Gérard TAILLE wrote:
Seeing your post it appears that the move here after is part of top moves for black
There is very little difference in the evaluation of the top four or so moves. Approaching the corner in the top left has better score and percentage evaluation than playing in the ko position in the lower right. The difference is only about 2 / 5 of a point.
I'll come back to why I think the obvious big approach moves and the ko situation have similar evaluaion.
Gérard TAILLE wrote:
We previously concluded that it was not a good idea for black to play first in the bottom left corner. Does that mean that something was wrong in our analysis? I believed that the value of this black move was far smaller than 15 but it must be wrong. How do you calculate such value (without trying to be very accurate OC)?
Our analysis was probably quite good, I think our conclusion was that it isn't necessary to start the ko immediately and you can see in the screenshots, I posted previously, how this is confirmed by KataGo. I might have analyzed the approach ko incorrectly, maybe I didn't count the moves right, I'm really not sure. I was mainly trying to give reasons why playing the ko wasn't necessary (that means right away, it will have to be resolved somehow eventually). We can't expect to get exact results by only counting points in the opening and therefore shouldn't be very concerned if that part of the analysis has shortcomings.
How to calculate this ko? Well, it really is complicated by there being many different possible ko. I think some feeling for which ko is the right one is necessary. Personally, I think I might have a better chance of picking the right ko based on my mood than on a detailed analysis. Here the principle is that the larger ko is more worthwhile for white, if it can be fought but white has local threats so this proofs to be possible, and that white has the privilege to start the ko, which means that the conditions for playing first in the ko are materialized earlier for white than black. When it comes to evaluating the ko then it is most important if it is heavy or not, this ko isn't heavy, black and white have pretty much the same size of territory at stake. The score has already shifted to match the fact that there is this ko in the corner; it is a sunk cost so to speak for black who might have expected to own the whole corner. That was a guide of some sort, at least it's the insights that I think are important to evaluate the ko.
When I try to evaluate the ko more formally I find that I have difficulty evaluating the step ko that white actually shouldn't play and I have difficulty keeping track of the structure of this very complex position. These two things mean that a more formal evaluation isn't really possible for me, unless I take the structure of the position as something given (i.e. which ko to play, eliminate alternatives and decide what will be sente ahead of the analysis) which then isn't much of an analysis after all.
One thing that I found when I did try, and this could be due to my mistakes, was that the ko position is a "number". That is, it is a position which both players can avoid playing in. I say this could be due to my mistake but it also makes sense that the position would at least be equitable, or close to so, if the conclusion was that the players could (and should) avoid playing in it, but this could of course be my bad arithmetic. For one thing it does fail one sanity check, which is that the other ko would seem to no give a number and white appeared to prefer it. However, that could be due to me simply mixing up all the temperatures.
Gérard TAILLE wrote:
After having analysed the position in the botoom right corner, assuming neither player has a ko threat available, I concluded that a black move at "a" as well as a white move at "b" is valued 15 points. At least it is consistant with top moves according to katago. In addition I think the count of this position is B+30.
In pratice I expect one of the player will very soon play in this corner.
I think after white plays "b" the value per move is about 15 points, I got 15 7/9 for what that is worth. On the other hand it appears to me that black playing "a" would be a move in a game that is a number
OK, we take he ko first if white has to find a ko threat and at some point this ko will dominate the game but I think it is better to not play in the ko until we have to. "a" in this sense has a negative value, like any move in a number like for example in a seki. The reason it is a number is that black is better off if he waits for white to start the ko.
BTW I played from this position, using moves as indicated by KataGo for 77 moves before playing in the ko was the only move
At a few points I had to let KaTrain look for alternative moves when playing in he ko was indicated but I think most of the time the alternatives where either better (in terms of score) or it was just about splitting hairs, but, yes, I tried to avoid playing in the ko.
Maybe we need a topic for ko fun
Login
Register
FAQ
Search
Gérard TAILLE
tenuki and white starts the ko and wins it:
tenuki
tenuki