I'm trying to take a closer look at a number of branches of joseki. with the aim of:

• Having fun trying to understand go better
• Improve my instinct in the opening
• Perhaps pick up some good habbits (shape, fighting, positional analysis)

I'm investigating and writing primarily for my own benefit, but I'd be thrilled if anyone could offer sugestions, corrections, thoughts, or even if people just read and take something from it.

First I intend to look closer at the position:

Click Here To Show Diagram Code

[go]$$B
$$ | . . . . . . . . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .
$$ | . . 1 . 2 . . . .
$$ | . . . . . . 3 . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .
$$ ------------------[/go]

And I intend to look at for example:

• Black 3
  • When do professionals play black 3
  • When do AI (probably ELF) play black 3
  • Meaning in relation to other pincers
  • Meaning in relation to other choices
• What happens next
  • What happens next in joseki books
  • what happens next in pro play
  • What happens next in AI play
  • What happens in special cases
  • What happens next in weak play
  • Can/should deviations be punished

So is popular with humans, but at least supperficially from my initial look it seems ELF thinks it should be less popular.

From a GoGoD search, it's the most popular pincer of the high approach and it's the second most popular move (behind the attach underneath.)
(the search was on a full 10x10 corner of the board, I've just cut the diagram down as the wider moves weren't so popular)

Click Here To Show Diagram Code

[go]$$B a - 56%,b - 14%,c - 11%
$$ | . . . . . . . . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .
$$ | . . 1 . 2 . . c .
$$ | . . . . a . b . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .
$$ ------------------[/go]

However - on looking at the games I found that I could understand (at least in a shallow 4k way) why the pincer was chosen, but ELF usually thinks it's still the wrong choice. On one occasion I found that ELF overlooked the pincer (presumably because it's so frequently a bad move) and when presented with the move it thought it was about 2.2% better than anything it was reading.

Click Here To Show Diagram Code

[go]$$B 44th tengen - Hane Naoki v Goto Shungo. Moves 1 to 7
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . 5 . . . . |
$$ | . . . 2 . . . . . , . . . . . , 1 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . 7 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . 6 . . . |
$$ | . . . 4 . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . 3 . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

This seems to me to be a pretty tame use of the pincer for example. I've had this exact opening in a couple of my games recently while I've been investigating the pincer , but in particular i found it in the 44th tengen - Hane Naoki v Goto Shungo. Black has a good reason to play 7 as a pincer as building infront of the shimari looks atractive (Choice of pincer is interesting to me though.)

ELF thinks this drops the black win % by about 2% though (not a truly game deciding swing.)

Click Here To Show Diagram Code

[go]$$B O Cheong-a v Gao Xing
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . X . . . |
$$ | . . . 1 . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . 2 . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . O . . |
$$ | . . . O . . . . . , . . . . X , . . . |
$$ | . . . . O X . . . . . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

This looks like a reasonable time to play 2. White might naturally want to lend a little support towards the group in the lower left and/or build something on the left. A wider pincer in this case still seems to leave problems on the left side and since another move is needed, the tighter pincer makes some sense to me.

ELF thinks that before this position is almost exactly even (white has slowly given away the slight initial advantage.) Looses white 7.6% in this case so ELF seems to consider it a real mistake compared to attaching underneath.

Looking at around 12 further example positions from pro games, ELF mainly saw the 1 space low pincer as losing around 2-3%, with one occasion where it chose the pincer as it's move, one occasion where it didn't consider the pincer but thought it was 2.2% better than it's favourite move, and a few more occasions where it considered it a fairly serious mistake (around -7%.)

I have mentioned Go Seigen's prescience regarding AI moves before, and several other people have pointed out their own examples of where he foreshadowed AI. So it's already a fair-sized body of evidence that he was on a different level. This may be another example.

Despite having it played against him many times, Go clearly disliked this pincer when he was in his prime, playing it just a couple of times. He played it a little more often in his dotage when he was reduced to fast games, and given what was going on at that time, with heavy research into other pincers, I speculate he may have then simply been trying to avoid opening traps.

In that regard, the first time Go played the pincer he startled the go world of the time (1939) with an opening trap of his own. It was considered a new move by him - wrongly, as it had appeared some 40 years earlier but his move depended on a ladder. At least we can infer that thought in terms of opening traps and that he had researched this opening thoroughly while delining to play it (he then abandoned the pincer again for a long time).

If we surmise that Go was roughly equivalent to AI level at his peak, when it seems that he may have been up to 2 stones ahead of the chasing pack, that may be a clue to how far AI is above the pro pack at the moment.

