Kikashi: taking advantage of the opponent’s plans

When and wherever explaining kikashi, the following joseki always seems to be given as a prime example:

The sequence from 1 to 17 in Dia 1 is a fairly often seen 5-4 point joseki. White takes corner territory in sente, black in exchange building outside influence with his stones. The exchange of white 1 to black 2 in Dia 2 is assumed to be part of the joseki, and after black 2, play usually moves on to another part of the board. Most readers can probably classify white 1 as a kikashi, or as a forcing move, but what exactly happened here?

When asked for justification for white 1 in Dia 2, an oft-heard reply is that “white 1 is sente, because black has to respond with 2, and then ▲ is forming a bad empty triangle“. This is correct, although not the whole truth. White 1 can also be said to be a “peep”, or nozoki in Japanese.

What if black resisted, like with black 1 in Dia 3? In that case, white could cut with 2, and if black played a counter atari with 3, white is able to make the larger cut of 4. After 6, the result would be a disaster for black in most games.

Instead of black 3, black could play at 6. Then white would have gotten some extra territorial profit in sente, also being able to continue at 3 later.

What if black played the solid connection of black 1 in Dia 4, preventing the white peep? Later in the game, white would be able to capture a black stone with white and 4, leaving a cut in black’s shape at A. If black used his turn to prevent the cut with black B, white would again have profited in sente. Looking back at Dia 1, the black hanging connection of 17 completely prevents the cut of A in Dia 4. For this reason, black 17 in Dia 1 is the joseki move.

Looking back at Dia 2, white 1 is indeed sente, and a peep, but what makes it a kikashi?

If “kikashi” meant the exact same as “forcing move”, the term kikashi wouldn’t have to exist at all in western go culture. The terms aren’t exactly synonymous, however. The Japanese word kikashi, 利かし, uses the character of 利, which stands for “benefit” or “profit”. In general Japanese language, kikashi more often means “to season” or “to enliven”, but it also has the meaning “to bring out a thing’s effect”, which hits the jackpot for us. The verbs 利く (“kiku”) and 利かない (“kikanai”) also come up in Japanese go language, apparently meaning “forcing” and “not forcing”. They share a connection, but do not hold all of “kikashi”‘s meaning.

I will have to attack the current Sensei’s library definition of kikashi a little bit:

Kikashi, a Japanese go term adopted into English, is a sente move or sequence that produces a certain additional effect. Because it has done its work, it can normally be freely abandoned unless it is part of a much larger chain or group. It is usually translated as forcing move or sequence.

While not being wrong, I would imagine I am not the only one who finds this definition too ambiguous; what is exactly the “certain additional effect”?  What about the second sentence in this definition — it can be abandoned, unless it can not?

Looking back at Dia 2, can the reader figure this out somewhat? White 1 is sente, and a peep, forcing black 2. While black got a bad empty triangle, it is not simply for the empty triangle that white 1 is kikashi. White 1 is kikashi because it can be used, more so than black 2: white 1 can be an effective ladder-breaker. This would be the so-called “certain additional effect” of our sample kikashi, here.

Let us have a look at another example. The sequence from black 1 to white 12 in Dia 5 is a well-known star point joseki. I’m sure many readers have seen black immediately push with black 1 in Dia 6, forcing white 2. I saw this kikashi example in the Japanese go theory book 依田ノート (“Yoda nooto”, Yoda notebook), where Yoda Norimoto 9 dan discusses the subject. He quotes an amateur player who justified Dia 6 like this: “It feels good to get a one-point profit in sente, forcing the opponent to respond”.

If black didn’t play black 1 of Dia 6, black would have more options for the future. For instance, if black later on got a stone at 1 in Dia 7, and white didn’t respond, black would have the tesuji of 3, which would reduce white’s corner territory in sente — white would need to create two eyes with white 8, unless he were able to escape to the centre and live there.

Even if black wasn’t able to get black 1 and 3 of Dia 7 in, he would still be better off to leave the kikashi of Dia 6 for a ko threat.

Still, black 1 in Dia 6 is a sente move that reaps a “benefit”, reducing white’s territory by one point. You could argue it to be a kikashi, albeit a bad one.

