Brilliant Moves in New Game Analysis Report

It seems that a definitive consensus has yet to be reached as to what empirically defines a “brilliant move”. This thread is primarily conjecture, educated guess-work and egoist defensiveness. BUT what I think we can ALL reach a consensus on is that “hikarunaku” is the most obnoxiously abrasive, pedantic pontificator I have ever had the displeasure of encountering. It is a testament to this communities maturity that this man was not chased off the thread with torches and pitchforks. I commend you all.

bruh

Brilliant moves are where the engine think it's even better than a best move or that you moved a very good move without the engine knowing it or finding it.

Hey, in one of my games I got a move the engine couldn't find but it wasn't marked as brilliant.Why??

https://www.chess.com/analysis/game/live/21024052229 Who knows why this is a brilliant move

Chess.com seems to have added a "Great Find" move category with a single exclamation mark (!) with a light blue color in addition to "Brilliant" Moves with a double exclam (!!).

Look at my analysis not in Beta mode.

ok

My opponent did two brilliant moves in a row. But both were bad

Hi

Have been trying to figure out why d5 was a (!!) .. seemed to be just capitalisation on opponent's weak move previously .. unless I'm missing something deeper ?

Rgds

I've only had one move marked as brilliant by the post-game analysis.  In that situation, I was in check and only had 2 legal moves.  I chose the better of the 2 (which seemed fairly obvious) and the analysis called it "brilliant".  I found that really surprising.  Perhaps just a bug? 

It's usually a move that is hard to find, like a queen sacrifice. OR, it's a move that is considered brilliant because it allows a great play afterwards, but if you don't find the play as well it usually becomes bad.

Like me in the Stafford.

What

Came here because I was curious but what I found was a comment fight..

I was playing against a 1400 bot for fun and this happened:

FEN: 8/8/5p2/5Pp1/8/4K3/8/6kB b - - 0 56

For those not using Beta, this is the method brilliant moves are assigned from what I've seen.

My copy-paste message for people asking about brilliant moves in the chess.com discord:

Here's the explanation behind 99.999% of brilliant moves.

From what I've seen, brilliant moves need to fulfill 3 criteria (at least)

1. The engine only sees the move's continuation after a decently high depth, this doesn't mean the engine can't find it, it means the engine doesn't find it on low depth.

2. The move needs to be the only move in the position that draws or wins.

3. The move needs to have more single continuations after black makes their move. Black can have multiple good continuations, but if after at least one of them white again needs to make an 'only move', it'll at least have a high possibility of it being classified as a brilliant move. You'll find it being classified as 'only moves' in your annotated pgns after a game report, it'll say {critical move}

Because of rule 1 and 3, some moves that we'd call "brilliant moves" aren't brilliant on chess.com. It actually needs to see the continuation to be called brilliant on chess.com. If it's so brilliant that the engine doesn't spot it, it won't count. For those using the old Game report, if you've played a move the engine doesn't find but it doesn't say it's brilliant, analyze on a higher depth than 18, it'll say it's brilliant if it correctly evaluates the position at depth 18.

The same thing applies for rule 3, a great "only move" positional idea doesn't always warrant single continuations afterwards, but it could lead further down the line to a better position.

Also, on a side note, since I've seen a lot of people talk about this,

Previous moves made don't matter, it doesn't look at any history, just at the current position and the move made. It may or may not check the local history of assigning brilliant moves (so it doesn't give 3 brilliant moves for a dumb repitition), but I haven't found enough evidence for that yet.

For those using beta and the new brilliant move algorithm, here's how the new ones are assigned from what I've seen.

1. Sacrifice anything, just anything. The more value, the better. It needs to be the best move (or close enough) in the position, it doesn't matter that much what the follow-up is, just sacrifice. It doesn't need to be an 'only move', it doesn't have to have single continuations. It just needs to be semi-decent.

2. It still does count moves which it can't find on low depth, but they dialled up brilliant moves on sacrifices by 10. 'You made a cool pawnbreak?' Yeah that doesn't sacrifice anything.

'You gave up your bishop for a pawn so it doesn't become a queen next move?' ABSOLUTELY BRILLIANT.

3. It almost definitively doesn't check history. Like at all, repetitions give the same brilliant moves over and over again.

Overall, this new beta algorithm is somehow more garbage than the previous one. It's WAY too generous, gives brilliant moves on every bishop opening sacrifice on f7, most sacrifices in endgames, you get the point. Very little importance is placed on actual good brilliancies, positional ideas (which are hard to code for, I know) and actual forced-ish combination. Some random nonsense sacrifice shouldn't give brilliant moves, while an insane silent move should.

Rant over

engines aren't perfect