Extending advices #8987
Replies: 2 comments 3 replies
-
Some of this has been proposed and discussed before. Brilliant moves are a very vague and nebulous concept. chess.com's system is pretty bad and quite often feels very random. Often times it qualifies the only obviously good move as brilliant. I don't think there's a good way to automatically detect brilliant moves and it's inherently a very subjective and human concept. I also don't think marking good moves is a good idea. It would just add more clutter and doesn't really offer any value. Any move that's not an inaccuracy is a decently good move. The same mostly applies to "best move" though that at least sounds a little bit more reasonable. But it still doesn't really offer much value IMO. You can already see what SF thinks is the best move very easily and very often there is no single clear best move. Regarding accuracy scores, there's already the ACPL, which feels pretty similar. But this also always has the problem that it doesn't really have much actual meaning. I think the blunder count is much more useful. But maybe converting the ACPL to an easier-to-understand range is still not a bad idea. Adding comments sounds like an interesting idea but also has the problem that it's quite hard to do well. chess.com quite often gives completely ridiculous comments that don't apply to a game at all. I think we can probably do better but again, I'm also not sure if this really offers much value. In the end, you will always only get one of a fairly limited selection of comments and you can usually just get a much better overview from the eval graph. |
Beta Was this translation helpful? Give feedback.
-
I have been recently thinking about this. It is, of course, annoying to get a vague description of requirements and be asked to implement them, but even more annoying that we can't figure this out. Imagine you see a game and you think a move was beautiful: sacrifices, flashiness, great checkmate at the end. You are elated. Until a GM passes by and scoffs: "Ne4 and you lose". There are no universal objective great moves. Yet I keep thinking about defining anything close and I believe it is a good exercise, if nothing else. All my thoughts lead one way: add multiple algorithms that show overlapping charts over the normal computer evaluation. They don't have to pick a move, just show some computed value in a line chart, just like the normal eval. Then individual people will find their interest in the intersections of those lines. Something akin to financial analysis for trading. Everyone has a system, an economic indicator, some lines that are drawn over the stock price line which they claim has meaning. We don't have to find the perfect algorithm, but maybe a combination of them will please more people. Kind of like having both the noob and the GM give their opinion on what move was brilliant and you get to choose which you care more about. |
Beta Was this translation helpful? Give feedback.
-
Hi everybody!! I just want to propose an idea, it's that:
1- we should add more advices in the analyse as 'best move', 'good move', and 'brilliant', and i think that will not slow the server because the evaluation of position is alredy avalaible (except the case 'brilliant' were we should compare the evaluation of our move and the evalution of that computer found)
2- also I think we should add more information to the game as the accuary percentage and maybe a comments about the game that base on the variation of the the evaluation durant the game (exp: if the variation is up all the game (the evaluation is bigger and bigger for white) it generate a saved commentaire that say: "you was leading all the time, your game was incredible!")
I hope guys that my idea was useful, and i wait your comments. A bientôt!
Beta Was this translation helpful? Give feedback.
All reactions