Game 4

[Splatted]

Can someone explain to me? What does that move accomplish?

It is a clever way of capturing P7.

To follow up on this a little bit since I spent a while mulling over the same question. forces black to defend himself by pushing from behind with and , which gives white the chance to atari p7. That forces black to choose between saving the R7/8 stones or the p7 stone which is obviously no choice at all.

I also wondered why black didn't simply connect at R6 for , but white would still be able to capture with a ladder and since the descent/cut at R8 would be sente and very big, black would probably want to play there anyway so the connection would have been a wasted move

[Knotwilg]

upWarrior: I'm awaiting the review at gogameguru. In the preliminary analysis, An Younggil finds the result to be even. That goes to show my positional judgment is worthless.

[oca]

Can someone explain to me? What does that move accomplish?

It is a clever way of capturing P7.

To follow up on this a little bit since I spent a while mulling over the same question. forces black to defend himself by pushing from behind with and , which gives white the chance to atari p7. That forces black to choose between saving the R7/8 stones or the p7 stone which is obviously no choice at all.

I also wondered why black didn't simply connect at R6 for , but white would still be able to capture with a ladder and since the descent/cut at R8 would be sente and very big, black would probably want to play there anyway so the connection would have been a wasted move

Wow... I will let that one for latter ' kilometers from my level...

[Uberdude]

Re move 20: black isn't forced to atari and let white capture the cutting stone: he could defend and fight but you lose quite a lot of points initially for a less certain profit from the fight:

$$c Fight
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . 7 X . . . |
$$ | . . . . . . . . . . . . . . . 4 . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . 5 X O 2 . . |
$$ | . . . . . . . . . . . . . . O X . O . |
$$ | . . . . . . . . . . . 6 . . O X 3 1 . |
$$ | . . . , . . . . . , . . . . O O X . . |
$$ | . . . . . X . . . . . X . . O X X . . |
$$ | . . . . . . . . . . . . . . a . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------

Click Here To Show Diagram Code

[go]$$c Fight
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . 7 X . . . |
$$ | . . . . . . . . . . . . . . . 4 . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . 5 X O 2 . . |
$$ | . . . . . . . . . . . . . . O X . O . |
$$ | . . . . . . . . . . . 6 . . O X 3 1 . |
$$ | . . . , . . . . . , . . . . O O X . . |
$$ | . . . . . X . . . . . X . . O X X . . |
$$ | . . . . . . . . . . . . . . a . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

Probably 4 is the shape, but if white has to answer 5 at 6 then black can come back to answer at 7. White a is sente here which means black's corner is very small and it helps the outside group in the fight. It seems white needs to do it before 6 in fact otherwsise black can hane there and cut and 2 stone squeeze. Difficult fight...

[Bill Spight]

Knotwilg, Bill, but do you think white is already ahead by move 81? If so, what went wrong? Is it that F14 and F12 are significantly worse than D14 and D12?

Caveats:

First, I do not think that the published methods of positional evaluation are accurate, and I suppose that they form the basis for professional judgement, even if they do not capture all of it. For one thing, if those methods were reasonably accurate, computer programs could have used them as a basis for much better evaluation functions than they had in the pre-Monte Carlo era.

Second, in the published methods that I have seen, pros subtract full komi from Black's score. But they should not do that early in the game, only subtracting full komi by the end of the game. I do not know actual pro practice.

Third, in the published methods that I have seen, pros do not make allowance for whose turn it is. One exception seems to be O Meien, IIRC.

OK. By move 27, I think that Black has gained around 6-7 points. I actually think that Black maintains a lead throughout the game, despite losing some ground in the top right, up to . I also think that he loses a bit by playing safe. For instance, I think that says that he is confident of winning if he can nullify White's thickness in the center. No need to contest the left side, which could lead to complications. Also, plays safe, instead of sliding into the bottom left corner or reducing the left side. I suspect that Gu Li thought that he was ahead at these points. It would be interesting to hear his comments.

OTOH, makes me think that maybe Lee Sedol thought that he was ahead at that point. It would be interesting to hear his comments, as well.

