Brilliants moves (!!) and others (Good, Book, Dubious) for game analysis

Hi,

First, thanks for developing Lichess! It's awesome.

I've seen this topic was discussed a while ago, but I want to bring it back.

The current game analysis tools only show blunders and misses. My proposal is to give users an analysis level similar to the move types chess.com has. They've definitely set a standard here.

The request:

• Show other move types like Brilliant, Good, Book, Dubious
• Rework the color scheme used to represent them

I'm a designer, and would be happy to help create the icons for these. Attached you'll find a mockup:

• The hue changes naturally describe the type of move.
• To avoid any chess.com claims, the Brilliant move color was changed to purple, which is more logical and connotes the same value.

Replying ahead, I've seen the main reason this hasn't been implemented is because many say "there is no such thing as a brilliant move." I understand the abstraction, but I want to point out a few things:

• Annotating games with icons like !! or ! is standard in chess. Check this article: https://en.wikipedia.org/wiki/Chess_annotation_symbols
• Saying "then go to chess.com" is worthless. My point is to make this fantastic platform even better.
• This isn't just an ego booster. When you're learning, understanding bad moves is great, but it's also good to know when you did well. It's pedagogic.
• The whole chess community uses "Brilliants" to quickly showcase their skill. Nobody says "look, I haven't made a single blunder" because that's boring.
• The chess community needs these fancier ways to show their games and attract viewers. You won't tell me you've never watched a game showing off Tal's brilliancies.

I have experience with other open source projects, like Blender (3D software). Blender used to make you select objects with right click. It was one of the most counterintuitive defaults out there. Do you know when Blender started to grow? When it changed the standard: the moment it let you select with left click. You can still select with right click, but it's no longer the default.

I leave also some links as reference that may help on implementing the algorithm for them:

• https://support.chess.com/en/articles/8572705-how-does-game-review-work
• https://www.firechess.com/blog/chess-brilliant-move-explained

I'd really appreciate the devs considering this feedback, as I believe it would have a Brilliant (!!) impact on the Lichess platform.

@Axaet said ^

Annotating games with icons like !! or ! is standard in chess.

Sure. And this always represents the subjective opinion of an expert This is the crux of the problem. Lichess is not an expert, and does not employ an army of them to annotate games. Instead, It uses computer algorithms to do things. So one would have to design an (objective) algorithm that acts like a (subjective) expert. This may be impossible. It is at best very difficult.

chess.com tried, and probably failed (I believe I've seen videos of GMs.mocking the analysis). But chess.com is not interested in doing the mathematically correct thing; they are interested in giving the (mostly young, inexperienced) users the sense that they are getting what they paid for. If the rating and advice is sometimes way off base, they can probably get away with that. But it doesn't mean Lichess thinks it is wise.

I'm sorry, but I don't agree with you @mcgoves.

I really believe that a couple of pragmatic GMs/Coaches that know data science and programming, can easily figure out a quite solid algorithm to determine these moves.

Just to pick a single analogy, this is the same as color profiles. There is a lot of theory based on human perception, and there are different color profiles to represent them (sRGB, ACES, CIE, AGX, etc). Are all of them 100.00% objective? No they are not, but they are practical enough to cover the color needs + they get polished from time to time.

Creating a good algorithm for chess is much more simpler. And remember that it is a game, the algorithm doesn't have to be 100% infallible.

Having a practial analysis that works 80%-90% percent of the time is like having a very good coach telling you which things are ok and which aren't.

An algorithm developed by some top GM's can be an extremely valuable tool.

Relying merely on simply knowing if you blundered or not is not pedagogic nor encouraging for those who are learning, and it slows down the initiative that the chess community is having currently.

For me, creating a solid pragmatc algorithm could be actually simple for the experts.

No, there is no simple algorithm for defining what is brilliant. Even agreeing upon what would be the target would be hard let alone coding it. Like chess.com first has some more complex algorithm then they found out it is not very good. And then the current definition basically piece sacrifice that ends up winning material. That is objective for sure but brilliancy always ?

Lets face : there is no generally agreed definitio what is blunder. And that is to some degree more objective problem.

"there is no simple algorithm for defining what is brilliant"

• There is: a move that:

1. wins,
2. is the only move that wins,
3. involves a sacrifice, and
4. is a quiet move, no check (+) or capture (x).
   It is easy to code.

"there is no generally agreed definitio what is blunder"

• A blunder (??) is a move that turns a winning position into a losing position.
  A mistake (?) is a move that turns a drawing position into a losing position.
  A missed win (?) is a move that turns a winning position back into a drawing position.

All moves that are neither mistake (?) nor blunder (??) are good moves or correct moves.

Book moves can be good, mistake, or blunder.
A book move is a move that has been played before.

A dubious move (?!) is a move for which the commentator was too lazy to find out it it is good or a mistake (?).

I still question why people want brilliant moves so badly...

All you need is a notation for “mistake”.
An engine move is “good”. An engine move can also be “brilliant”, depending on the situation. For both a notation isn’t needed, that’s chess.com-ish.

@ResurgenceOfVoid said ^

I still question why people want brilliant moves so badly...

As OP said, it encouraging for people to see they did good. I moved from chess.com to lichess after learning about the technical process of making it (some yt video) and chess.com started being very clunky at times. Brilliant moves, even if more of a gimmick, is one thing I miss, especially bragging to friend like "look here!!". I understand this might be frivolous, but would be (imo) a nice addition and maybe a reason for others to come here.

@tpr said ^

"there is no generally agreed definitio what is blunder"

• A blunder (??) is a move that turns a winning position into a losing position.

sure, that nice definition and very objective. But not generally agreed on.

Turning winning game into draw can result from not seeing 7 move combination. a mistake for top GM but hardly so for average lichess player. Under normal circumstances he/she will not find it even using all the time available.

For practical purposes blunder is:
A move that overlooks an immediate tactical consequence—such as a fork, pin, discovered attack, checkmate, or hanging piece.

Also you brilliant move would almost match what Nunn used in his endgame books. works there well. But for move being brilliant there wont be something people agree upon

at end of the day languages are set contract between users on language and people do different ideas what has been agreed.

"a mistake for top GM but hardly so for average lichess player"

• A mistake is a property of the move, not of the player.
  The average lichess player makes more mistakes than a top GM.
  A top GM makes more mistakes when he plays bullet than when he plays classical.