One characteristic of the commonest line in this joseki, which involves a lot of very close contact, is that it offers several ataris that amateurs love to play but pros avoid. However, even pros boob and one of the famous barbs of the famously barbarous Nozawa Chikucho was mocking a fellow pro for a mistake in this line: "Tamura Kahei doesn't know joseki." I gather the inference wasn't just that he overlooked something but that he played like an amateur. Eeeeew!!

Nice quote and background John

I believe when people say "Michael Kyle doesn't know joseki" it retains both of those meanings!

I stated as one of my aims that I'd like to maybe pick up good habits from learning some sequences, but you're correct - Joseki sometimes involve examples of poor technique that is thought to work well in these specific cases. I hope I can spot this as I go along.

Haven't got into all the reading and justification involved at all yet, but here's something that looks fun:

While I was asking ELF to evaluate the various board positions I noticed that in almost all situations, after the pincer, ELF thinks that the game should proceed like this:

Click Here To Show Diagram Code

[go]$$B ELF thinks this is very normal
$$ | . . . . . . . . .
$$ | . . . 5 . . . . .
$$ | . . . . . . . . .
$$ | . . 1 . 2 . . . .
$$ | . . 4 . . . 3 . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .
$$ ------------------[/go]

isn't something I've seen before, and sure enough it's not in the the GoGoD database once.
It seems to serve a similar purpose to when white makes a 2 space extension along the side in response to the 3-3 attachment, but maybe with a bit more prospect of sealing white in(?)

Click Here To Show Diagram Code

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

Ichiriki Ryo, 8p - Iyama Yuta, 9p, B+R (Komi 6.5)
42nd Japanese Kisei, title match #2, 2018-01-25

Very interesting that Elf has independent discovered this unusual move, I recall AlphaGo also likes it several times in the teaching tool. I've not seen pros trying it out yet.

Click Here To Show Diagram Code

[go]$$B ELF thinks this is very normal
$$ | . . . . . . . . .
$$ | . . . 5 . . . . .
$$ | . . a . . . . . .
$$ | . b 1 6 2 . . . .
$$ | . . 4 . . . 3 . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .
$$ ------------------[/go]

How does Black reply to ? "a" looks too thick, and "b" looks too thin.

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

Visualize whirled peas.

Everything with love. Stay safe.

The solid extend, I suppose that's ok being solid because black also needs to play the hanging connection instead of descent (to avoid painful peep). And although the marked stone is now a kosumi where it is sometimes a knight's move when played afterwards, that solidity does have plus sides in countering the thinness of the 2-space extension and creating a follow-up enclose at a. The 'joseki' seems to conclude with 3, black plays a 3-3 invasion (of course), and the pincer-cum-extension of 10 looks like a common continuation locally (play then continues elsewhere).

Click Here To Show Diagram Code

[go]$$W
$$ +---------------------------------------+
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . X O 1 . . . . . . . . 9 . . . . . |
$$ | . 2 . X . W . . 3 , . . . . . X . . . |
$$ | . . . X . . . . . . . . . . . . . . . |
$$ | . . . . a . . . . . . . . . . . . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . 0 , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . 5 O 7 . . . . , . . . . . X . . . |
$$ | . . 4 6 . 8 . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ +---------------------------------------+[/go]

For the original knight's move AG gives the following (black) win %:

Click Here To Show Diagram Code

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

a: best at 48.1
b: 49.1
c: 49.4
d: 50.0

About the common human joseki of b, one of the key reasons AG doesn't like that so much for white is black's pressing move at a in some fighting variations (AG thinks the peaceful human joseki has multiple mistakes from both sides), gennan made a nice video on that: https://www.youtube.com/watch?v=5ouxL3G_tIo. So playing high to prevent that has some logic.

Edit: As well as the butting move of Bill, AG also suggests kosumi out for black in response to knight's move, leading to this 2nd line crawling for a big corner that I wouldn't feel comfortable doing, particularly given the bad ko aji at a.

Click Here To Show Diagram Code

[go]$$B Another choice
$$ +---------------------------------------+
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . 9 8 6 0 . . . . . . . . . . . . . . |
$$ | . 7 X O 5 . . . . . . . . . . . . . . |
$$ | . . 2 3 . O . . . , . . . . . X . . . |
$$ | . . 4 X . . . . . . . . . . . . . . . |
$$ | . . . . 1 . . . . . . . . . . . . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ +---------------------------------------+[/go]

Click Here To Show Diagram Code

[go]$$Bm11 contd.
$$ +---------------------------------------+
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . X O O O 2 4 6 . . . . . . . . . . . |
$$ | . X X O X 1 3 5 7 . . . . . . . . . . |
$$ | . . O X . O . . . , . . . . . X . . . |
$$ | . . O X . . . . . . . . . . . . . . . |
$$ | . . a . X . . . . . . . . . . . . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |[/go]

There is an old human joseki like this with the kosumi out first in response to the pincer, and then the knight's move and 3-3 attach: http://josekipedia.com/#path:qdodmcnepf ... rdqbreqfrb. With that the corner player tries to finish off the corner without so much 2nd line crawl. If you are going to crawl so much, is 8-9 really a good exchange? I suppose this is another example of AG valuing sente highly and 2nd line crawling as not so bad.

