Something I don't really get about tewari...

As I understand, the idea behind tewari is: a good move is a good move, regardless of order. So you can change the order of play to analyze the position, and see if the moves still seem good.

Here's an example from SL (http://senseis.xmp.net/?TewariExample6):

Click Here To Show Diagram Code

[go]$$W [Hikosaka Naoto] (W) - [Mimura Tomoyasu], 2007
$$ +---------------------------------------+
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X . . . . . . . . . . . . . . . |
$$ | . . a . . X . O . . 7 . . X . . . . . |
$$ | . . . O . . . . b , . . . . . , X . . |
$$ | . . . . . X 1 3 . . . . . . . . . . . |
$$ | . . . O . . 2 . . . . . . . . . . . . |
$$ | . . . . . 4 . . 6 . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . 5 . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . O . . O . . . |
$$ | . . . O . . . . . , . O . . . , . . . |
$$ | . . . . . . . . . . . . . X . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ +---------------------------------------+[/go]

White didn't want black to play 'b', hence the attachment at .

Tewari analysis gave evidence that this is bad:

Click Here To Show Diagram Code

[go]$$W Tewari
$$ +----------------------
$$ | . . . . . . . . . . .
$$ | . . . X . . . . . . .
$$ | . . . . . X . O . . .
$$ | . . . O . . . . . , .
$$ | . . . . . X 5 1 . . .
$$ | . . . O . . 6 . . . .
$$ | . . . . . 2 . . 4 . .
$$ | . . . . . . . . . . .
$$ | . . 3 . . . . . . . .
$$ | . . . , . . . . . , .
$$ | . . . . . . . . . . .[/go]

Changing the order, we see that the 5-6 exchange is bad - who would play 5 in this scenario?

What I don't get is the reason that the ordering doesn't make an impact on the analysis. For example, in the tewari version of the position, if white plays 1 as above, black doesn't necessarily have to jump at 2. Maybe he takes the 3-3, for example:

Click Here To Show Diagram Code

[go]$$W Tewari
$$ +----------------------
$$ | . . . . . . . . . . .
$$ | . . . X . . . . . . .
$$ | . . 2 . . X . O . . .
$$ | . . . O . . . . . , .
$$ | . . . . . X . 1 . . .
$$ | . . . O . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . , . . . . . , .
$$ | . . . . . . . . . . .[/go]

Note: This may very well be inferior in this example, but at least I don't feel as much pressure as black to respond. I'm more likely to consider alternatives to the jump, whereas I'm very likely to respond to the attachment played in the game.

From this, I see a benefit to the game play, which cannot be found in the version where the moves are switched: the attachment that white played in the game induced a response from the opponent in a way that the tewari version (with switched order) does not.

My observation from this is that the order of moves plays a role in "forcing" a particular response from the opponent. A good example is the timing of a probe - you *force* your opponent to make a decision, and act accordingly. Given this, is it really a fair comparison to swap the order of moves in order to prove that a particular play is bad? The order of moves seems to play a big role in *controlling* the flow of the game, even if the resulting board position results in a shape that seems weird or inferior under a different order of play.

Does my confusion make sense? If not, maybe I can word it in a different way...

_________________
be immersed

I don't think you are confused in your analysis, but maybe you are expecting too much from tewari. Tewari (there are two types anyway but let's stick with the move order one) is not a dynamic technique for finding the best move to play. It is a static test to determine whether there are any inefficiencies in the position. The inefficiencies may go back further than the moves just played.

You can consider it a kind of DNA test, as in Who Do You Think You Are. It may tell you a little about yourself but what it really opens up is your ancestry, which may include, say, a skeleton in the closet. Similarly in go, your real mistake may have been a move much further up the tree than you are looking, e.g. the stone at H17.

That information may not be useful in the current game (i.e. too late or too complex to be useful) in correcting the mistake, but it still has value in that at least some pros keep a tally of inefficiencies as a way of estimating the score. Amateurs probably make too many mistakes to make this procedure worthwhile, but pro examples I have seen will typically highlight three or four inefficiencies. By assuming the rest of the play is perfect they seem to get a pretty accurate assessment of who is ahead, and the technique is especially useful in quick games. We amateurs can do this by playing over pro games, and I find it's then quite easy to make an assessment in about the first half of the game (but having a commentary that highlights the mistakes helps enormously, of course).

In other words, once you know a move is a mistake, you can still extract value from that mistake without knowing how to correct it and without considering alternatives. So tewari is not just about improving your josekis.

