I agree, I was thinking it might be enough to just say “this is not joseki”, but still produce hints from it
I’m also going to implement a shortcut for the one color mode that claire_yang described here
1 Like
By the way, @okonomichiyaki, I sent a pull request earlier that enables the responsive fill mode by default on index. It also adds some metadata tags for auto one-boxing.
1 Like
I think that’s a great idea. Just something that reveals every purple hint automatically, and putting exclusion hints initially is not necessary since exclusion is implied, but the exclusion hints should still show up for subsequent guesses.
This way of playing should not be thought of as wrong or cheating, but rather just a way to play a variant of the game within the same framework.
2 Likes
Is it intentional right now that all the hints disappear when you start making moves in a new sequence? I’m not sure what the logic should be exactly, but at least the colors on empty, unused intersections should stay, I think.
2 Likes
Here’s what the auto one-box looks like
I had to explicitly add the index.html since Discourse seems to have cached the missing metadata of the link without it.
3 Likes
hmm not sure I understand, do they disappear for you?
this is what I see after doing one submission, then going back to the root (empty board)
oh nvm, I see it, if you branch off
2 Likes
Yeah, once you embark on a new sequence you lose the hints.
By the way, if you are open to going crazy with colors or symbols, some ideas of things to highlight:
- A known next move in the sequence (put a plus on it?)
- A known move played at the wrong time, so it will definitely become purple on submit (pink?)
- A known move no longer playable in the current sequence since you are past that move number (dark green?)
- A purple move played at an untested time, unknown if it will be purple or green (yellow?)
3 Likes
Highlight them when? As soon as they are placed on the board?
1 Like
some good ideas, personally I think it might get a little bit busy with colors.
but I like the idea of highlighting the next moves with known colors (just purple/green), since Besogo has letters already for next moves, just circling with the known color could be nice
3 Likes
As soon as they are placed, based on the information revealed earlier. Basically reduce the need to click around your earlier submits to remind yourself of what you have learned.
4 Likes
I played like this, to me it’s more fun to optimize the luck factor like that as pure luck.
Now after some practice we will go in some kind of standard opening with a limited but efficient choices of josekis. That still can be interesting
4 Likes
yes I’m very curious if there is some “optimal first choices”, maybe depends on the lengths? which means knowing the length ahead of time is interesting
on the other hand, just trying out many sequences to learn “oh is that a joseki?” is fun for me, as a less experienced player without a big library in my head
5 Likes
My guess is the modern 4-4, low approach, back off and attach, hane, double hane is a good guess for various 4-4 joseki, probably hit a good few of the 3-3 invasion points, as well as a few of 3-4 pieces.
That or maybe the small avalanche is quite compact.
2 Likes
Slightly OT but:
Imagine a conversation of “have you studied joseki” and your answer being yeah, but to get good at Josekle 
8 Likes
I’m sure we’ll get some interest from players on other go servers
5 Likes
Josekle #6
Too few moves
Too many moves
Splendid
I admit that my second attempt was not exactly joseki 
Are there plans to add feedback on the board instead of just in the tree? I saw some people discussing it earlier but I couldn’t quite figure out if it had been agreed that would make it too easy
3 Likes
do you mean show known next moves? like over the lettered spaces without stones that Feijoa was suggesting?
I personally think it’s a great idea but haven’t had a chance to look at the code for it yet.
2 Likes
If I’m understanding that earlier conversation correctly, I think it was about the variations tree rather than the board itself? 
I’m just gonna make some diagrams because I’m not sure I can figure out quite how to express it in words, or whether somebody has already said the same thing.
Here’s basically what I was thinking. Assuming a Josekle solution like this one (numbers shown for reference):
And a first attempt like this one (again, numbers shown for reference):
Which yields the following result:
I imagine it would be helpful if the hint shown at the beginning of the next attempt includes numbers for the green moves to show when they were correctly played:
And also if the hints persisted during the following attempts until they are covered up:
I’m not sure whether this would come off as visually irritating to some, or if it’s a huge programming challenge, but the visual aids this way would remind me a lot more of the persistent Wordle hints.
7 Likes
I think something along those lines are what @Feijoa was proposing as well, and your diagrams really help clarify the concept.
I think it’s a good idea to explore this concept and experiment with some variations upon it. Originally, I was envisioning the green hints aggregated in the root node having a number, but I didn’t get to programming that as a matter of convenience. In the rare case that a joseki has overlapping stones (due to captures and then later filling), we’d have to figure out how to display that. Perhaps, just something like putting all of the numbers separated by a slashes? Maybe all of this persistence (or forward propagation) of hints could be something that could be toggled on/off. I think there are some questions about graphical design, game design, and efficient presentation of information that would need to be considered.
The persistence of the “red X” hints could lead to a great deal of clutter. One possibility is to suppress those when exploring out a new variation, but only immediately show them (before submission) when a stone is placed on a spot previously excluded.
From a technical perspective, this is all definitely doable, but we’d have to slightly rework some of the underlying implementation. Currently, we’re using markup symbols (with reworked graphics) to present the hints, and the logic of deciding whether or not a position has already been submitted is based on just checking whether or not there are markup symbols present in the view. With automatically forward propagating some hints, we’d have to use another new attribute set on each node to determine whether it had already been submitted or not. Alternatively, I guess we could just allow nodes to be redundantly submitted.
On another technical note, another way of automatically forward propagating the markup would be to rework how those are stored within the game tree. Currently, markup is stored as an attribute specific to each node, and forward propagation would require some logic to mark new child nodes. However, we could rework that as something automatically inherited (much in the same way that the stones state is automatically inherited such that each node just stores the change). That way, if hints are marked on a node and on root, those same hints will automatically propagate over the entire tree, unless overwritten by later hints placed directly on that node, or by subsequent board play that could act to clear them.
5 Likes
Okay, it was about time for me to write some code. I’ve added buttons to zoom in and out, pull request is in.
If anybody is eager to try it out, visit my Josekle fork.
I’m not entirely settled on the zoom steps and the default zoom, feel free to improve those.
3 Likes