This is stimulated by the disagreement in the thread on Game 4 of the Mlily match between Gu Li and Yi Se-tol, in which amateurs appear to look askance at Yi's positional judgement whereas pros seem to think his play was fine.



As I read it, in summary, Robert was unequivocal that Yi's play was bad and in the absence of precise count-based explanations from the pros, he claims his method is the best way to judge the position. Other amateurs say "we haven't a clue why, but if the pros say it's OK it must be OK."

I lean towards the latter camp, but I also anyway oppose rather strongly the idea that in the present state of knowledge the essence of early go positions can be usefully captured by numbers. To me it's still like trying to evaluate a painting by Rembrandt by saying he used 12 colours and 75% of his brush strokes were diagonals.

However, it is incumbent upon us all to have open minds, and I would like to invite Robert to give his evaluation of the position above (Black to play, no komi).

I chose it because I think it demonstrates quite well the aspects that have come up in the discussion. There is some thickness and clearly defined territory, and the final areas are fairly well adumbrated already (I say this on the basis of knowing the final position) but several of the positions are hard to define as thick or thin and even the boundary lines not in the centre are hard to pin down. In other words, there is still quite a lot of territory to play for.

Also, I have the pros' non-numerical assessments (they seem to agree), and in that connection there is a item of assessment that is typical of pros and not amateurs which was alluded to in the other thread and which will apply here if you also wish to guess Black's next move - which is after all the point of the evaluation.

So, Robert, here is a chance to convince us without (I assume) you knowing how this game continued and ended. Fortunately you would be the last person I know to look it up before answering.

I hasten to add that I myself feel rather at sea in this sort of position, and so would grasp at any means of giving a compass bearing.

(I have not said that Yi's play in general was bad. I do not claim that numerical territory and influence counting is my method, but that it is an important part of my method. Every judgement method should also have numerical information; this, however, does not imply that my method of positional judgement is the best, because other aspects of my positional judgement can still be good or bad. When different methods are bad at different aspects, one cannot say in general that one method is better than another; one can only say that the judgement of some aspects of a particular method is better than the judgement on these aspects in other methods.)

I download the SGF and study it a bit to see if I can provide a judgement within reasonable time or whether I would need more free time than I can currently invest.

Since the SGFs are not shown properly here, you need to download them and view them in your own SGF editor.

Very nice to see an illustration of your principles in action, Robert, and despite the fact that you may have come to a different conclusion as the pros (you see it as an even board whereas the pros apparently see one party ahead) I doubt many others here could give such a well backed up opinion. Also impressive that you did the analysis and made the diagrams so quickly.

_________________
Patience, grasshopper.

John never stated what the professional consensus was, so it's a bit presumptuous to automatically assume that Robert's differs from that consensus.

5 minutes thinking plus 64 minutes editing:)

Although it is not explicit, "Also, I have the pros' non-numerical assessments (they seem to agree)," suggests to me that the pros saw one side as having an advantage, whereas Robert writes: "In this rough description, the position (with Black's turn) is fair." Please also note my use of the word "may." I hope I wasn't being presumptuous.

_________________
Patience, grasshopper.

Hi daal, I was also going to point out your "may," but you ninja'd to it. Also, I re-read it, and I wonder if the wording could be improved on? Because I got a similar feeling as the usage "You may have won this time, but..." Maybe I'm totally confused.
(How about something like: "Whether or not you came to the same consensus as the pros, ...") As Robert mentioned, so much in editing. I read John's wording to mean at least 3 possibilities:
- the pros agreed one side was ahead;
- the pros agreed the game was even;
- the pros agreed the situation was very unclear (difficult to say who's ahead).

What suggested to you it was the first case?
John just said they seem to agree, but he didn't specify on what.
Or did he?

_________________
http://tchan001.wordpress.com
A blog on Asian go books, go sightings, and interesting tidbits
Go is such a beautiful game.

Tchan, I didn't follow Game 4. I didn't look at that game. I didn't know whether the board in this thread was from Game 4, or another, completely unrelated game.
No player info in the SGF in this thread.
I still haven't looked at Game 4, and I don't know who won, or what happened in that game.
I read this as a stand-alone, self-contained new thread.

I thought John picked a different game for this exercise.
It could be from Game 4; maybe I missed something in this thread that explained it is from Game 4.

