Add things like best move, excellent, good, great, brilliant, etc.

Let me give you an example.

You have a position where the best engine line is a forced mate in 20. Next best line leads to heavy exchanges and a completely winning endgame for you. Which one is the best move, the one that leads to forced checkmate or a second one?

But here is the problem, forced mate in twenty leads you into absolutely wild complications that require you to play ten only moves, otherwise you lose. Probability of making a mistake is shooting through the roof.
From the engine's perspective mate in twenty is the best, it sees all the moves. From human perspective, easy win is the best.

So things are not as simple as one would think.

@bufferunderrun said in #11:
> Let me give you an example.
>
> You have a position where the best engine line is a forced mate in 20. Next best line leads to heavy exchanges and a completely winning endgame for you. Which one is the best move, the one that leads to forced checkmate or a second one?
>
> But here is the problem, forced mate in twenty leads you into absolutely wild complications that require you to play ten only moves, otherwise you lose. Probability of making a mistake is shooting through the roof.
> From the engine's perspective mate in twenty is the best, it sees all the moves. From human perspective, easy win is the best.
>
> So things are not as simple as one would think.

Yea, you are right. It is not simple, but I really want lichess to label the good moves I made not the blunders and the mistakes and the inaccuracies. I think realizing your own problems is important, but positive reinforcement can also goes a long way.

chess.com have brought out such feature which means it is definitely doable.

It's probably more of a marketing gimmick than anything else.
Mistakes and blunders have meaning, they are useful, you can learn from those mistakes. Good moves do not IMO. I think we would even struggle defining what a good move is. Your opponent blunders a queen, you capturing it, is this a good move? It is just a normal move. Hard to find move, is it a good move? Hard to find for whom? For you, for me? I'm not sure that adding semi-random indicators adds any value to the analysis.

@blundererxd said in #12:
> chess.com have brought out such feature which means it is definitely doable.

Doable, yes; but I question its honesty & judgement.

On the other hand, it's never flagged any of *my* moves as "brilliant". Hmmm...

the idea is a bit like world peace. we all agree that it's a good idea in principle. it's just that nobody has any reasonable idea how to achieve it.

@glbert said in #15:
> the idea is a bit like world peace. we all agree that it's a good idea in principle. it's just that nobody has any reasonable idea how to achieve it.

It seems like chess.com had some ideas.

@mcgoves said in #14:
> Doable, yes; but I question its honesty & judgement.
>
> On the other hand, it's never flagged any of *my* moves as "brilliant". Hmmm...

Yeah, I don't even expect it to be good, I just want a bit more of an reference which could potentially make me feel better on a bad day.

@blundererxd said in #16:
> It seems like chess.com had some ideas.

yes, some very bad ideas. i have some bad ideas about world peace as well, wanna hear them?

@glbert said in #17:
> yes, some very bad ideas. i have some bad ideas about world peace as well, wanna hear them?
Yes

@glbert said in #17:
> yes, some very bad ideas. i have some bad ideas about world peace as well, wanna hear them?

Merely you opinion.

At least they have some ideas, yet we struggle to even take this plan off the ground.

Chess.com is the number 1 server of chess players (meaning most amount of players are there). I cannot allege that this feature alone made them that famous, but if they are really as bad as you have implied, they won't be the no. 1 server. It is time for us to reflect and reconsider our evaluation of ourselves.

I believe in lichess.org's ability to make the best decision for their developers, for their users, and for chess!

@bufferunderrun said in #3:
> How would you define it algorithmically? It's a very subjective thing.

It's been defined before. One way is to say that a brilliant move is one that has a bad evaluation at a certain depth and a good evaluation at a certain higher depth (especially, an evaluation that is better than the engine's best move at lower depth). We consider a move good when its CPL difference from the engine's best move is small and bad when it is large. Such a system allows brilliancy to be defined in terms of depth and CPL thresholds, which would be tunable parameters.