Saxmaam's Slow Go

In the review link above, “opportunity missed”, I can see text saying “Variation: black invasion”, and clicking will take me to move number 46, but there is no actual variation or commentary, or anything I would consider to be an actual review. Am I missing something here?

Looking at the game, I stand by my comments above on whole board thinking :slight_smile: If you want to add some comments to the review – what you were thinking about during the game, which moves could maybe be improved on, which of your opponent’s moves surprised you – then we can have a conversation about it if you like.

Another suggestion: at the Go Teaching Ladder, look through games by people at or just above your level, see what types of mistakes they make, what questions they ask, what answers the reviewers give.

well, first I labeled the invasions backwards. other than that, if you go forward from those shared variations, you can see that I made an invasion that barely lived, and my opponent made an invasion that was rather more successful. i labeled them so I could put them side by side and compare.

that opponent just steadily outplayed me. it seems like i got through the opening ok (maybe not). and then after move 20 was a gradual erosion. feedback is welcome.

I did share a missed opportunity link with an ai playout in the original game (not the komojo review). unfortunately, I like the way the ai works better in the original game than in the review page. at least when I’m reviewing by myself

I’ll malkovich a game and share it afterward soon.


1 Like

and re: the “not a review” remark. I’m experimenting with using some of these links as flashcards. so you’re right. definitely not a review but useful to me regardless.

I join on this that it’s more productive to work on a failure you made in 20 games as 20 failures in one game.
And to detect this, a stronger human player may be more qualified as an AI.


Ah, flashcards! I get it now. Nice idea :slight_smile: Sorry for misunderstanding.

The difference between those two invasions is that white went in too late and too deep. Black invaded a relatively open area with strong friendly stones nearby for support. So it’s not just about finding the right followup moves for the invasion. It’s more about the strategic judgement: not letting black enclose such a big area in the first place, cutting it down to size a few moves earlier.

1 Like

Reminder: area with 2 access is not that big. Area with 1 access suddendly getting much bigger.


I might need a bit of context to understand that. :grinning:

It’s a reference to the proverb Invade a moyo one move before it becomes territory.

(That page is a bit disorganised and needs editing! But you’ll get the general idea.)


Moving according to this proverb is definitely something I need to get better at. It has been one of my main downfalls in even games with slightly-higher-rated players (the other being that I’m bad at reading late mid-game invasions, both attacking and defending)

March 2024

3/6. I finally played IRL with people who know how to play go. It was great! Go club meets late for me, but in this case it was worth staying up with the grownups. First game I took a 9 stone handicap, and second game I gave a 6 stone handicap.

I’ve been playing with AI SENSEI lately. It certainly doesn’t replace doing traditional tsumego. Reading from a couple of books when I have a little time. Busy lately.


April 2024

4/8 This last couple of months I’ve been experimenting. Talking with a friend about “learning how to learn”. I can definitely spend a lot of energy in improperly targeted study and wind up gaining little. I have managed to rank up 1 stone — from a solid 13kyu to a solid 12kyu — in the past 8 months. I think. Noise in the ratings is a thing.

AI Sensei made a huge improvement at a time that I was struggling to make good problems. It now accepts+suggests multiple correct answers when appropriate. This was such a huge game changer that I became a subscriber. There is a learning curve to making problems from my games that I can actually learn from. There’s a learning curve to understanding what the AI is telling me, too.

I reviewed Graded Go Problems for Beginners volume 1. That was pretty tedious, but I did put a select few problematic life and death problems into AI Sensei. I’m just not that good at reading.

And I’m reviewing Janice Kim’s second book now.


That’s quite a good start imho. Reading is quite fundamental in our game and whatever is your interest in other parts of it, it’s almost like compulsory to even enjoy reading.

What you do for it is a good start too, so i want to encourage you to sone regular practice (exercises) like a few hours a day, seriously searching to read and find the solution starting by easier and evolving to more difficult.

Then and not before, you can diversify to other subjects (mainly checking the fundamentals)

1 Like

Working with AI Sensei for the last couple of months has caused some changes in the way I play my correspondence games. On a bad day, I’m still likely to play reflexively and miss big threats or opportunities. But on a good day, I see the correspondence game move as a whole board problem, which is an improvement.

It occurs to me that there could be a whole new class of problems wherein the solver is asked “what is going on here?” Is black trying to expand his boundaries? Or is it a capturing race? Is black’s big peninsula about to be captured? Is the group you’re trying to kill actually be about to swallow you?

Well, I guess that’s really what the whole board problems are about, whether it’s stated that way or not.


And miai! So much of what I miss relates to miai opportunities or miai threats.

1 Like

More thoughts on working with AI Sensei.

For the most part, lessons I want to learn from AI Sensei are not memorization lessons. Ideally I’d approach those problems (the non-memorization ones) with a fresh mind. I don’t want to think “oh, it’s THAT game, I should play a clamp right now at this point.” But SRS is designed for memorization.

Fortunately the platform makes it easy to push a problem into the future by 4, 7, 14, or 28 days, regardless of whether I succeeded at solving it or not. I’m using this more and more.

1 Like

AI Sensei

Another challenge trying to make good problems and learn from the AI comes from the cognitive load of looking at a whole board from a fresh perspective. I don’t worry too much about making great problem statements when I first make a problem. But as a problem is presented and re-presented over time, I do try to make these useful.

For example, just reading the situation is a variable skill for me. One day I realize that “white just cut off five black stone in the middle, and black needs to choose a new direction of play”. The next day, that situation might just look complicated and ambiguous to me. So it’s useful to put that understanding into the solution part of the problem to remind me on those fuzzy days.

another use for the problem statement is to turn a whole-board problem into a local problem with a hint. EG “find the best local move” or “what can you do about white’s weak group?”. Not every problem needs to be a whole board problem. It depends on what I want to learn from it.

Right, the problem is: do I play a move to try and save those five stones, or do I leave them as dead and play elsewhere? (Or, at a more advanced level, do I leave them along even knowing there’s a way to save them, because playing elsewhere is bigger or more urgent?) This one aspect of the whole-board thinking were talking about a few weeks ago.

It’s hard to encapsulate this sort of judgement in a tidy problem position. Most of us learn it slowly through lots of games and a few game reviews. I wish I had a good answer to this :slight_smile: AI is getting there.

I’m still learning from AI Sensei, and learning to learn from AI Sensei.

Reading seems to be a weak part of my game. I have concluded that if I can’t read a capturing race, I should just go with a simple liberty count, even though there may be moves that tip the balance from that default resolution.

I’m still getting better at making problems. It’s mentally exhausting to solve too many whole board problems in a row. So I’m getting better at making problem statements to reduce this load.