Also, even if it's from Game 4, it could be from a different moment in the game,
than what was discussed in the other thread. I thought that's part of the point of this exercise: to examine a new board position that hasn't been discussed in that thread.

I read John's introduction, including the part you highlighted. It does not specify whether anything that follows would be from Game 4.

Well, I am also one who thinks that Gu Li was in the lead after . If I am sticking to my guns it is mainly because, by comparison with other amateur dans, I tend to evaluate outside influence more highly. So if even I think that White's thickness is not sufficient compensation for Black's territory, maybe it is not. To be sure, the position is playable for White, but that is not the same as equal.

I certainly do not think that the essence of any go position before it is scored can be captured by a number. However, in a funny way the value of early go positions can be reasonably estimated. For instance, I think that the value after an initial Black play on a 4-4 point is around 14 points. Not only does the value of komi suggest that, I have a method of calculation that gives that figure.



This position is not a good one for my methods. In a way, it is too late in the game. That is because I do not have a good way to assess the strength or weakness of groups. Early in the game that is not so much a problem. In this position, though, the White group on the bottom side is weak, and Black can mount a sustained attack.

If the group were immortal, so that Black had no threat against it, now or later, then my guess is that White would be around 18 points ahead. But plainly it is not. If it were completely dead, so that White could not even make any territory by threatening to save it, my guess is that Black would be around 24 points ahead. That is a huge difference. (Earlier in the game you are not likely to get such huge differences. )

Shall we think of the group as half dead? No. That would mean that Black to play could kill it. Maybe a better guess is that it is 1/3 dead. In that case we could estimate White as being around 4 points ahead on the board. But there is a very large uncertainty in that estimate.

What about the value of Black's sente? Normally, in a quiescent position at this stage of the game it would be worth somewhat less than full komi, around 4 or 5 points. That would make the game about even. But if White runs into the center, he would be running into Black's sphere of influence. Black will have a strong attack. Sente might even be worth around 10 points. Again, there is a great deal of uncertainty.

So maybe Black is around 5 or more points ahead in the game, give or take many points. (White is ahead on the board, but Black's sente puts Black ahead in the game.)

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Ed, I haven't read the other thread either and I still read this as meaning that it's from game 4. I have no idea whether it's a different moment in the game from that which you think it is from. So I just think of this as a new thread as much as you do. BUT unlike yourself, I do think that it's about a moment in Game 4.

JF in his first sentence mentions Game 4 then he talks about Robert being unequivocal that Yi's play was bad (which later on RF tells us he hadn't said so). Then he later talks about "the pros' non-numerical assessments (they seem to agree)".

So from this I gather:
Something about game 4
Something about Yi Sedol
Something about pro's opinion on Yi's play being fine

If JF does not reference any other game for the position he chose, why would you think that he had picked a position from a different game? Would a different game actually be more useful to see the value of RJ's methods compared with Yi Sedol's positional judgement in a particular moment of Game 4?

_________________
http://tchan001.wordpress.com
A blog on Asian go books, go sightings, and interesting tidbits
Go is such a beautiful game.

The position referenced in this thread does not have anything to do with game 4, apart from the fact that there is an assessment to be done, and professional opinion to compare it with.

ok, thx skydyr

_________________
http://tchan001.wordpress.com
A blog on Asian go books, go sightings, and interesting tidbits
Go is such a beautiful game.

I find it hard to imagine pros agreeing on the second, and certainly not on the third possibility.

_________________
Patience, grasshopper.

I didn't. (And, why not?) I try to keep an open mind. I didn't know where the position came from, and I kept it that way until there is clear evidence one way or the other.Yes, actually, very much so. The more data points, and the more varieties, the better. My understanding is it's less about comparing with the assessment of a particular pro, but rather with a consensus among some pros.

Why? (I still haven't looked at Game 4, or the board in this thread, or any of the analyses. Maybe it's very clear to you. )

Ed, this is pure speculation on my part about what I think pros would or wouldn't agree on, and also speculation about John's intention with his post, and I could be utterly wrong on both counts, but it has nothing to do with the game itself, and quite frankly, my speculation is really not the interesting aspect of this thread. The interesting part of this thread is the question whether numerical assessments, particularly Robert's, lead to similar conclusions as the pros. Let's wait and hear what the professional consensus was.