To further explain kikashi in words of one syllable, here’s a third joseki example. The moves from 1 to 25 in Dia 8 are an old 3-4 point joseki. After black 25, white most often plays elsewhere.

Later in the game, black has the options of A and B as in Dia 9. Both of them are kikashi. Black A prevents white’s advance to the centre in sente, forcing white to remove the three dead black stones off the board. Black B prevents white’s advance to the upper side in sente, forcing white to remove the three dead black stones off the board. Black cannot get both A and B in sente: it’s either-or.

It would be an amateurish mistake for black to simply play either A or B at the first opportunity. The correct way of play would be to first figure out whether the upper side or the centre is more important for black, and then pick the better kikashi. If neither kikashi was good for the time being, black should leave both of them unplayed until one became clearly superior to the other.

Indeed, kikashi are not simple forcing moves that are played because they can be played in sente, but we have to be mindful about what the kikashi’s actual benefit is. For argument’s sake, I wouldn’t classify black 1 of Dia 6 as a kikashi: it, in fact, loses black much more than it gains, and to me, a “bad benefit” is an oxymoron.

A kikashi is a sente move that benefits the one who played it.

The sequence from white 1 to black 16, here, is another joseki. After black 16, white usually plays a move like B to counterattack black; even when white doesn’t defend his corner group, black cannot kill him without a ko fight. Black 16 is increasing the black chain’s liberties while threatening the ko (which starts with black 2-2 in the corner). If white answers with A, he will have responded to the threat black 16 made, and black 16 becomes a kikashi; black got a benefit, namely additional liberties for his group, in sente. On the other hand, if white answers with B or the like, he will not have responded to the threat black 16 made, and black 16 would not have been a kikashi.

A move is not kikashi unless the opponent responds to it in the predicted way.

Earlier I argued that if an attempt at a kikashi is not beneficial, it’s not a kikashi. This diagram here features white 6, which at the time could be called kikashi, being sente, forcing black to respond in a predicted way, and reaping a benefit (strengthening the white 2 stone). However, after the sequence has been played up to black 11, the kikashi of white 6 turns bad: the exchange of white 6 to black 7 is beneficial for black. Black’s way of play here is effective, having made white waste his kikashi. Similar techniques can be found in many a professional game.

There is a pseudo-proverb of “play kikashi before living”. What is exactly meant with it? By now, the reader will already know that a kikashi is not a kikashi unless it reaps a benefit.

Let us look at the sequence from white 1 to white 17 in the diagram above. White 1 goes to live in what was originally sort-of black’s corner, and black cuts white’s stones starting with black 2. White’s solid connection of 5 is a possible way of living in the corner: this way, white gets the kikashi of white 7. This is an example of “play kikashi before living”. Black went out to cut white’s stones, and white makes use of understanding black’s intentions by playing 7, which simultaneously strengthens the white stone to the left and creates the possible white connection at 8. When black cuts at 8, white then moves on to live with 9 etc. White 15 is yet another kikashi, and when black has completed his dumpling shape, white finally lives in the corner with white 17. The white stones towards the left will not get in any trouble even though black was able to cut white.

Instead of the black cut of 8 in the earlier diagram, black should in this case play as shown here. Now, it is actually the ▲ stones that became kikashi, reducing white’s territory!

Here’s an artificial position where black has an opportunity to effectively play kikashi before living. Can you find it?

Before creating two eyes, black should first peep with 1 as shown here. If white responds with 2, black then creates two eyes, and he may get use of the cut of A later on, if the white right-side stones become attackable. If white responds with A to black 1, black would already be alive and he could play elsewhere.

A kikashi forces the opponent to commit more to his game plan, and so benefits at the opponent’s expense.

This here is the game against the Japanese 8 dan which I already showed some weeks ago.

[sgfPrepared id=”0″]

As the reader may guess, white 50 is meant as a kikashi. Black stated with 49 that he won’t let white live easily, while also surrounding territorial profit in the upper right corner: white 50, then, is asking black to commit some more in his corner territory. Black has to be careful about his response.

