Tsumego Cluster Map

Tsumegos can often be classified with respect to tesujis, and be coupled together with small-but-crucial differences, like a liberty or the possibility of a ko. In this sense, tsumegos remind me of (data science) cluster maps, where each cluster would have a tesuji or any other answer theme. I believe this would be an amazing way of visualizing relationships between shapes while better organizing and memorizing tsumegos, not to mention better understanding the whole.

I don’t know of any initiative in this direction by any other website — if you do, please share. And I do believe this would be an amazing addition to OGS, all the more because there could be tight and interesting integrations with the already functional “Puzzles” section of the platform — ideally, I believe, the user would examine the variations while still on the map page.

What do you think about this idea?

1 Like

You can find this kind of thematic organization in books or in some puzzles already and It’s made by their authors under their control.

To generalize the idea to a set of publication is an interesting idea, although i don’t know how practical it is (pattern recognition well working? Producing interesting categorization? Cross-referencing? ) if you do it in a automated way.
I didn’t meet any experimentation yet on that.

Cho Chikun encyclopedia of life and death is organised like this for example. You often get a puzzle and the next handful are very similar

1 Like

Classification could be done by users using tags, but in some cases this could spoil the solution (ko, seki, liberties shortage and so on).

If I understand what the OP is talking about :innocent:


I know there are books which follow this same thematic presentation of tsumegos. However, I do not believe they are enough to rule out the map I mentioned as redundant. From my point of view, this interactive map would be the dream of those who were truly trying to write those books. And having this map be interactive and online would make exploration of the content far quicker than dealing with physical material.

The map could be done Wikipedia style, as @Lys mentions, but I would rather have very experienced users curating it. And, when it comes to spoiling solutions, perhaps we could have the option of only showing tsumegos the user has already solved, or any other unlocking dynamic.

These go pro writers?
I have a little story on this. In a go camp during the dinner i was sitting near a 9p Sensei Saijo and a 5p Sensei Catalin. So they were eating too but pros first characteristic is too be more interested in go as anything else, not like us. They are still playing when you don’t. So at same time as we were all eating they had a goban nearby and some training between the 9p and his “student” 5p.
They were playing with josekis, modifying a move for another, getting influence or territory, getting in some hot fights, all this at a hallucinating velocity through dozens of sequences.

Let me doubt on your opinion.

You first have to make the shape recognition efficient. You can have different shapes in the same position so you will maybe need to prioritize. Maybe some will be less meaningful as another one in our brain? As i answered before, it’s an interesting idea, i just not sure it will bring more in the side of practice for progress.

A problem collection has more concern as simply being exhaustive. The author has something to transmit, his own selection of tesuji/shapes, a way to make you walk through them, an intensity by a well weighted repetition…

I guess there are many YouTube videos examplifying what I said, but just so we have a concrete example, here is Lukas Podpera 7d EGF on Practical Tsumego:

Sensei’s Library has a large section on this topic, the L-shape.

Can I ask a pretty silly question, what is a cluster map?