A bralliant move is basically a move that goes deeper then the engine's depth. Such as 35 move checkmate, when the engine can only see twenty moves ahead.

Oof, true enough. I'd saved the game containing my first brilliant move to my library, but now it shows as no longer brilliant. I guess I won't check on my other brilliant moves to preserve what illusion is left.

A lot of people talk about the depth, but it has nothing to do with it but the statistics of each path of the tree. The algorithm doesn't analyze every single movement combinatión simply because there are too many of them, instead, it checks a few sample of plays and recommends a path that has the highest chance of success. A Lot of people talk about how their Brilliant move was the only move that didn't end in a blunder. This has nothing to do with a brilliant move, instead, is the amount of legal plays available, no matter if good or bad. The more legal plays are available at your turn, the higher the chances of you finding a brilliant move (Otherwise the engine wouldn't even suggest a move when you are only with a king against two queens, wouldn't even call it a "best" move when you already lost the game)

This is called The Candidate Moves Technique, (There is a book about it) and is how the masters play, just that the engine can take a bigger sample than the human mind in less time. Basically pick some plays, analyze them deep into the game and rate their chance of success, if none or low, discard and check the next branch (picking a different move and then a sample of consequent legal moves).

According to this engine, a brilliant move is a play that wasn't part of the sample of plays used for the game analysis AND turned out to be better than any of the analyzed branches. It has nothing to do with depth and a "brilliant" move can even  win a match in less than a few moves as a lot of people have already discovered. (If an analyzed branch has guaranteed checkmate in 10 moves is automatically picked as best and you don't need to check other moves. The engine already determined it is the best play because it guarantees a mate in 10... when another move may do the same thing in just 3)

And even simpler example would be: Imagine you have only two legal plays. The engine checks the first one, analyze the entire resulting tree of moves and determined that basically all the plays of that tree are good so it will immediately pick that move without even considering the other legal play, purely because statistically, the analyzed tree shows that all the subsequent moves are going to have good results for the analyzed depth. Then you go, and move the other thing and you have three possibilities. 1st, it is a bad move anyways, 2nd, it is a good move although the other one was still better and 3rd, when it turns out it was ACTUALLY a better move, which the engine rates as brilliant.

This is exactly the opposite as how people think the engine works. The engine prioritizes Depth of as many plays as it can in a limited amount of time (Approximatedly) as of now, there are three engines available: Komodo, and stockfish, which are the ones that can run locally in your machine with a limited sample, and a third one that runs on the cloud (For diamond members) that is the only one that analyzes the entirety of legal plays of the match.

We, instead, rely on our current and possible 2nd moves thinking instead of which piece is more convenient for us. Loosing a queen may be devastanting, and it is even totally discarded by the engines, But sometimes it is that sacrifice that makes us win the match.

There you have another example. Unless it is a trade of queens, there is no way the game actually recommends the sacrifice of a queen, so it just doesn't even analyze the branch.

(BTW if you run the cloud analysis you will never find a brilliant move... because obvious reasons)

(Beware of typos, English is not my first language)

All  this talk about brilliant moves being best only after the computer analyzes at a deeper level is nonsense.  Today my king was in check in the endgame.  The king had to move to one of 4 squares.  I picked the most obvious choice, to get opposition on his king while attacking the rook.  The obvious choice to a sub 1000 player.  It was called brilliant.  

So just to be clear, brilliant means absolutely nothing.  It was probably intended to mean something like what was discussed in the hundreds of previous posts, but it doesn't work.  

here is the brilliant move.  Its a joke.  Brilliant means nothing.  It is a flaw in the programming and should just be eliminated
.  There is nothing better than a best move.  

Please Subscribe and Watch this: https://youtu.be/qWleniWRMHY

idk

why is Kg1 bad in that position?

? the engine saw that... engine already was going to playa Bxg3....

OMG

We played a match and in one of the move h5 on black, opponent lost his winning position and exposed his weekness in the h. We have analyzed this.. See attachment

Its just that stockfish depth isnt gonna instantly find it. Later it will. And for saying computers dont make brilliant moves. First off situations where brilliants moves are played are basicly always after a 'not so good' move. And second off you dont analyse ai matches with depth 18

But in my game here https://www.chess.com/analysis/game/live/19157231475?tab=analysis the brilliant move was the only clearly winning move, so I don't understand how the engine could've missed it?

I got a new brilliant move but the game was not very clean. Also, the move was not very brilliant, it was very obvious.sonalirv vs. mqs1234 | Analysis - Chess.com

A brilliant move is a move that the engine thinks is a blunder, mistake, or in any way a not useful move. However, the engine looks at the next move you make and realizes that the previous move is significant in your favor. Since it only shows the power of the moves you've just made, it can't show that you made a brilliant move unless you go into analysis.

Not true. I asked for a report in a puzzle where the engine was not able to solve before, and the report is considering the move as "Best move".
I'm still not understanding what is a brilliant move, but one thing I noticed is that brilliant moves happens in positions that are not reached frequently. 
This is one example from my games where I tried to use the traxler attack, but my opponent did some mistakes and created a weird position to play. (https://www.chess.com/analysis/game/live/16323851621?tab=analysis)

imo Brilliant moves are when you gain HUGE positional advantage though no material has changed. Brilliance in Chess.com is a lot like kero kero bonito; zeitgeist says it isnt the best, but after it occurs there is significant power in the execution of the existance of it. thus brillisance is the 2nd best which actually is the best. Idk tho, i see others up here checking out top plays comparative to engine analysis (to a 'certain' depth). I reckon it is likely that the engine works out the best move and then is beaten by say a piece of positional advantage due to the non-forced sequence (eg. where although a move is not forced, but every other move actually confers positional advantage for the orign player) is considered best. Whereby the depth calculated by the computer is registered as worse than the best move as displayed by the computer at a given depth. In all honesty, I feel "brilliant moves" are more close to best moves given very difficult positions that a computer is struggling against. All my best moves have happened when it is relatively early and there are many pieces on the board. Full disclousrue, i have had maybe 2 Brilliant Moves on ananlysis and i don't play good chess (i be rated 1000)