If black simply defended with 1 as here, white 2 becomes a difficult move to handle. If black uses the common technique of the hane of black 3, up to black 7, white’s result becomes excellent. White has A in reserve so that he can make two eyes for his group anytime he wants, but since the group already appears to be safe, he might as well play the sequence of white B, black C and white D next, going on to live on the right side. This way, white 50 will have become an excellent kikashi.

If black doesn’t play the common technique shown before but something else, white will play as shown here. After black 11, white will have made his upper-side group strong, and while white 50 now isn’t really a kikashi, it doesn’t lose white anything either, as black also has the weird-looking stone of 1. Most important of all, this way the white upper-side group  got strong in sente: this result is still a major white success.

For these reasons, in the real game I didn’t simply defend against the white kikashi, but fought back with black 51 and so on. White would have been able to live in the corner, but in that case the upper-side white group would have gotten in serious trouble.

To start reaching a conclusion for this essay, I would like to wrap up the definition of kikashi as “A sente move or sequence which reaps a benefit while forcing the opponent to commit more to his plans.” I may come back to this text to make minor structural changes to this definition, but anyone who has read this far should be able to see my intention behind these words.

More skilled readers may be interested in knowing that professional go is all about kikashi, and if a professional game’s result isn’t caused by a simpler mistake, the reason behind the result will more often lie in one or more successful kikashi.

Not too seldom, somebody finding a fine kikashi will transform entire joseki. Remember Dia 8 in the beginning of this essay?

In a fateful game between Go Seigen and Takagawa Kaku in 1957, Go Seigen diverged from the then usual large avalanche joseki (Dia 8) by playing black 13 in the diagram on the right of Dia 8. If white simply responded with 14 and proceeded to play the classic large avalanche joseki as normal, up to black 27, the exchange of ▲▵ nets black an extra two points. A two-point hard gain is huge in terms of professional go, and for this reason, the whole large avalanche joseki got transformed, more or less due to one simple kikashi.

33 thoughts on “Kikashi: taking advantage of the opponent’s plans”

  1. Fascinating stuff… It bears some completely new light to playing go for me, although I suspect I will need a lot of time, study and games to fully grasp it. As of late I am learning a lot of new stuff that was just there in front of the unsuspecting and innocent me… ^^

    Thanks for this gift! You may not know it, but in Spain instead of Santa Claus we celebrate the Three Magi day, which is this Sat, so this post arrived in time as a wonderful present… Have a happy new year and lots of success!

  2. Kikashi is one of those SL pages that I’ve given up all hope of ever becoming a stable page with an accepted explanation, so this essay found an avid reader in me.

    As often, the examples are clearer than the wrapped up definition. An important thing I’ve understood here is that kikashi is an after term. The intention may be there, but the exchange will decide whether a move actually was kikashi. The profit aimed at is also tangible in all the examples.

    The difficulty is now to differentiate it from sente. Sente too depends on the attitude of both players: if I do not answer, your move will not have been sente. The suggested differentiator is “the profit for who played it”, but that looks self fulfilling: why play a sente move if it does not benefit you, i.e. in good play. We’re trying to define a term that is part of good play but not synonymous with it. Okay, it’s not gote. But it is more than sente too.

    What I’m missing, and what has been present in some of the SL definitions over the years, is the idea of being disposable. I see from quite a few examples of yours that a kikashi need not be a disposable stone. I accept that and will not challenge your definition of kikashi, but to me it becomes meaningless as I cannot differentiate it from (good) sente anymore. That is, I would never think of dia.6 and “kikashi or not” as kikashi but as sente. On top of that, dia. 6 seems to be a bad sente and it still is kikashi?

    Another differentiator may be that the opponent is offered a choice and whatever his choice, you will adapt in a beneficial way. But that to me is a probe.
    So, is a kikashi a probe which has become sente?

    For now, I’ll keep thinking of kikashi as a disposable sente move, because that’s the only meaningful concept I see.

    Thanks for this marvelous essay.