[John Fairbairn]

Second, in the published methods that I have seen, pros subtract full komi from Black's score. But they should not do that early in the game, only subtracting full komi by the end of the game. I do not know actual pro practice.

I think this is a vitally important point that merits much wider discussion.

First let us note that Bill correctly makes an implied distinction between what is published (i.e. is aimed at amateurs) and what pros actually do. Although he says he doesn't know actual pro practice I am sure that he is inferring it, and in general terms this is quite possible on the basis of published commentaries.

For example, in commentary after commentary on games in the no-komi era (I'm sure you know I've read more than my share of them, and I was reading yet another this morning where this point came up) there is some reference along the lines of "Black is now ahead". Initially, I used to think, "Hang on - Black was ahead to start with. There was no komi. Where did he go wrong before he recovered?" I can even recall sometimes going back to the beginning of the commentary to check whether there was komi.

What I now realise, though, is that while Black may start in an advantageous position, he actually has to earn the right to say he is ahead. It's more subtle in komi games, but the need to earn bragging rights still applies.

That's Point 1. Point 2 is that it is very, very rare for pro commentaries to give a count until very late in the game. One minor exception early in the game is when an area is clearly cordoned off and you may get a comment along the lines of "Black's moyo has consolidated into 70 points of sure territory", but even then you very rarely see a count for the whole game (instead you get comments like, "So White has to try desperate measures").

This is not so strange when you consider that Japanese (and Chinese and Korean) don't really have a word for early counting in our whole-game sense. Instead they refer to "evaluation of the situation". However, while 形勢 is really just a normal word for 'situation' it is impossible for an Oriental go player not to see both 形 (shape) and 勢 (influence/strength) in this, and to be influenced in how he views the process. The commentaries bear this out by repeatedly talking of evaluation in such terms (形 here, however, means shapes in general, not just 'good shape', and if necessary that can be made plain by using the phrase 碁形).

Of course individual players may prefer to count more precisely but I would hazard a guess that Yi Se-tol, given his style, is not one of them and that he tends to think in terms of shape and influence, especially attacking potential. In the case of the present game, for example, I think I'd expect him to review the lower right not by saying, "I thought it was worth 30 points but it looks like it may only have been 25", but by saying something like, "Hmm - I thought it was thickness but I ended up having to defend that group a little, so maybe it wasn't so thick after all."

Any sort of counting in joseki is absurd in a whole-game context, but even counting fuseki stones in a whole-game context looks iffy. It seems as if the goal of a student for the first 100 moves is not to be able to say X = B - W - komi, but simply to assess reliably things like whether "Black is thick everywhere but White is thin", "White has chances to start a fight", "Black's moyo has great potential", "White is now pleased to have the initiative", and so on.

[RobertJasiek]

Of course, W lost: he used too much of his thickness for making too small territories in front of it, because B preventing efficient W extensions from the W walls.

Bill, what, for positional judgement, do you call "accurate", so that then you say not to know of accurate methods anywhere? Is half a point accuracy not good enough, or do you wish to include influence, strategic options and fighting aspects in what shall become an accurate judgement?

[Uberdude]

Of course, W lost: he used too much of his thickness for making too small territories in front of it, because B preventing efficient W extensions from the W walls.

So you disagree with An Younggil's analysis that white's result in the lower right was ok and that white lost the game with bad choices at the lower left and top in late middle game? I'd be interested to compare your commentary with your numerical evaluation techniques to Younggil's.

[pwaldron]

Point 2 is that it is very, very rare for pro commentaries to give a count until very late in the game.

To add to this, I'm not even sure that pros have a definitive count in their own mind. I remember watching Michael Redmond do a game commentary and he commented that one side was ahead. Someone asked by how much and got a fairly generic answer in response ('quite ahead', or some such).

I piped up and asked Michael how many points komi he would want to switch sides and that stopped him in his tracks. He looked at the board for a good minute before coming up with a number.

[Uberdude]

The book Nie Weiping on Go: the Art of Positional Judgement features whole board counting in the opening.

