Why are these moves joseki?

Two very simple joseki here:

Why is ts the triangle marked stone preferred over the square marked one?

And - if your opponent plays the square one: How do you profit out of it?

1 Like

I think simply because usually letting the other player approach on the third line (H3/R9) in sente or almost isn’t good.

1 Like

The high extension is certainly playable at our level, if we want influence.

1 Like

I would say it’s a matter of safety versus speed, as are many moves in a game of go.

In these particular examples, strong players usually opt for safety against an equal opponent.

1 Like

Typically we say that if you want to play the square move Q11, then you shouldn’t play the second-line keima S16.


Since AI I think the S16 keima is considered a little bit slow anyway. Attach at R16 is suggested instead, as it is more forcing.

1 Like

Third line finishes the position and doesn’t leave the standard weakness of a 4th line stone being approached with knight’s move on 3rd line which Sofia mentioned. As well as h3 it can sometimes be possible to cut at E3 directly.

Having said that, the knight’s move in lower left is a joseki too, just not so common as solid to 3rd line. It’s easier to play if white made hanging rather than solid connection though, because that gives black some forcing moves to help deal with the cut in future.

4th line in top right is not joseki with the slide, as the slide makes it heavier (can’t jump into 3-3 to trade if they exploit weakness of the 4th line) so standard thinking is to only play it directly without the slide. However, I recall Go Seigen not dismissing it, and yet again his prescience of AI is remarkable as they sometimes play it too.


This topic offers me an opportunity to ask some questions I have been thinking about for some time.

If we take Sensei Library’s description as a starter I see some confusing things.

Joseki is an English loanword from Japanese, usually referring to standard sequences of moves played out in a corner that result in a locally even exchange. The term may also be used for sequences that occur in other parts of the board (sides and center) and at various stages in the game (opening, middle and end), although such usage is usually made using explicit wording, such as endgame joseki. It is also possible, although uncommon, to use the term joseki to refer to sequences whose outcome is not equal.

The definition of joseki is ambiguous, to say the least.

Which sequences are considered playable joseki has changed throughout history, owing to the effects of rule changes most notably komi, continued research from professional players and, most recently, from artificial intelligence analysis. A joseki that has fallen out of fashion is normally labelled as ‘old’ or ‘bad’. If a sequence is not considered joseki, it usually means that either one side (of the exchange) is left at a clear disadvantage with little or no redeeming benefits, or that the position is too uncommon to have developed an agreed upon continuation. Proper evaluation of joseki considers the trade-offs between: sente and gote, territory and influence, speed and solidity, local and global balance, and aji.

What a joseki is has changed over time. What once was considered joseki doesn’t have to be joseki anymore nowadays.

I am curious about the process in which is ‘decided’ which (corner) sequence is joseki and which one is not.

  • How is a consensus reached?
  • How does a joseki debate evolve?

Usually it is agreed amongst top pros, would sometimes write books on the topic. These days strong AI also weigh in on adding new and refuting old joseki.

1 Like

Top pros play sequences in their games. Other pros watch those games and get inspiration. Study groups of pros review games and discuss sequences. Study groups that include insei / yungunseng also study sequences extensively. And now the AI helps study sequences, too.

In periods where joseki evolve slowly enough, pros can write joseki books. Famous books that have been translated in western languages include Ishida Yoshio’s “Dictionary of basic joseki” (1975) and Takao Shinji’s long-awaited “21st century dictionary of basic joseki” (first volume 2011, second volume 2012, and third volume cancelled because of AlphaGo, I think).

Those dictionaries list sequences in order, with a useful index to navigate them, and comments about the sequences. There are usually three kinds of sequences: the “balanced joseki”, the “this is locally better for one player, but might be playable if you really want this direction of play”, and the “this is bad for one player, don’t do this”.


It’s not just top pros who invent joseki. Other pros do too, though it can’t really be called an accepted joseki rather than some random pros innovation until the top pros approve of it and start using it too, and IIRC the avalanche joseki originally came from a Japanese amateur, the pros had dismissed it out of hand for being bad shape playing yourself into hane at head of two, but on further analysis concluded it was playable and from then it developed into one of the most complex and well studied josekis.

Also with the evolution of joseki, there is often a lag through to western amateurs. There are players here in 21st century Britain who learnt from Japanese books from the 1970s and are still playing joseki that have been relegated from pro play (except in very special circumstances) for decades. Teachers like Hwang Inseong are good for reducing that lag.


And KataGo :wink: