Discussion reference:
viewtopic.php?p=163920#p163920

Bill, just read the appropriate books that explain counting during the opening and middle game, such as Positional Judgement 1 - Territory.

A black 3-4/5-3 corner enclosure has the value (2 | 11 | 2), a white enclosure (-2| -11 | -2), where the first parameter states the stone difference, the second states the territory count and the third states the influence stone difference. (Especially during the opening, this can be transformed to a value triplet with stone difference 0.)

This is not a complete description, because aji, options, fighting aspects etc. are not described by these numbers. It is also not a single number description, as the miai value attempts to do. However, the description can be interpreted and applied reasonably easily by (experienced) players. For probably a couple of centuries, you cannot achieve the same with miai values, because you cannot often determine them during the opening or middle game.

This sounds like confusing counting local positions with evaluating the overall position.

In other words:

Global territory: use imagined global reduction sequences to settle all the local territory regions of a player.

Global influence measured by the influence stone difference is the sum of the local influence stone differences, because only the significant outside influence stones are counted, and outside (wide essentially empty regions adjacent to both players' live stones) is given in the global position.

IOW, local and global counting techniques are closely related.

Without diagrams and more detail, it's impossible to figure out how you're coming to a count.

For territory, I have explained the basics in different threads here. Search and you can find it.

For influence, it is much easier: the influence stone difference is number of Black's significant outside influence stones minus the number of White's significant outside influence stones. (Do not perceive borders visually, but e.g. a ponnuki can have also further significant influence stones behind the hull.) Since influence is defined mainly by degrees of connection and life, the significant stones are identified as those with great impact on connection and life for affected intersections and their stones. (Therefore, usually 1st and 2nd line stones are not sufficiently significant.) It often is not necessary to be exact; a value plus-minus 1 is good enough.

I'm talking about your count for this specific position. You seem to just be hand waiving. I can't follow your starting position of what you're trying to evaluate or even your end result of how much better black or white is.

Why don't you start with posting the position and at least the end results of your evaluation if you want some discussion. It would help to describe your thoughts on the process if you want useful discussion.

I am surprised you said Lee's wall in the game was only worth 4 influence stones: as it's basically a wall of 6 stones I expected something nearer 6. I understand a wall down to the first line isn't getting so much value out of the 1st line bit (but even that can be useful for reverse money jumps into territories), but Lee's wall of 6 all seem like significant influence stones to me, so why only 4?

It makes sense to me that, with Gu Li's group in the lower left, the wall wasn't going to influence very much.

Yes, but afaik that more global consideration is not part of Robert's evaluation function of counting significant influence stones. (Btw the wall does have the effect that black is more reluctant to enter the lower left corner at 3-3 as it will weaken his outside group which would be unpleasant with that looming white influence to the right). I suppose the point he's getting at is that this position:

Click Here To Show Diagram Code

[go]$$B
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . X , . . . . . , .
$$ | . . . . X . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ -----------------------[/go]

has more outward facing influence than:

Click Here To Show Diagram Code

[go]$$B
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . X , . . . . . , .
$$ | . . . X . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ -----------------------[/go]

despite them both having 2 stones facing the outside: the first is more efficient and the influence of the 2 stones overlaps less. His evaluator would also say the normal shimari has more territory so better for that too. So if you have a solid wall of 6 stones you are saying that their influence overlaps about 2/3 giving 4? This seems a reasonable sort of judgement to make but I thought Robert's evaluator was trying to be more algorithmic and rely less on the intuition and judgement skills of a Go player.

Also the wall was made in sente which has to have some value associated with it at that point in the game if you're trying to get a count. That's why I'm still unsure whether he's trying to make up a local score or a game score. At the time he felt white was behind, and I was curious why.

That's local.

Wall in the game assessed by the number of influence stones relevant for the influence stone difference: only stones with SIGNIFICANT OUTSIDE influence are counted. Therefore, stones hidden behind the bottom black group are not counted.

A white wall of 19 stones adjacent to a black wall of 19 stones has 0 influence stones, because the wall "protects" only a neutral region, which is not on an outside.

Shimari: every shimari (adjacent to a wide empty space) has the influence stone difference 2. Different shimaris can have different degrees of influence, but the influence stone difference does not measure with this precision. If you want to measure precisely, apply my formal influence definition separately to each intersection (in the neighbourhood of the shimari) and determine the black and white n-connected, m-alive and the t-territory values. For each intersection, you get a 5- or 6-tuple of values. Presumably a keima shimari gives better average values than a kosumi shimari, and this can then be interpreted as a confirmation of greater efficiency (in an ideal empty position).

The influence stone difference is closer to human subconscious thinking - the formal influence definition and its values is closer to algorithmic thinking.

Sente: Sente can, but need not, be stated as another parameter. Alternatively, a local position can be assessed without sente; instead one can have the view of two possible cases: Black moves next locally or White moves next locally. In the global judgement, the right to make the next turn must be considered.

I wonder how the tygem estimate button would count it. I think it is kind of peculiar that when someone plops down their hoshi stone, it counts it as three points, my san san gets ten points. I know that's pretty far from what most people count those openings, but it isn't far from how I think about it, where a hoshi stone actually only gets influence.

The opening hoshi's 14 points miai value is split in 7 points of 'current territory' and 7 points of future territory to be made by its influence.

http://senseis.xmp.net/?CurrentTerritory