[SUGGESTION] AI Summary table must take in consideration Winning %

OGS Development
1

I suggest this table to consider winning % too to consider a blunder, and not only points. Why?
Context: I have a game Where at some point I were winning and just decided to play endgame and safe moves. But because the suggested move is considered big, all the turns I don’t “punish” is considered a blunder. I’m not sure if this is correct at all. On my mind, If I’m making moves because I know I’m ahead an just closing the game, and my moves don’t harm my winning %, should be considered a blunder? Maybe this should go to something like (Losing points moves) or something like that. BTW, maybe could be more interesting to base the criteria of the complete table on %?

What do you think?

2 Likes
2

Sounds very reasonable.

3

How can you distinguish between a player that’s intentionally playing safe so it’s not a “blunder” by this classification, vs a player who doesn’t realise they could punish a certain move but happens to be winning?

I see it more useful as a teaching tool to tell me (and quickly to be able to jump to) moves where I could’ve played better.

I use the graph to quickly jump to mistakes in the game, but probably with a bit of practice I can get used to using the table to jump through big mistakes

1 Like
4

Well, AI can’t know this, but a movement who lose % and points, is what in my mind means blunder, and is not the same as a movement onlye losing points. At the end, you have to win by 0.5. I understand what you mean for this I suggested if a movement don’t loose % but loose points, enter in another category different to blunder

1 Like
5

Yes, moves that lose points and gain % should be considered as good moves.

Note: conversely, there are moves that gain points and lose %, these should also be considered as good or great moves.

1 Like
6

But if you’re up by 20 points, nothing really loses percentage except a big blunder. Any move that could lose less than like 10 points towards the end of the game could be considered like a “safe” move.

It would make the categorisation fairly useless to improve your endgame for instance, unless the game was close.

Or in a handicap game, none of your moves will be mistakes until it starts getting close to even, at least to a point where katago might have a chance of losing the game.

I mean you could split it into mistakes while winning vs mistakes while losing, but doubling all the categories wouldn’t be great.

You could add a filter to throwaway mistakes while winning entirely.

I probably wouldn’t use it personally. I like the fairly objective nature of “there probably was a way” you could’ve made/saved/ destroyed a few more points regardless of whether I was 20 point ahead, 10 points behind or it’s a half point game.

1 Like