    1. Thanks for your thoughts!

      You are right in that we are only able to tell a kikashi after we’ve seen the response to it, as we can only tell a sente after we’ve seen the response to it. When analysing a game, it is often possible to find suprising moves which turned out to be kikashi, and directly contributed to the final game result. However, the term “sente” has absolutely nothing to do with profiting anything. A move or sequence is sente if its player retains the initiative. A move or sequence is kikashi if in addition to being sente, its player has profited without loss. You ask why play a sente move if it does not benefit one; to me, this line of thought is straying out from the meaning of sente. Sente is only about the initiative. Even professionals do play bad sente moves.

      The SL-added characteristic of “kikashi being disposable” is something that I cannot find in Japanese go literature. The 用語小事典 (small terminology encyclopedia) by Nihon Ki-in, 1997, says on page 61: “利かし: 現在先手で打てる手で、しかもなんらかのプラスにこそなれ、損のない手を、現に打っておくこと。” My amateurish attempt at translation: “Kikashi: playing in advance a move, which can be done in sente at the present time, causing its player no harm, and which (later) has an additional gain.” My apologies for the structural change in the translation; the Japanese text wasn’t translatable in the same content order. At least in this definition by the Nihon Ki-in (indeed, the book doesn’t even show the name of its author in its cover!), nothing is mentioned about the kikashi being disposable. Sure enough, a kikashi stone can often be disposable, but that’s not a defining characteristic of kikashi.

      As I also wrote in the essay a bit later: “For argument’s sake, I wouldn’t classify black 1 of Dia 6 as a kikashi: it, in fact, loses black much more than it gains, and to me, a ‘bad benefit’ is an oxymoron.” The Japanese definition for kikashi has this too: “…causes its player no harm…” (損のない手). Dia 6 in actuality doesn’t gain black anything, as black has the one-point sente move anytime he wants, but it loses black a lot of future potential. Dia 6 is an example of wrong usage of the term “kikashi”.

      I don’t see what probes have to do with the essay — I’m certain I didn’t mention them at all, and their meaning is safely different from that of kikashi. A kikashi is a sente move which gets a benefit while causing no loss, and is played after noticing the opponent’s intentions, making him pay more for whatever he wants. A probe is a move which makes the opponent choose what he wants. Sometimes a probe can turn into a kikashi later on, but that’s a different matter. The definitions are completely separate.

      Again, I would like to stress that a kikashi makes use of the opponent’s plans. “A sente move” means a move, after which its player retains the initiative, but this wording doesn’t imply anything more about what was actually played, and if what was played is beneficial or not. “Disposable sente move”, including the premise that whatever the move is, it is correct and good, will map a part of the moves which are called kikashi, but not all of them. For now, I’m confident that my definition of “A sente move or sequence which reaps a benefit while forcing the opponent to commit more to his plans” is a more accurate mapping, and while longer, it doesn’t require the premise of “good play”.

      1. the ki-in definition actually gave an interesting aspect, sort of like “a move that does nothing now but something good later on”. Does this have something to do with aji?

        1. I don’t think there’s a very direct connection there. Look back to Dia 2, for instance: white 1 is definitely a kikashi, but I cannot imagine (anyone) saying that the white stone of 1 is (or has) aji. However, it will still be a ladder breaker if a suitable ladder emerges to the lower left, meaning that it could do something good later on.

      2. It is true that disposability is not a defining characteristic of “kikashi”, but kikashi stones are often disposable, and that bears mentioning. As for “probe”, the point of a probe is to get the opponent to make a commitment, something that you believe is a characteristic of kikashi.

        1. About disposability: Volume 5 of “Sakata no Go”, which I think may be still in print, is about how to sacrifice stones. Chapter 1 of that volume is about how to play kikashi. There is a definite connection between kikashi and sacrifice. :)

  3. nitpicking more activated; parsing…

    syntax error on line 3, missing ‘”‘: add ‘”‘ after ‘forming a bad empty triangle’

    is becomes excellent: remove ‘is’

    Also, “A kikashi is a sente move that benefits the one who played it”… doesn’t every move give benefit to the one who plays it? That could be made more specific. Do you (sort of) refer to ?