[gowan]

Point 2 is that it is very, very rare for pro commentaries to give a count until very late in the game.

To add to this, I'm not even sure that pros have a definitive count in their own mind. I remember watching Michael Redmond do a game commentary and he commented that one side was ahead. Someone asked by how much and got a fairly generic answer in response ('quite ahead', or some such).

I piped up and asked Michael how many points komi he would want to switch sides and that stopped him in his tracks. He looked at the board for a good minute before coming up with a number.

This is an interesting point. I doubt that such a "komi" would necessarily be equal to how many points the other side was behind, so naturally Redmond had to think about it.

[Knotwilg]

The Redmond anecdote suggests that pros don't really calculate but rather feel who is ahead, by merely looking at the disposition of the stones. They cancel out certain structures which they recognize, even across the board, by having played and analysed hundreds, even thousands of games. Or rather, they keep track of the exchanges done, in an almost physical way. Since the "badness" of their moves never accounts for more than 2 points, the accumulated difference will rarely go beyond 10 points, or they will resign. That means they need to callibrate the difference between 10 and -10 and probably between -5 and +5, which is a colour palette that can be maintained easily. So I would think they rather think in these terms, like "quite ahead", than calculating the actual difference by assigning counts to each group or territory. That would be the reason why Redmond needed to think about the actual difference, while he felt very clearly that one player was "quite ahead".

[RobertJasiek]

No, the Redmond story tells something different: that also professionals need time to count accurately. Redmond was reported to calculate accurately at other times, when he was playing and had the thinking time for it. But during a live commentary, it is possible that he does not always have the time, until asked specifically for the count.

Uberdude, after move 27, the W wall has only ca. 4 significant influence stones. K3 is not really a threat yet. Therefore, the W excess of influence stones (B also has some!) is small and cannot compensate B's big lead on territory.

[Bill Spight]

Bill, what, for positional judgement, do you call "accurate", so that then you say not to know of accurate methods anywhere? Is half a point accuracy not good enough, or do you wish to include influence, strategic options and fighting aspects in what shall become an accurate judgement?

It is true, as John Fairbairn says, that pro commentaries seldom talk numbers until late in the game, and that is, IMO, appropriate. However, many years ago I was surprised to read in an article in one of the go magazines about an instance when some young pros (less than 5 dan in those days) were discussing a game and had laid out over 20 moves when another young pro walked in, took a quick look at the board which, as I recall, had a White wall in the top left, and said, "Oh! White is 3 points ahead." That was just what the other young pros had agreed on. That tells me that pros know heuristics for estimating the numerical value of positions early in the game. In this case these pros came up with a precise estimate of three points. That does not mean that it was accurate.

It is possible to be accurate, even though there is a great deal of uncertainty about the outcome of a position. For instance, suppose that White has enclosed a Black group which is unsettled. In one play Black can live and make 2 points; in one play White can kill and make 28 points. An accurate estimate of the local position is 13 points for White, even though the eventual result will be 15 points different.

Now, the traditional estimate of the value of a single stone on a 4-4 point is 10 points. We now know that that is too low. The actual value is around 14 points. (At least, early in the game, before surrounding stones may have altered its value.) Suppose that, early in the game, several stones have been played on the left side of the board, while Black has nirensei on the right side. Anyone using the traditional estimate for the value of those stones could underestimate the value of the whole board for Black by as much as 8 points.

Consider a 3-4 5-3 shimari. Traditionally, that is valued as 11 points, because that much territory is relatively secure. However, the traditional value does not take into account the possibilities for development from the shimari. Nobody knows the actual value of the shimari, but 25 points is more like it. Maybe 30 points.

[MJK]

Well, professional players have played thousands of games throughout their life; they must have a strong intuition about the 'tendency' of the game. If An Younggil thinks the position was even, it is likely to be, and also pros talk with each other about such big matches, so what An has written is based on not only his own opinion but also other pros'. If there had been disagreement he would have said that some pros have different thoughts rather than clearly stating 'the game is even'.