Monkey Jump
A monkey jump is a large-knight jump from the second line to the first line into the opponent's would-be territory, reducing it by a considerable amount. The stone on the diagram cannot be cut off. If White has a large territory to the right, the move can't be ignored and has to be replied to.
A monkey jump is proverbially worth 8 or 9 points in sente, although the exact amount depends on the position, and it may very well be gote.
In some cases a small jump to a is reasonable as well. This is sometimes known as a small monkey jump.
Sente or Gote?
In the diagram above, is usually not sente, simply because White does not need to answer at all. It also seems that White can answer it in sente:
But Black can play better than this, as shown in the next diagram:
Later, White will probably get this exchange in sente. But this is still a gain for Black. See Value of a monkey jump for a detailed discussion.
If the marked stone is one line higher, as in this diagram, Black can keep sente with this simpler (and better) line, as White needs to defend the cutting point at a.
Monkey Jump Failure
The stone on the second line has to be strong. e.g. monkey jump (Black a) doesn't work here because of the weakness at b.
White can play at . If Black responds at
, then the atari at
and the throw-in at
will capture all the black stones.
Thus in response to
, Black can only connect at
, then White plays at
, and you can see the difference yourself!
See also Endgame Tesuji 4 for a more subtle case where the monkey jump is not an outright failure, but clearly suboptimal.
Malweth I also saw a Monkey Jump failure in one of my current DGS games:
Here I cannot respond at 'a' because white can connect via 2. If I play Ba,,Bb white can connect out with two liberties. Black playing at c after b gives white three liberties to my two.
The Monkey Jump into a High Position
What's White's proper defense here?
is wrong, as Black can push further into White's territory at
. The
marked squares are miai for Black, and White cannot cut
off. Try it out!
Chris Hayashida: What about at
? Seems to me that Black can't live by taking the inside. Actually, I found my answer. If Black extends upward (down in the diagram) to
, and White extends to a, then the black hane is hard to deal with. It would work if the ceiling were one line lower, though.
It looks suprising, but works:
putting the two marked stones into atari is correct. If Black continues
with
,
cuts it off.
Therefore, Black should play as follows:
In an actual game Black should be satisfied by the exchange of the marked stones and keep the plays 1 through 10 in reserve. There is no need to play them out immediately.
WTD: I believe Black's play in 'Correct sequence' above, is not optimal. If is played at 6 (in that diagram), the result is 2 points worse for White.
The full sequence is shown in 'Better Black reply', and at 5 gives the same result. Black can still tenuki after
, but if Black plays next locally, it should be at
, rather than 7.
However, if White plays kosumi in response to the monkey jump as here, Black must play to avoid being cut off, then
is required. Apart from tenuki, black has three alternatives for
(at 6, 7, or 9), each giving the same result. This is equal in value to the 'Correct sequence' above, but without the drawback described in 'Better Black reply'.
Relevant sequences are shown for a narrower but very similar position, at goproblems.com in problems 3762 and
4105, the creation of which led to this analysis.
Alternative to the monkey jump
If White has far superior ko threats, then this sequence can be expected. So the monkey jump is better by 2 points. (The two points marked with a Square).
If Black can win the ko, White can reply like this, essentially reverting to the original monkey jump diagram. So the one-space jump is never better. (This may, of course, be different in other circumstances.) (White 11 connects at c.)
Harpreet: If I'm not mistaken, black would play like this if there are enough ko threats for him. 5 at 6 is not possible.
Which is the bigger move when ko threats are present?
WTD: --- lmj = large monkey jump ------ 1sj = one space jump
Black has 2 ko threats at 'u' and 'v' ('u' is the larger), whilst White has none. Once (1sj) is played, then
, &
follow. White must then decide between
at 'z' if she wants ko, or at 'y' if she doesn't.
White wants ko.
If White plays at 'z', Black must play
at 'y', and 5 - 7 then follow.
makes the larger ko threat.
White now has the option of sequences (a), and (b). The best one for White depends on the size of Black's threats, and the value of sente.
carries out the larger threat.
answers the larger threat.
at 'x' prevents Black filling the ko, and is effectively an internal threat.
makes the smaller threat.
at 'x' wins the ko.
carries out the smaller threat.
fills to the right of 'x'.
White doesn't want ko.
Relevant sequences are shown for a narrower, but very similar position, at goproblems.com in problems 3844,
4857 and
5781; the
feedback from which formed the basis of this discussion. The top left interface is also slightly different causing a small variation in threat size requirements, but the overall argument is the same.
What is sufficient ko threat advantage? White does not need threat advantage, as she takes first in the ko. It is Black's threat advantage that is determining. With only 1 Black ko threat, however large, 1sj is never better than lmj. This is because White can save the threat AND win the ko, by playing D19 ('x' in 'Sequence 'b' ...' above) after Black retakes.
The minimum requirement for 1sj to be better than lmj is 2 threats. When sente is worth nothing, the smallest threat sizes that work are 15 and 11. The maximum gain Black can get by choosing 1sj is 2 points.
In summary: With this formation in a real situation, it will be very unlikely for 1sj to be superior to lmj. The maximum advantage Black can get is only 2 points, and to realise this he must have 2 more large threats than White, and risk fighting a difficult ko. Also, the best defence against lmj into this formation is very poorly known, and Black is likely to get 2 more points than he should do, which would neutralise any advantage he might get from 1sj.
How to stop the Monkey Jump
The correct White response to depends on whether there is a white stone at
a or not. If not, c (or sometimes d) is usually the correct
response, and if there is, b is the first move you should think of.
If there is no white stone at a, the monkey jump is about 10
points in sente, while if there is a stone at a, it is the regular 8 points
in sente.[1]
WTD: In 'Monkey Jump Workshop' by Richard Hunter, c is generally dismissed as a reply to the jump, because it loses points unnecessarily. I don't think it should be considered unless special circumstances apply, such as the invader exploiting a gap in the defenders formation, or forming a separate live group.
You should try to read out the answer precisely, as you should neither give up two points, nor lose the game by producing one of the catastrophic failures we have seen above.
Try out your reading in the Monkey Jump Problem Series.
Examples illustrating the rule of thumb above follow below:
WTD: If is played at e, in '10 point monkey jump' (above), the result is 2 points better for White. This can also be achieved if White replies to the monkey jump at
here (d in 'Defense summary').
It comes down to whether Black can capture or not. With the squared white stone,
is safe.
[1] dnerra: These values assume that White's hane, after preventing the monkey jump by a block on the second line, would be sente.
The term Monkeyjump can refer more generally to the large knight's move (ogeima), a loose but fast extension, often used in endgame on the first line (as illustrated above), but also used elsewhere.
Bill: Not so. It is specific to the play from the second line to the first, according to the Go Players Almanac.
Hu: I agree with Bill, but I think the writer might be referring to how these moves can be used to undercut formations near to the edge but not necessarily right at the edge. Such usage would be less general than ogeima, but a slight generalization of the Monkey jump as it is usually thought of.
Bill: Such a move is simply called a suberi, or slide. The monkey jump (saru-suberi) is a particular kind of slide.
Contributors: fhayashi, AlexanderYoshi, Jonathan Cano, unkx80, ArnoHollosi, IaGo
WikiMasterEdit by dnerra, 2003-12-01.