Note: these are of course subject to ongoing review, and input is welcome

One broad goal is for the OGS Joseki Explorer to contain relevant and useful joseki.

A fear I have is that this usefulness might be diluted by sequences that were "played once by a pro or AI", or "played by AI" but actually are situational, or are "interesting for some reason" ... rather than being truly established as "joseki" in the sense of "sequences of play that are credibly established to be among the best that you can do locally" (quote from the Charter).

The "Joseki Explorer" loses some value if it has too much in it!

It should be a filter through which only "the best" are seen, to save the user from having to do the filtering themselves.

So a guideline would be to err on the side of including positions and sequences only where this credibility is established.

However, we should not be too didactic about this. If a sequence or position is useful to document for analysis and awareness, then we should do that. In borderline or outlying cases, suitable descriptive text is important.

Our "tags" also help people find "the best", by marking these as "Joseki", allowing users to filter out other "intersting but not actually joseki" positions.

There are three kinds of moves/sequences that are candidates for inclusion:

• Joseki
• Interesting and valuable sequences to know (even if they are not joseki)
• Mistakes and refutations

"What is the definition of Joseki for OGS Joseki Explorer?" is the most common question in discussions about what to include.

Joseki: Sequences of play that are credibly established to be among the best that you can do locally, such that if you play moves from one of these sequences and your opponent does not, you can expect to emerge better off, all other things being equal.

"Credibly": A sequence or position is credible if we have a source for it who is or was a professional player. We also accept sequences that are "demonstrably common" in professional play even if there is not a professional source that specifically mentions it, but this is not well defined, so is less ideal.

Note that just having a professional describe a position does not make it Joseki. It also needs to be "established".

"Established": We don't have numbers for this yet, but it has to actually be, or have been, commonly used in professional play. Use in AI-only play alone is not enough - we want it to have passed through the filter of "humans have worked out how to use it".

Some sequences are not Joseki, for many reasons, but are arguably worth including - common "situational sequences" are an example.

These will typically be marked as "Good" rather than "Ideal", and not tagged as Joseki.

Question: do they still have to be credibly sourced? << Under consideration.

There are some moves that appear - through their attractive shape or logic of the response - to be "correct" but are in fact substantial mistakes.

This can be included and marked accordingly, but only sparingly.

Clearly not every wrong response to a Joseki will be listed - only those that have a compelling reason why it appears they would be played.

Mistakes and refutations do not require the same strict level of credible sourcing, because practically speaking this level of information is not readily available however they must be validated by AI.

Please follow the principle: all variations must go down and to the right

I can't tell you how many times I've read on Josekipedia "this position requires the ladder" and been stuck on trying to figure out "what ladder??".

In OGS Joseki Explorer we help the user with this by linking to the position that shows them the ladder in the description.

The plan is to pop up that position in a mini-board... when I work out how.

TL;DR:

• Enter the new sequence in Explore mode
• Click Save to enter Edit mode
• Enter the relevant details
  • Type in a description
  • Choose position type, tags and variation label
• Click Save to save the new sequence

The "Position Description" is rendered using "Markdown".

This allows you to do simple formatting such as headers, italics etc.

There are also some "special joseki" markups...

<X:K17>

(link text)[222]

At the moment this just renders a labelled link - the link text with a link to the position - but ultimately it should show a thumbnail of the referenced position.

The intention of the numeric "variation label" is as an initial guide about what's likely to be the best move.

All other things being equal, '1' would be most likely to be the best move.

This usually means it's the most commonly played at present.

Josekipedia's definition of position types has stood the test of time, and seems workable for us to use:

IDEAL: A correct move. This is a perfect move that cannot be shown as inferior to another choice. Of course, multiple Ideal choices often exist for each position. A series of Ideal moves forms a joseki.

Note: recent update on policy: https://forums.online-go.com/t/multiple-mistakes-in-ogs-joseki-position/30883/41?u=eugene

GOOD: A move that is almost perfect, but can be shown to be slightly inferior to the Ideal choice. This move may be played if outside circumstances (like a big wall) dictate an unconventional approach. Or possibly it's a way to avoid complications while only sacrificing a small advantage. A Good move cannot be part of a joseki.

TRICK: A trick play is a move that challenges an opponent to find a non-obvious refutation. Should they fail to do so, they will often incur a large disadvantage. With the correct response, the position typically ends slightly unfavorably for the side that attempted the trick, although this is not always the case.

QUESTION: This position is worthy of discussion, there are questions about it. It may be removed if the discussion concludes it does not fit into one of the other categories.

MISTAKE: A move that is certainly incorrect. Ranges from a little bad to completely horrible. Mistakes are only included when they are likely or common. Every possible mistake does not belong.

(tbd. Please put the "Joseki: Position Is Settled" tag on the final position of a joseki sequence)

All "Joseki: Done" positions should have a "Source", which indicates where we can confirm that this is a credible joseki.

The "Traditional" source is a "get out of jail free". It means "everyone knows that this is a credible joseki, it's demonstrably played all over the place". Positions marked in this way invite questions in comments about whether that's really true.

A "Source" is expected to be the high level "published location" of the information about the position. This can be a book, or an internet source of information.

It is not expected to be a detailed reference to the specific position, because otherwise we'd end up with one "Source" tag per position. It is supposed to be a way that people can filter to chose only positions from a certain "Source" that they like.

The reference to the specific position within the Source should be in the Description. So for example the Source for the play at Tengen is "Dwyrin", because he's the guy who said it's a valid opening, and the Description contains a pointer to where exactly he said that.

Exactly what constitutes a "Credible Source" is under discussion. It has been agreed that just being an amateur Dan does not qualify - indeed, concensus is that amateurs are not a Credible Source at all.

An unsourced sequence of moves may be included in the Joseki Dictionary if that sequence starts from a misplay that a human player is likely to make, contains only those moves necessary for punishing the misplay, and has been validated by AI.