Review request

Hi. There were a couple of things I messed up here, but I feel specially bad about my handling of the invasion that starts at 69: Swaggy_P vs. lacrymology

I got an AI review, and the recommended moves, but following that goes way over my head. I need a human touch to understand the principles, rather than just following a string of moves :slight_smile:

Note: my ranking is all messed up, IMO. I havenā€™t played in a long time, and Iā€™m not sure how I got up to 9k, but Iā€™m probably more around 13 today.

Thanks

2 Likes

I left a quick review in the game. The main points are:

  • At the beginning of the game, if all groups are safe, play elsewhere. The board is big. Adding stones to a living group is small at that stage.
  • A moyo can rarely be converted entirely into territory. If you build a framework with N intersection, expect to make N/2 points.
  • Therefore, unless you are sure that you can kill, donā€™t make bad shapes (i.e. shapes with many cutting points). Attack from a distance and make solid shapes.
5 Likes

Thank you. I feel that helps. Letā€™s hope it does :slight_smile:

1 Like

Moves 11 13ā€¦ shows that you are not aware of how crucial are your own liberties.
One way to consider Go is how to keep your stones breathing with as many liberties as possible.
Check ā€œempty triangleā€ on the sensei library wiki.

Thanks. I donā€™t know if Iā€™d put it quite like that. Rather than ā€œnot aware of the importance of libertiesā€ I feel it more as ā€œnot aware of the danger of cutting pointsā€: I honestly thought that the cut at E17 was safe for a few more moves, if I answered at E18.

At least pushing down the 5th line like that felt pretty good. I saw his shape was more solid, but I was pretty happy. If I had fixed the cut at 20 instead of overreaching, wouldnā€™t you say thatā€™d been okay?

My bad, my comment was aimed at blackā€¦
So you played well on this side (white) until G18 which is far too slow and small. You could instead play in many other places. Later tenuki at R12 is doubtful, if black hane there he will ruin your shape

Sorry for blackļ¼white confusionā€¦

1 Like

If you have questions about the AI review, I donā€™t mind trying to help you understand it or answer the questions if I can :slight_smile: I can run OGSā€™s review on the game also, and you can ignore it if youā€™re not interested in it :slight_smile:

Breathe deeply and repeat: ā€œIā€™m not trying to kill! Iā€™m not trying to kill!ā€
Better with Dwyrinā€™s voice. :smile:

Harassing that invasion could be more profitable than trying to kill and fail.
Keep it small while building against your walls.

Donā€™t worry: we all got a little bump when OGS ranks were revised. Iā€™m 9k IRL and 4k here.

1 Like

I donā€™t quite get the OGS AI review at all. I mean, yeah, I can see what were the bad moves, and what would be better, but I only got likeā€¦ 4 of those? if there was somewhere to see how to interpret it, it would be great.

Anyway, I meant I loaded the SGF into KataGO locally, and saw through that. The problem with the AI approach is that itā€™ll usually point you to ā€œthis move was biggerā€, but itā€™s not so great at seeing how itā€™d work locally. I know Iā€™m not a genius at evaluating the biggest move on the board, you know? Iā€™d just like to get better at fighting at this point :slight_smile:

Full AI review is only available for OGS supporters, though if your opponent is a supporter you will also benefit from the review.

Itā€™s a small perk, but as you mentioned you can just load it in Katago yourself anyway.

I think katrain has a feature that you can force the AI review to only consider local options if thatā€™s of interest to you. Itā€™s a different gui+ for running say katago

As @qnpnpmqppnp said it only shows a few top suggestions and variations for non-supporters.

But as I said, for that game specifically Iā€™ve now also run the level 4 review on that game, since at least if thereā€™s some questions on it for instance, weā€™re both looking at the same thing :slight_smile: Of course you can just share screenshots for instance too.

That said though, a lot of the time in the opening, knowing that a move is bigger is a pretty good way to improve. I think when I was moving up in ddk ranks, just knowing to play away a move or two earlier already can gain you like 10 points or something.

Case in point, move 34 in the game Swaggy_P vs. lacrymology

Of course human reviews are better are being able to separate out these things, and explain in words. So locally here one might expect something like

local continuation for black

or if you want maybe, but it could depend on the point of the game and whether one wants sente say

Or the monkey jump might be the best as a reduction, but again in theory one is supposed to ignore this this early in the opening. Itā€™s not really gaining any points for black

But anyway, if you have some specific questions about using the ai review, it is something Iā€™d like to explore on how to explain better, and make accessible to players of all levels.