I was surprised, some years ago, to find that a lot of strong amateurs, at least in the West, do not place much stock in tewari. IMX, however, when I was an SDK one of the most eye opening things I learned was shape. I also inculcated the idea of efficiency. In fact, from 4 kyu up to 3 dan I was a shape player. (Then I learned from Sakata that shape is not enough. )

As a shape player I was often intent on making my opponent take bad shape, that is, locally inefficient shape. That's one reason that it bugs me for people to say things like, The Table Shape is a good shape. No, it's not. You can't just say that. It has to be efficient to be good shape. Which means you have to take the opponent's stone into account.

Example:

Click Here To Show Diagram Code

[go]$$Bc Old tsuke nobi joseki
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . 8 . . . . . . .
$$ | . . . . . 4 . . . . .
$$ | . . . O . 2 3 . . , .
$$ | . . . 6 5 1 . . . 7 .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

makes a table shape, but it is bad shape, because it is inefficient. If that is not obvious — and it was not obvious for many years — tewari shows why.

Click Here To Show Diagram Code

[go]$$Bc Tewari
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . 2 . . . . . . .
$$ | . . . . . 6 . . . . .
$$ | . . . O . 4 5 . . , .
$$ | . . . 8 7 1 . . . 3 .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

Up through is joseki. But then and are obviously bad. True, Black has alternatives to , but is good. That's all you need to know. Tewari is a tool to bring inefficiency to light. Since my aim was often to induce inefficiency, I used tewari a lot.



Well, it does. The plays up to are not random. Nor is . The point is that - is normal, if not joseki.



This is a question of judgement. If at "a" does not look at least OK, then you don't get the tewari comparison.

Click Here To Show Diagram Code

[go]$$W Hikosaka Naoto (W) - Mimura Tomoyasu, 2007
$$ +---------------------------------------+
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X . . . . . . . . . . . . . . . |
$$ | . . . . . X . O . . 7 . . X . . . . . |
$$ | . . . O . . . . . , 8 . . . . , X . . |
$$ | . . . . . X 1 3 . . . . . . . . . . . |
$$ | . . . O . . 2 . . . . . . . . . . . . |
$$ | . . . . . 4 . . 6 . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . 5 . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |[/go]

Moi, I don't need tewari to tell me that, in this position (Edit: That is, the resulting position), the - exchange is inefficient. And I daresay neither did Hikosako. But he judged that getting both and was worth it. So you have company, but the commentary disagrees.

BTW, Mimura continued with with the aim of making a decidedly bad play.

Anyway, tewari requires judgement. Often you can use joseki as a standard for comparison, and - in the tewari diagram are joseki or close to it. The fact that you might prefer a different play for is irrelevant unless you think that in the tewari sequence is not good. (And you admit that may be better than the 3-3. That's all the tewari needs.) You have to rely upon your own judgement. Tewari is just an aid to judgement.

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

Visualize whirled peas.

Everything with love. Stay safe.

Hi Bill,This local situation is interesting to me, because given:

Click Here To Show Diagram Code

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

Click Here To Show Diagram Code

[go]$$B
$$ | . . . . . . . . . . .
$$ | . . . W . . . . . . .
$$ | . . . . . O . . . . .
$$ | . . . O . O X . . , .
$$ | . . . O X X . . . B .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

It's not clear to me why the exchange ( - ) is bad for White.
If the local result is bad for White, maybe the problem is not the exchange ( - ).
Maybe the problem is :

Click Here To Show Diagram Code

[go]$$B Bad for White ?
$$ | . . . . . . . . . . .
$$ | . . . O . 2 . . . , .
$$ | . . . . . 1 . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

Because starting with this attach,
White ends up with:

Click Here To Show Diagram Code

[go]$$B
$$ | . . . . . . . . . . .
$$ | . . b a . . . . . . .
$$ | . . . . . O . . . . .
$$ | . . . O . O X . . , .
$$ | . . . O X X . . . X .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

And given this local situation, it's not bad for W to add a move (say, at (a) or (b)).
So the old sequence with the one-space jump = (a) is the price W pays for starting with .

Which would agree with the tewari's "bad" exchange of ( - ):

Does this line of reasoning make any sense ?
( Really asking, not rhetorical. )

Ed

I think this is, in some ways, a poor example to use to discuss tewari, so maybe any line of structural reasoning is academic. Certainly White 8 is regarded as too solid (C10 is given as the correct move). But to justify that you need to know that Black cannot push in and cut, and that takes a lot of tactical nous. For example, the Kitani joseki dictionary gives 11 variation diagrams, nearly all going to over 40 moves, with quite a few eye-popping tesujis along the way, and usually while the conclusion favours Black it's not a killer variation - just "Good for Black" or "Black is thick." And since there are lines where White can get a ko, the relationship with the rest of the board can come into it.