_________________
Patience, grasshopper.

Thank you, Robert. I would assess your evaluation as very creditable - probably as good as the pros', maybe better, and easier to understand. But different.

The game is Game 4 of the Honinbo title match, after the Atom Bomb game. Hashimoto Utaro is White, Iwamoto Kaoru Black. I deliberately chose it as not being special in terms of content, so as to create a fair test, but one thing special about these games was that, as they weren't published at the time, both players simply recorded their thoughts after the game, and so we have direct evaluations from both sides and untrammelled by any philosophising from observers.

Obviously we are still not getting the full SP, but the way the players gave their assessments seems fully in line with the way we see it in more normal commentaries.

First, Iwamoto (perhaps being modest as Black) thought White could be rather pleased with his position when he played the last move (White 62) shown on the left side. He said the "balance of territories has now become equal." Whether or not you agree with this, there are problems with it, which I will come to in a moment. He also thought that whilst Black now seemed obliged to attack the White group on the lower left side (but initially from underneath rather than on top as Robert suggested) in order to reduce the White moyo, he could expect little joy from that. Presumably for that reason he did not attack the lower right group at once, but played a move (Black 63) at the top (Robert's move C = L18). However, he did not call it a reducing move, as Robert did, but focused instead on another aspect, which was turning the White stone in the upper right into a mochikomi stone (ending up with the effect of a wasted invasion). White had played that as a probe but it was premature. If he had deferred it he could have answered Black 63 in different ways. The reducing aspect must have been there, but apparently not uppermost in his mind. This was so important to him that he played the Black 67-White 68 exchange to confirm the capture, even though this strengthened White's moyo. Note that Robert did cover this point by referring to aji intersections, but he appeared to prioritise the moyo over the aji at this juncture.

There are a couple of tactical points to make, too. One is that Black's lower left group was not as secure as it looks, and both players seemed aware of that (one result was that White got more on the left side than Robert bargained for). Another is that White had a wicked move at M8 (White 78) which Iwamoto did not see, and that impacted both on the moyo and the safety of Black's lower right group.

My own impression is that at this stage the pros and Robert coincide well enough, but here (as in many other commentaries) the pros place a lot of emphasis on factors such as mochikomi stones. I sense they give much more weight to these than amateurs ever do, and I speculate that the reason may be that they can assume that in their games virtually every move both sides make is at least satisfactory, and so a rather obvious case of a wasted stone (a connection on a dame point is another example) is enough for them to conclude that the scales have been tipped.

But if that's the case, their method is as useful to us amateurs as a fart in a perfume factory. Most of us buff our nails if we can make a couple of satisfactory moves in a game!

Reverting now to the "balance of territories" problem I mentioned. From wide experience, I don't think this implies a close count of the type Robert makes. I think that at this stage, and much later, they usually rely on heuristics such as mochikomi or virtual territory (look at the area of your moyo and count half for yourself on the principle that the opponent will invade in one half but live with only two points and in the process will allow you to form a solid wall round the other half - there are variations where you count a third).

Further, in this game, there were several subsequent comments that the game was close, though edging in Black's favour. That suggests to me an ongoing refinement in the assessment, simply because more moves have been played, and (as is common elsewhere) these comments nearly always coincide with remarks about things becoming settled, so that the losing side is running out of areas to skirmish in (as well as territory being easier to count, of course). This also suggests to me that the early evaluations are tempered to a very large degree - perhaps much larger than Robert allows? - by the scope for messing around. Some sort of territory count must be in the mix, but may even be the most trivial aspect in many cases.

However, one further complication is that we are told repeatedly that the game is close, but then Black makes a horrible mistake (199) which was worth 10 points, according to its perpetrator Iwamoto. Since Black won anyway by 3, and this move was near the endgame, it seems to suggest Black was actually about 13 points ahead until then - which is nothing like close. Since other examples of this can easily be found, we are faced with various possibilities, including hyperbole ("ten" points means 1 point or OMG!!!!) or a weird definition of "close" (e.g. close by amateur standards). Robert's method avoids this woolliness.

Here is the full game.

G2 and F2 weakened the lower left black group, so of course White can make a bit more territory on the left side afterwards.

I considered attacking the lower left white group from above or below and saw little difference in the developments, so I simply mentioned an example of an attack.