I review every game I play with AI, even some games I just watch or are for fun, like in person rengo :slight_smile:

1 Like

I think katrain has a feature that you can force the AI review to only consider local options

Yeah, Iā€™m using Katrain. Forgot about that option, tho, thanks

That said though, a lot of the time in the opening, knowing that a move is bigger is a pretty good way to improve

100% and I get that. Honestly, at least in my mind, I, for the most part, only kept playing there because my opponent kept responding, and it was going well for me, IMO (reminder: Iā€™m white in this game :slight_smile:).

I definitely got carried away, the hane after the jump was bad, but the question the AI isnā€™t answering to me is ā€œwhen should I have defended the cut at E17?ā€ Now I feel move 20 was unforgivable, and as long as I wasnā€™t going elsewhere on the board, that was the time to defend. I think at the time I was annoyed I had to do it in gote.

1 Like

The game was going great for you.

What a nice moyo. Opponent who came into the center of it really deserved to die. Moyos arenā€™t especially popular these days because itā€™s considered too easy to destroy or reduce them. But in practice people die all the time in them, even low dans, I recently forced even 3d and 4d to die. Sometimes even EGF dans die. But good shapes in killing can be hard to explain, a lot of them are intuitive.

1 is often good shape for attacking, one space from the lone stone. Closer and it might strengthen black, further and itā€™s not as severe.

Always peep at everything. You want your opponent to have heavy eyeless shapes. But always remember that opponent might ignore your peep. Or extend instead of connecting straight away. White three stones arenā€™t in danger, the only thing to worry about is getting squeezed but white is strong there so free moves black places there face pretty strong white so it should be ok. If you think about it, it would be pretty sad if white had to defend there. Theyā€™re so strong and still have to defend?

Black making this shape is a bad sign for white. Itā€™s becoming close to alive now. Note that thereā€™re no white stones around it so it has place to expand. Thatā€™s an important quality here, not just empty intersection inside of it. If it was more surrounded by white stones, the shape would be much sadder.

This feels so overcommitting. Ideally you want to kill opponent in easy and fun way. Where at any point, if things go south, you can back off and be ok with the result. Which is why AI might suggest playing A. After all, black is not alive yet and white does solidify area towards the bottom. But the move you played commits to killing, it gives black a bunch of free moves like Black A. And like if black lives then you wouldā€™ve liked to have spend this move elsewhere. So itā€™s like youā€™re saying ā€œI definitely kill you, black, this move is crude but it doesnā€™t matter since youā€™re dead now, you just donā€™t know it yetā€. This type of committing move shows that situation is slipping from whiteā€™s control. Though white is still pretty strong all around, I wouldnā€™t be sure I could live with black, maybe itā€™s still ok.

And so on and so forth.

Fighting is hard because itā€™s so involved. Thereā€™re a lot of considerations, shapes and reading. I suspect some teachers avoid going into it. I often see lessons skip through fights with only basic commentary (well, maybe live commentary go into it because they have lots of time). But itā€™s an interesting topic. Itā€™s fascinating to see a pro show the actual reasoning for their moves with dozen-move follow-ups.

Good practice for this type of fighting is shape game. Black has stones on all first line intersections and has to kill all white to win, white has to live with something to win.

Also Antti made a lecture on something like that.

3 Likes

I think my example was that white lost maybe 10 points by playing one too many times in the same area. Itā€™s hard to visualise that ten points, and usually X points that the ai says. Probably a good bit of the time itā€™s some kind of average. Itā€™s hard to see where those ten points appear exactly from the high enclosure, because the determined opponent could probably live somewhere, but presumably g18 only really secured like 3-4 points, H18, H19, J18, J19, so the enclose just has to sketch out more than that.

I think the funny thing seems to be you donā€™t have to. The wall you get for sacrificing the two stones is supposed to be worth more in influence, or eventual territory than the two stones = 4+ points.

In terms of how to ask the Ai this question, if it shows up as a suggestion, thatā€™s the time to play it :slight_smile: If not then just keep playing the cut for black at different points and see the response.

I guess you definitely defend it when four ish points in gote seems big, but you could end up also naturally defending it if you have some low stones on the side.

Gives up the two stones



Actually funny enough in terms of naturally defending the stones on the upper side, it works much better if you have a solid connection at F12 over the tigers mouth at F11.

Imagine a stone at around J18 and compare some variations of trying to turn on the second line


In terms of how you might naturally end up with a stone around there hereā€™s a random variation

But even then it might still be better to just give up the two stones because of the other cut at F14. You could end up being reduced a good bit in that area.

Itā€™s kind of hard to decide.