9x9 confusion

I am super confused on how to use this move F6 stone on move 6, It says it doesnt lose any points so its a really good move in the opening but when they respond with hane on G6 I cant play E7 without allowing myself to start losing the match , it says to play D3 but why did I even play F6 if im going to allow him to be able to play E7?

I hope this makes any sense, I just dont understand what i want to do with my F6 stone if i cant play e7

I am super confused on how to use this move F6 … I just dont understand what i want to do with my F6 stone

So why did you play it?

1 Like

Because I understand its a really good move lol, but i am asking for help on learning how to use it better

If you don’t know why it’s good, then why are you sure that it is?

1 Like

Iol ok i guess i dont, maybe we should see what other people say.

The AI says its a great move

The AI said it lost 12% winrate actually, so not all that great.

The AI says its a great move

No, the AI doesn’t say it’s a great move, according to the analysis in the very same game you linked. It just says it loses barely anything.

But just because a move loses barely anything in points doesn’t mean it’s good. Wasting a small ko threat also usually barely loses anything in points, because the chance of a ko coming up where a small ko threat makes the key difference and where there aren’t other similar-sized threats or an alternative way to play - isn’t that big. But wasting a small ko threat for nothing is still a bad move even at times when it doesn’t lose much.

For the case in the game:

Indeed, analyzing a bit deeper with AI offline - it doesn't really like F6, but if black responds defensively, it doesn't lose much, as long as white doesn't make the mistake of trying to follow it up with a second move.

There is a small benefit to making the exchange, which is that if white played D3 directly, black would attach at D7 to reduce white’s corner and build the top. After white makes the exchange wF6-bG6, black can’t attach at D7, because then white’s wedge at E7 would be more severe and leave black with too many cuts.

So if you have to give a meaning to the move and why it doesn’t lose a lot of points, despite making black’s corner stronger - that’s the answer apparently. But because it does make black stronger and give up other possibilities by white for how to play, it’s not even the favored move.


When you say

I assume you mean the AI review.

Actually, it does lose 0.4 points compared to move 6 at e7 which really does lose zero. But in terms of win percent it loses 12% (again, move 6 at e7 loses 0%), which could even be considered a mistake. I guess the margin in terms of points is smaller because it’s 9x9.

I think this is because AI didn’t like f6 in the first place so when it suggests D3 it is saying “f6 was not so great, but having played f6, you’re now best off playing d3”

Ok i see , I have the AI set up to tell me how many points a move loses or gains , not winrate %… so when it says my stone is worth -0.4 i figured it wasnt a problem… since I usually cannot ever find moves that are perfect play.


I guess it might seem obvious to other people that it was a bad move but I didnt realize -0.4 was such a problem. Thanks again for helping

1 Like

Yes, I usually look at points as it’s more intuitive and on 19x19 I don’t usually worry about 0.4 points and similar small losses. But, like I say, I guess the margins are smaller in smaller boards


No need to apologize. I found AI review quite confusing at first. Still do TBH :laughing:

I guess it might seem obvious to other people that it was a bad move but I didnt realize -0.4 was such a problem. Thanks again for helping

You learned the wrong lesson here. :slight_smile: The lesson isn’t that -0.4 points is a problem. The lesson is that the number of fractional points the AI says a move is good or bad by, tells you not so much about whether the move is good or bad for someone at beginner level.

If you really want to know the specific reason why it’s only 0.4 points (and why the AI thinks even that Black should have responded differently) - see the hidden section in my post above.

But if you read through it - it so far above beginner level as to be almost useless. I’m a dan player and it flies way over my head - I wouldn’t think of that in a game, and even knowing it now, I would still have no idea when in a game to apply that tactic or not, and how to tell if it was worth it in a new situation.

And as in the example of a wasted ko threat - the number of points a move loses says very little about how bad it is, in terms of learning. Yes, moves that loses fewer points, are in some objective sense, less bad, but you can have a move that loses 5 points that even a pro would not be able to judge and would consider plausible, while on the flip side you can have a move that loses only 0.2 points on average that even a beginner can learn is pointlessly wasting a ko threat or is giving up a tiny bit of potential for no gain, or otherwise doing nothing useful.

Review tools that encourage users to sort mistakes by size in points and give no other aid I think are to some degree doing their users a disservice for this reason. It’s great for finding big blunders, but the magnitude of something in points often doesn’t match the simplicity or complexity of the concept behind a move, or how learnable it is.

Get your games reviewed by human players, and ignore the AI, at least until you’re a lot stronger, except perhaps to find big blunders where you yourself in retrospect do think you can understand why once it’s pointed out.


I find it hard to ask someone to review my games because I feel like the only way i could actually improve is to find the mistakes on my own and i dont like bothering people because usually i get frustrated because advice usually isnt the type i want in the first place. if that makes sense… Anyway thanks again for helping me here I do think I understand why that move wasnt very good, its just a bad waste of my tempo during the opening , it helped him create territory and left me open to attack aswell.

Thanks again i hope I can be as good as you guys one day :slight_smile:

1 Like

I think people are being a bit harsh to you above to be honest.

Like this feels a bit harsh. Just play the moves you want to play, and if you’re not sure how to use them, play them 2,5,10 more times in a game and decide if you like the outcomes :slight_smile: I think the AI can be good at suggesting ideas, in the same way other people can. No guarantee either an amateur player or low playout AI’s suggestion will work out or be playable at our level, but sure try it anyway :slight_smile:

I also tend to ignore winrate, 12% doesn’t really mean much to me.

Here’s the same game with a level 2 OGS review. It changes form -0.4 to -0.7 points and from 12 to 14% points in winrate.

I don’t think it’s necessarily a bad move, it kind of helps white’s shape in the middle in exchange for Black getting solid points on the right.

The other option is to try to crosscut after Black hanes, but what it achieves is probably unclear. It’ll most likely be a sacrifice of cutting corner stone, but if the stone/group can get 2,3,4 liberties inside Black’s territory, Black might be forced to capture them. In some scenarios maybe they can help contribute to a shortage of liberties on the cut groups.

I think what hexahedron said is probably fair though and something to keep in mind

One might need to run analysis deeper/for longer in order to get a proper answer to whether a move is ‘good’ or not, especially in complicated situations.

There is a margin of error on these things too. If you run the review again it might throw it up as a -0.3 mistake or a -0.5 etc. The more playouts (the higher the level of review on OGS) the more certain it can be, but one might want to look at it offline or elsewhere that might run higher playouts in some cases.

On the margin of error note, sometimes when using the interactive analysis even clicking only score 0.0 moves the score estimate can still change in favour of one player or the other :slight_smile:

I’ve had cases where you follow the level 2 review with the interactive analysis and while maybe one move loses -0.4 or 0 points, over the course of a sequence it adds up and you realise you’re down 5 or 6 points.

1 Like

I agree with the bot’s tenuki, since it looks to me no local continuation, which mainly refers to tiger, extend or cut, leads to an interesting result for white.

To understand why it’s still a good move, maybe another similiar scenerio would help. Say both players are fighting for a ko, which is the last endgame on the board. White has 5 more ko threats than black. Now it’s white’s turn to take the ko but white wasted a ko threat instead. This doesn’t change the result since white has a lot more ko threats, but wasting a ko threat doesn’t help white as well.