    Good stuff, I’m sure forcing moves are a whole field of game theory, and each new angle to them (like go kikashi) gives new insight. Also, this one is literally a sente-essay (;

    1. Thanks, syntax fixed!

      The sentence “A kikashi is a sente move that benefits the one who played it” is still only getting to the definition I laid out, and not the whole definition itself. The fact that there is a tangible benefit is the one point with that sentence. I guess I’m on a tangent relating to the Sente Gains Nothing page, but that wasn’t my intention. I would prefer to use the final bold text as the definition I’d found for kikashi.

  4. One question, two comments:

    In diagram 6, did Yoda call B1 kikashi? I would find it surprising if he did.

    As for the ambiguity in the SL definition, that is on purpose. As Sakata said, pros do not always agree whether a play is kikashi or not. That means that you cannot give a precise definition. :)

    “Play kikashi before living,” is a real proverb. The point is to play moves that are sente before you live but would be gote afterwards. Not to play them before living loses points. It uses ”kikashi” in a different sense than its principal usage.

    1. It seems Yoda is careful to only talk about B1 as a forcing move. Since this Dia 6 is getting this much attention now, I believe I may have to change my text from “kikashi example” to “forcing move example”, too.

      I do believe we may be onto something here — it is certainly a feature of kikashi that they are sente if played right away, but often gote if played later.

  5. Nice article Antti. Have you read the book “Beyond Forcing Moves” by Takagi Shoichi? If you haven’t I expect you’d really like it. It’s one of my favourites, but unfortunately out-of-print now. I wrote a bit about it over on L19: Don’t worry if that example I used there seems rather easy for you, it’s got plenty of gems from great masters like Shuei and Dosaku that will make even a 6d swoon with admiration 😉

    1. That’s mean, now I’m interested in the book, but if it’s out of print I won’t be able to read it! Maybe I can borrow it from somebody…

  6. I would just like to add that to me it’s always been a very important property of kikashi that they are moves which are only sente at the moment they are played (i.e. they will be most likely become gote as the game proceeds), that’s why you make the forcing move instead of preserving the forcing move as a ko threat.

    1. You make an excellent point there, and one I completely omitted in the essay. What you describe is certainly an important characteristic of kikashi.

  7. One more comment, I am afraid. :)

    I do not understand why you use the avalanche joseki sequence as an example of kikashi. I just checked a book of Go Seigen’s game (“Gendai no Meikyoku, vol. 5), where he refers to that play. He simply says that W14 got the order of play wrong. White should play the cut at 16 first.

    1. That’s purely a historical reference I made. The cut of 16 before W14 is indeed the correct order of moves, but that is a joseki which has been developed as of fairly late. Go Seigen finding the two-point kikashi is what caused the development of the variation that continues with the cut of 16.

  8. “[…] can be done in sente at the present time[…] “. I would like to put also some attention to that “at present time” – part since the same move played later may not be as effective anymore (for example something else is more urgent). Terveisiä Japaniin Saksasta

    1. Kiitos!

      Excellent point you make there, and one I will either have to include in the current essay, or maybe (more preferably) in a sequel. I still have a fair amount of kikashi diagrams I want to present, as well as high-level techniques related to the term.

  9. Hey Antti :) I think that that Japanese definition you quoted above adds something essential that you left out:

    “Playing IN ADVANCE a move, which can be done in sente at the present time, causing its player no harm, and which (LATER) has an additional gain.”

    The key points seem to be “in advance” and “later”. The element of foresight is what makes it a kikashi. Is this reasonable?

    I’m also completely confused by you saying that white 1 in diagram 2 does not have aji. Maybe you’ll have to write an essay on aji as well because this seems to be completely against the meaning of aji in English. You say right away that it can be useful as a ladder breaker and “do something good later on” – what else do you call that, if not aji?

    1. That’s reasonable indeed, and as I omitted that from my essay completely, I guess I’ll have to write a sequel or something. :)

      Aji as “possibilities left in a position” is certainly a very broad concept. I can see it referring to white 1 in Dia 2, but something puts me off about saying that that white has stone has aji in relation to, say, the lower left corner — even though that can still be argued to be the case. I don’t want to challenge the aji term any more here, as that’s not our subject (for now).

