Tewari Example 1

Not joseki  

In two games of mine all players were strong amateurs (U. S. 5-dan plus). In both games the play went the same through . White has deviated from joseki. Let's do a tewari analysis after .

Pairing stones  

The stone was played in response to a black stone at the marked point. Let's eliminate that stone (the black stone is already gone).
Now let's do the hypothetical replay.

Replay  

at a would be joseki, but is clearly bad. Taking the ko, or a, or even b would be better.

Replay 2  

Now, after the criticized move at , Black exchanges for (the "superfluous stones" that were removed). Isn't this a favourable exchange for White? In other words, how unfair have we been to White in criticizing her move, by removing two stones that perhaps weren't that equally superfluous. Or yet in other words, what are the opportunities Black lost with for in Replay 2, with respect to the opportunities White lost by playing in both "replay" diagrams ?

--Dieter

Well, both and are poor moves at this point, but does seem to be worse than . Tewari is not exact. You cannot always say that two plays are exactly equal, but often they are nearly so, and that is the case here. Under certain circumstances, depending on ko threats, offers a little extra protection. But if we consider the corner to be White's territory, the difference between and is negligible.

If were worth substantially more than , it would be because of the flaw at the marked point. But if White needed to protect that flaw, then would not just be bad, it would be terrible! ;-)

Without tewari, we might think that White's play was OK. It solidified the corner and left Black without a clearly live group. But tewari shows us that, even given that is somewhat doubtful and the pairing of the stones slightly favors Black, White has lost at least half a move.

-- Bill

I don't agree with your argument. In particular, "But if White needed to protect that flaw, then would not be just bad, it would be terrible! ;-)" - there's nothing bad about having to protect that flaw, given that it is going to be protected anyway. Andre Engels

Bill: Since the flaw is going to be protected anyway, the exchange of and is almost equal.

dnerra: Bill, I do not understand either how you can come to the conclusion that this exchange is almost equal. It makes the corner independently alive!

Possible continuation  

For example here white can cut at with almost no worries -- a move that would be much more difficult to play without the marked stone, as then black could kill the corner by peeping at the same spot (in case he has good enough ko threats, which I know to be an unrealistic assumption at the moment). So the marked stone is a gain, and you have to compare that with how much worse is compared to . That gets me really confused...

In summary, I agree that black is a little better off, but tewari doesn't help me much in arriving at that conclusion. And I think the loss is much less than half a move, in fact, I would happily pay it as komi if we were to have a game :)

Black sente  

After , White clearly owns the corner.

In addition, it is White's earlier protective plays in the corner that allow to threaten to cut off.

White kikasare  

Here is a sequence without the two paired stones.

is at best a middling yose. In addition, it fails to take the ko. At the start of the game, that is a loss of at least a half move. In addition, White seems to have been pushed around a little bit with the rest of the plays.

(Even if you play at a, which has better aji, the above remarks apply.)

Joseki according to Kogo  

After this Black can start ko or exchange a for b. Pretty close to the original diagram on this page, suggesting that move 7 shouldn't be the one to blame, but move 9.

Bill: comes in for criticism, given at a, not per se. When tewari permutes the order of play, just because a play may be bad given one order of play does not mean that it was bad in the actual order. In fact, the culprit is frequently a later play that does not work well with a previous play or plays.