Also there is a tactical element in that White's extension to K3 is part of the equation as the cut at G3 has to be available for Black.

There is also the interesting side question of why C10 is the correct move and not D10 (as in a big handicap-game position). The reason is to do with inviting White in to the gap and then attacking him as he cannot make a base, but that's a lot to ask for a handicap player, and so the tight move has become a crutch that weaker players rely on, to the extent that it can be called joseki in the true sense (or a "handicap-game joseki" as it is often referred to).

There is not a single example in pro-pro play, with the wide White extension, but with tighter extensions White 8 does appear in pro play, though it is still quite rare.

Well, that's the value of tewari. If it was clear, then there would be no need for tewari.



Well, when I was just learning go, the tsukenobi was popular, at least among kyu players. It may have been in a magazine aimed at kyu players that I first saw the tewari I have shown. In any event, I did not run into anybody who played Wa or Wb next in the last diagram. As John indicates, the extension to C-10 was joseki, and that is what you saw.

I gradually gave up the tsukenobi for the immediate jump to "a" or sometimes the keima to "b", because, as John says, the extension to C-10 allowed irritating invasions. So what was the point of the tsukenobi? Usually if I played the tsuke, it was with a different joseki in mind.

Click Here To Show Diagram Code

[go]$$B Tsuke osae
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . 5 . . . . .
$$ | . . . O 6 2 3 . . , .
$$ | . . . . 4 1 . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

As John says, after White extends to C-10 Black can push and cut. The go magazines showed how to handle a number of those sequences. But there was another way I learned, which I usually used if the situation arose.

Click Here To Show Diagram Code

[go]$$Bc Sacrifice two stones
$$ | . . . . . . . . . . .
$$ | . . O , . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . 6 . . . . . . .
$$ | . . . 4 5 . . . . . .
$$ | . . . 2 3 W . . . . .
$$ | . . . O 1 W B . . , .
$$ | . . . O X B . . . X .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

Let's do a tewari on that. We remove the two and stones.

Click Here To Show Diagram Code

[go]$$Bc Tewari
$$ | . . . . . . . . . . .
$$ | . . O , . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . O . . . . . . .
$$ | . . . O X . . . . . .
$$ | . . . O X . . . . . .
$$ | . . . O X . . . . , .
$$ | . . . O X . . . . B .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

The stone is only four spaces from a four stone wall, which is perhaps just slightly inefficient, but OK, given the White strength nearby. Meanwhile, White has gotten to crawl along the fourth line. This result is good for White.

Now, it may well be that, in general, the tsuke is not best. The modern way — and AlphaGo's, too, unless it is ahead — is to avoid settling positions early on. Here is another example of that, from the old days.

Click Here To Show Diagram Code

[go]$$Bc 20th century joseki
$$ | . . . . . . . . . . .
$$ | . . . , . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . 3 . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . 4 . . . . . . .
$$ | . . 2 . . . . . . . .
$$ | . . . , . 5 . . . , .
$$ | . 6 . 1 . . . a . 7 .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

This joseki was popular when I was learning go. What I only learned much later was that it had replaced an earlier version, which still appeared as joseki in Suzuki-Kitani, but which was no longer considered joseki a few years later.

Click Here To Show Diagram Code

[go]$$Wc Older 20th century joseki
$$ | . . . . . . . . . . .
$$ | . . . , . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . X . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . 1 . . . . . . .
$$ | . . O . . . . . . . .
$$ | . . . 5 . 2 . . . , .
$$ | . 7 3 X 6 . . . . 8 .
$$ | . . 4 . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

Pros stopped playing because it strengthened Black. The simple keima in the previous diagram allowed the invasion at "a".

There are other examples where jump attachments stopped being joseki.

----

I somehow got the impression that the tewari I showed was discovered only in the early 20th century. You certainly saw the table shape in old games. However, the extension to C-10 occurs in Okigo Jizai. But Hattori also includes the table shape, as well as various other extensions. OC, they are in handicap games, where Black might take a small local loss to solidify his territory. But I think that that tewari was known by the early 19th century.

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

Visualize whirled peas.

Everything with love. Stay safe.

It is usually attributed to Dosaku (i.e. late 17th century) but I've seen nothing to back that up. As time passes and legends grow in the usual fashion, we may see Dosaku quoted as the inventor of smart phones.

The fact that Hayashi Yutaka avoids referring to Dosaku (or anyone else) may be significant, and there is also the fact that around the 1920s (when proper go journalism started) tewari and ishiwari and other terms were used for the same thing. That suggests two possibilities to me. One is that it was a new concept which had no settled term yet. The other is that tewari and ishiwari referred to the two different strands (dynamic and static) which are now subsumed in tewari. These strands may have been older and really only came to light once go journalism started, and it was a journalist who merged the two. I'm inclined to favour the latter possibility. However, Honinbo Shusai was around in the 1930s and very active in supporting go journalism and so he may have been the one who passed on an old Honinbo tradition, and one that they may well have wanted to keep secret. There's the plot of a go novel I there somewhere

The Table Shape is a good shape when that shape is needed. Context is everything. For example if W knows that they win the game by sealing off the left hand side securely, it's a good shape in that game.

I think the other kind of tewari (stone removal) is more convincing.

I was referring to that specific tewari.



Thanks, John. Actually, ishiwari is the term that I first learned. When I ran across tewari much later, I figured that I had misremembered or misread something years before.



The idea that Shusai may have spilled the beans is interesting, John. We live in an exoteric age, don't we?

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

Visualize whirled peas.

Everything with love. Stay safe.

Well, we can do stone removal instead of permutation.

Click Here To Show Diagram Code

[go]$$Bc Before stone removal
$$ | . . . . . . . . . . .
$$ | . . . O . . . . . . .
$$ | . . . . . O . . . . .
$$ | . . . O . O X . . , .
$$ | . . . O X X . . . X .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

Click Here To Show Diagram Code

[go]$$Bc After stone removal
$$ | . . . . . . . . . . .
$$ | . . . O . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . O . . . . . , .
$$ | . . . O . X . . . X .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

Click Here To Show Diagram Code

[go]$$Wc Before stone removal
$$ | . . . . . . . . . . .
$$ | . . . , . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . X . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . O . . . . . . .
$$ | . . O . . . . . . . .
$$ | . . . O . X . . . , .
$$ | . O O X X . . . . X .
$$ | . . X . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

Click Here To Show Diagram Code

[go]$$Wc After stone removal
$$ | . . . . . . . . . . .
$$ | . . . , . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . X . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . O . . . . . . .
$$ | . . O . . . . . . . .
$$ | . . . . . X . . . , .
$$ | . O . X . . . . . X .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------[/go]

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

Visualize whirled peas.

Everything with love. Stay safe.

Yep. Looks slow for white.



This modern preference (okay, not so modern, that second joseki is quite old itself and a bit out of fashion today) for unsettled positions is a bit harder for me to get my head around. I suppose a willingness to fix the position when one is ahead is natural as long is there is confidence in the lead. Maybe in the pre-komi days with no time limits it was easier for black to gain that confidence? Stylistically, even with modern players it seems to vary from player to player. The "My God, it's full of aji!" position may be comforting to one and nerve-wracking to another.

I don't think that I was clear.

In one kind of stone removal, you take away an even number of stones from each player that (more or less) count the same and judge the result. In the second kind of stone removal you take away an even number of stones from each player and compare before and after positions. If the after position is better for one player than the before position, then the stones of that player which were removed were less efficient/effective than the stones of the other player which were removed.

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

Visualize whirled peas.

Everything with love. Stay safe.

To counteract the common impression that tewari is about josekis, and also to show how tewari is used as a supplement to counting, here is a summary of a discussion on a game between Fujisawa Hosai (Black) and Kitani Minoru. Takagawa Kaku also took part in the discussion.



Black has just played the triangled stone. This came immediately after the completion of a ko fight in the top right where Fujisawa had to give way and so ended up, he thought, having made a loss. This was explained by tewari. (For those who insist on counting the abacus way instead, Black has captured 6 stones and White has captured 8, and the komi was 4.5, and Fujisawa himself counted the top side as close to 50 points.)

First, in the top right, the position was first shown as follows to show where the captured stones were and which stones can be removed in pairs:



The result is:



Here we can see that Black has not only made the silly exchange of the circled stones, an obvious loss, but the triangled stone is redundant - the loss of a whole move.

Black did get the move in the lower right, previously triangled, but here too tewari shows a loss for him (Takagawa's analysis):



Removing the triangled pairs, we can see that Black has inefficiently added two not very useful circled stones without reply. So the loss of the ko in the corner was "no big deal" for White, Mr T said.

Despite these inefficiencies by Black, which he lamented (starting the ko without sufficient ko threats was his biggest gripe), he won the game by 6.5. The commentary doesn't say, but there were no dead White groups (he even captured some Black stones) so I imagine that Black making territory on the fourth line was the decisive factor. It's just that Black should have won bigger.