      1. I am beginner at writing in English, sorry :-).

        Black 1 in Dia 6 gets a benefit and is forcing, so one is tempted to call it “kikashi”, but at the same time one knows that it is not a good move, and a candidate to kikashi is always a good one. (I said “candidate” because it will not be so if not answered in certain way, the way in which the opponent will be kikashed-the verb exists in Japanese.). I think the confusion comes from not distinguishing “benefit” and “net benefit”. Black 1 gets a benefit, but does not get a net benefit, i.e it loses more than it gains.
        Each theoretical concept of go is related with each other, but I think “kikashi” and “aji keshi” are special case. Every move is in some sense aji keshi, since it reduces the possibilities for future play, in the same way as every choice in any field implies lose of future choices.
        Black 1 is more aji keshi than kikashi, so it does not deserves the name “kikashi”, it is aji keshi.

  10. An interesting and enlightening explanation. I believe you have also nailed something important about pro games here. Indeed, many moves in pro games that seem strange can be explained as kikashi per this definition.

    I do not think my moves as kikashi or not but forcing the opponent to commit to his bad plans is always very entertaining. Getting proper compensation for his better plans is another aspect. Kikashi serves both goals.

  11. please include in your sequel that a nozoki to an ikkentobi is often bad and unkikashish! (unless you have a different opinion)

  12. Moving away from “disposable sente stone”, I’ve tried to come up with another translation in English. “Forcing move” doesn’t sufficiently catch the idea that the subjected party pays an extra price “for something he has chosen to acquire already”.

    How about this: “surcharge”.

    The opponent is “surcharged”, forced to pay extra for what he has bought, unless of course he doesn’t buy it.

    1. Hmm, surcharge the opponent before living, get surcharged… It would take a bit of time to sink in, but I could imagine using that instead of “kikashi”. It’s actually a pretty descriptive translation, too!

  13. I’m sorry, but I think I don’t get at all your definition of kikashi. In some examples that were given, I fail to see the part “sente now that can be useful later” of the Nihon Ki-in definition. Turning inward to gain 2 points is not useful later, rather right now, so I’d just call it a (good) sente move. Using your definition, would you say that killing an enemy group in sente is a kikashi move?

    1. Well, if we’re going into semantics, you won’t do anything with the two points right now, only later in the endgame. It’s not very evident in the Nihon Ki-in definition, but that two-point turn in the large avalanche is indeed “kikasareta”, “got kikashi’d”, in Japanese go language.

      As for you example case, that (surprise surprise) depends a bit. It’d be kikashi if by playing the sente killing move, you’re forcing your opponent to invest more in his game plan (for example defending a huge moyo from getting penetrated). The idea of making your opponent pay more for realizing his plans, and taking advantage of that, is key to kikashi. A regular three-point sente yose on the first line for example is different, because usually we’ll presume that the player to whom it is sente for gets it anyway.

  14. I’m not sure about the origins of the SL definition. But in Strategic Concepts of Go it says (and I quote): “A kikashi is a forcing move played to produce an effect. That is, a kikashi is a play which must be answered, usually in just one way; the exchange of the kikashi and the answer being useful in some way to the player of the kikashi. The terms kikashi and sente may seem to have the same meaning, but kikashi is applied to moves which are more or less incidental to the main flow of play. Once played, kikashi stones can typically be abandoned without any great loss.
    Timing is important in playing kikashi. Usually there is only one correct time to play kikashi for maximum effect, and ifthis chance is missed it may be lost forever. As the paradigm suggests, “Strike while the iron is hot!””

    The book goes on to give examples emphasizing timing. And then noting: “There is a very thin line between kikashi and aji keshi” and “…there often occur moves which seem to be kikashi but are really not. However, if one answers such a move as if it were kikashi then, in fact, that move becomes kikashi.”

    1. That description does in fact seem accurate and precise. If I come around to writing a second kikashi essay, I might just take this wording and explain it by examples!

Leave a Reply

Your email address will not be published. Required fields are marked *

Please confirm you are a human by solving this: