Ricky Demer's analysis regarding the monkey jump's value

   

I find myself skeptical of the claimed values for the monkey jump being correct when

if white prevents it, a subsequent hane-connect by white would be gote

and

black barely has ko threats and black will not get any bigger ko threat

.


I added the circles on the left. The next two diagrams and notes are otherwise taken exactly from the miai values list page.

4.42 (4 5/12)

[Diagram]
Monkey Jump  

(Assumes infinite White territory to the right.)

[Diagram]
Prevent Monkey Jump  

(Assumes infinite White territory to the right.)


After white prevents the monkey jump, the remaining play is just the 1-point hane-connect. Thus, the resulting temperature is strictly less than 2, so the players are assumed equally likely to get it.

[Diagram]
After Preventing Monkey Jump  

In other words, preventing the monkey jump reaches this count in gote.

[Diagram]
Preventing Monkey Jump  

Putting black stones at the circle s here only stops their intersections from being black territory and gives black up to three extra removable ko threats. Aside from the obvious shift by 2, that does not affect the count.

[Diagram]
Monkey Jump  

Aside from the same shift by 2, making the circle s _here_ black can't make things any worse for black. Thus, making them black in both cases can only _increase_ the difference between this diagram and the previous diagram.

[Diagram]
Simplified Monkey Jump  

Accordingly, I simplify to this diagram.


[Diagram]
the response I'm looking at  

I am considering white responding to the monkey jump immediately with W2.


[Diagram]
if B3 is tenuki  

If B3 is tenuki, then I consider white immediately playing W4 and white protecting W4 by connecting at circle. I assume that despite protecting W4 that way, the area to the right of W2 will be solid white territory. Under that assumption, at the cost of one net move, white has done locally two points better than having prevented the monkey jump.

[Diagram]
if black does not block the descent  

If black continues locally anywhere other than W4, then I consider white responding immediately with W4. I assume that the area to the right of W4 is solid white territory (even if B3 is in there). Under that assumption, at the cost of _zero_ net moves, white has done locally at least one point better than having prevented the monkey jump.


[Diagram]
black blocks the descent  

The remaining case is black continuing with B3 blocking the descent. In that case, I consider white responding immediately with W4.


[Diagram]
if B5 is tenuki  

If B5 is tenuki then I consider white immediately playing W6 and white protecting W4 and W6 by connecting at B3. I assume that, despite protecting W4 and W6 that way, the area to the right of W2 and W4 is solid white territory. Under that assumption, at the cost of one net move, white has done locally two points better than having prevented the monkey jump.

[Diagram]
if black does not capture W4  

If black continues locally anywhere other than at W6, then I consider white immediately connecting with W6. I assume that the area to the right of W6 is solid white territory (even if B5 is in there). Under that assumption, at the cost of _zero_ net moves, white has done locally at most 1 point worse than having prevented the monkey jump. (Recall that the value of preventing the monkey jump was supposed to be over 4.)


[Diagram]
black captures W4  

The remaining case is black continuing by capturing W4. In that case, I consider white responding immediately with W6 and then white being willing to let the temperature drop. Note that 6-(supposed_value_of_preventing_monkey_jump) = 6-(4+(5/12)) = 1+(7/12) .


[Diagram]
if black's next local move is random and at high temperature  

Suppose that with temperature strictly above 1+(7/12), black plays the next local move, and that that move is somewhere _other than_ the five marked points. In that case, I consider white responding immediately at w. I assume that that will make the circle s and the area to the right of them solid white territory (even if B7 is in there). Under that supposition and assumption, a white play at b would get _at least_ 4 points more than the territory I just mentioned, so black can't do better than B7 as in the diagram and playing b at temperature strictly above 1+(7/12). That would leave white _locally_ at most 6 points worse than preventing the monkey jump. However, 6 points is strictly less than the temperature at which black got b plus the supposed value of preventing the monkey jump. (i.e., it would not be worth white having spent 4+(5/12) to prevent this.)

[Diagram]
if black's next local move is filling the ko at high temperature  

Suppose that with temperature strictly above 1+(7/12), black plays the next local move, and that that move is filling the ko. In that case, I consider white responding immediately with W8. I assume that that will make the circle s and the area to the right of them solid white territory. Under that supposition and assumption, b is _at least_ a 3-point play, so the "so black ... prevent this.)" from the previous diagram's analysis applies here too.


Note that 2*(2+(17/24)) = 1+4+(5/12) = 1+(supposed_value_of_preventing_monkey_jump) .

[Diagram]
if black waits for low temperature  

I assume that black is _not_ komonstrous. If black lets the temperature drop below 2+(17/24), then I consider white taking the ko until black ignores that or does something locally other than just retaking.

[Diagram]
if black ignores W1  

Suppose that with temperature strictly below 2+(17/24), black ignores W1. In that case, I consider white immediately playing W3 and white protecting W3 by connecting at black+circle. I assume that that will make the area to the right of W1 and white+square solid white territory. Under that supposition and assumption, at the cost of 2 net moves at temperature strictly less than 2+(17/24), white has done locally 1 point better than having prevented the monkey jump.

[Diagram]
if black responds to W1  

Suppose that with temperature strictly below 2+(17/24), black responds locally to W1. In that case, I consider white immediately filling at black+circle. I assume that that will make the area to the right of W1 and white+square solid white territory (even if B2 is in there). Under that supposition and assumption, at the cost of one net move at temperature less than 3, white has done locally at most 1 point worse than having prevented the monkey jump. (Recall that the value of preventing the monkey jump was supposed to be over 4.)


[Diagram]
black enlarges the ko  

The remaining case is black playing a circle while the temperature is at least 2+(17/24). (I am aware that black playing the lower-right circle would let white crush the ko. However, I would need this analysis for the other three circle s anyway.) In that case, I consider white responding immediately by taking ko with W2. For this case only, I assume black's external ko threats are all small.


(The following analyses apply regardless of which of the four circle s B1 is at. I simply picked the one that does not give white local threats.)

[Diagram]
if black backs down from the ko  

Suppose that black responds locally to W2. In that case, I consider white in turn responding immediately by filling the ko. I assume that the area to the right of W2 and white+square will then be solid white territory (even if B3 is in there). Under that supposition and assumption, at the cost of _zero_ net moves, white will have done locally at most 1 point worse than having prevented the monkey jump. (Recall that the value of preventing the monkey jump was supposed to be over 4.)

[Diagram]
black plays a ko threat  

Suppose that black instead plays a small external ko threat. In that case, I consider white ignoring it and white immediately playing W4 and white protecting W4 by connecting at black+square. I assume that the area to the right of W2 and white+square will be solid white territory. If black does not now carry out black's ko threat, then there was no point in black using the ko threat, so I further assume that black now does carry out the ko threat. Under that supposition and those assumptions, black spent one net move at temperature at least 2+(17/24) to carry out black's ko threat, in return for white being _locally_ 2 points better than having prevented the monkey jump.


Compared to having prevented the monkey jump, white's overall score from preventing the monkey jump at temperature 4+(5/12) is -(4+(5/12)). On the other hand, white's overall score from the above diagram will be T-S+2. Recall that if black lets the temperature drop below 2+(17/24), then white scores better than preventing the monkey jump by starting the ko _before_ black enlarges it. If T is at least 2+(17/24), then the only way one can have T-S+2 <= -(4+(5/12)) is if 9+(1/8) <= S .

Since 2*(4+(5/12)) < 9+(1/8) , it follows that black having extremely small secondary ko threats is not sufficient for it to be worth white preventing the monkey jump at temperature 4+(5/12), assuming black is not komonstrous.


I phrase that conclusion carefully, to avoid subtleties such as the following diagram, which is based on [ext] https://www.littlegolem.net/jsp/game/game.jsp?gid=1833985&nmove=113 .

[Diagram]
thermographically interesting situation  

What is the miai value of circle? (Note that the diagram is _not_ hyperactive.)


(In case any of this ends up being cited for whatever reason, I'm mentioning that my legal name is "Eric Demer".)


Ricky Demer's analysis regarding the monkey jump's value last edited by RickyDemer on July 21, 2018 - 17:18
RecentChanges · StartingPoints · About
Edit page ·Search · Related · Page info · Latest diff
[Welcome to Sensei's Library!]
RecentChanges
StartingPoints
About
RandomPage
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Goproblems.com
Login / Prefs
Tools
Sensei's Library