Great sequences. Thanks very much for sharing them.

I had considered using the teaching tool as one of my resources, but Elf and lizzie help when I want to ask a stupid variation question, or when I want to try out sequences in a specific board situation. (I just need to keep watch that elf isn't missing an important ladder somewhere!)

From memory (I'm away from my computer) elf does like the solid 'a' connection too, but I'm not sure it extended across the top like master very often - maybe that was to do with the whole board in each case.

alternative ELF play:

fascinating stuff.

As a brief asside - I often like to play openings and/or patterns that I feel are going to force my opponent to play something i'm interested in exploring. (winning is secondary)

Eg. in order to look at these patterns I've been trying out:

Click Here To Show Diagram Code

[go]$$B In order to investigate the pincer
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . 5 . . . . |
$$ | . . . 2 . . . . . , . . . . . , 1 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . 7 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . 6 . . . |
$$ | . . . 4 . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . 3 . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

(incidentally, the sdk players i've played recently seem to be terrified to try the low approach so I can be fairly confident of at the moment. I guess I won't gave a good inducement when i want to study low approach pincers)

It occured to me that if I want to investigate these AI sequences without the risk of my opponent playing the human alternatives at (kosumi, descent to the 2nd line, attach to the pincer stone all seem conceivable in kyu world) I could try:

Click Here To Show Diagram Code

[go]$$B Forcing the 3-3 attachment
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . 5 . . . . |
$$ | . . . 2 . . . . . , . . . . . , 1 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . 9 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . 6 . . . |
$$ | . . . 4 . . . . . , . . . 7 . , . . . |
$$ | . . . . . . . . . . . . . . . 3 8 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

I'm assuming that would be a safe bet as white would probably expect:

Click Here To Show Diagram Code

[go]$$W White would love this?
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . X . . . . |
$$ | . . . O . . . . . , . . . . . , X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . 7 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . 1 . . . |
$$ | . . . O . . . . . , . . . 2 . , 5 . . |
$$ | . . . . . . . . . . . . . . . X 3 . . |
$$ | . . . . . . . . . . . . . . 6 . 4 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

I've been told (and please correct me if you don't agree) that this is old/ special case joseki, and is usually considered good for white, since white get's a pretty solid group and black didn't get enough compensation. I would also particularly love this for white in this game since white takes that solid position in front of the shimari and blacks position on the lower side isn't doing anything special.

Although since I'm playing the pincer, dispite the fact that AI tells me that it should end up slightly worse than the attach underneath, I suppose you could say I'm playing a trick move.

This is kind of incredible to me. Looking at ELF in lizzie, after whites kosumi this is a one way street.

However I couldn't seem to get ELF to pick the kosumi in the first place as you have. My Elf did fewer than 1k read outs on the kosumi while doing over 60k read outs on the 4-4 bump. Gomoto: I'm using..



..That network. Are you using a different one?

The variations in josekipedia look more sensible to me but ELF seems to think that each one is a non-trivial mistake (i think greater than 3% for each one)

I've only just spotted the really interesting little difference here.

Elf has strong feelings about about in this sequence

Click Here To Show Diagram Code

[go]$$B Master and Elf agree
$$ +---------------------------------------+
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . 9 8 6 0 . . . . . . . . . . . . . . |
$$ | . 7 X O 5 . . . . . . . . . . . . . . |
$$ | . . 2 3 . O . . . , . . . . . X . . . |
$$ | . . 4 X . . . . . . . . . . . . . . . |
$$ | . . . . 1 . . . . . . . . . . . . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ +---------------------------------------+[/go]

Elf agrees with Master the whole way in this board position. To play at 8 would be a fairly large mistake (-6.8%)
I suppose that you're loosing the ko aji, and maybe the extra line white has to push ends up a little close to the corner stone.

However in the board position we've been analysing in Lizzie (Ichikiri Ryo v Iyama in the Kisei final) Elf's choice between these two is dramatically reversed.

Click Here To Show Diagram Code

[go]$$B Whole board thinking in joseki..
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . . . . . . . . . . . . . O . . . . |
$$ | . . . . . . . . . . . . . . . O X . . |
$$ | . . . O . . . . . , . . . X . O X . . |
$$ | . . . . . X . . . . . . . . O X O b . |
$$ | . . . . . . . . . . . . . . . X a . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

In this case Elf likes a, leading to white maintaining the narrow advantage from the start of the game. If white plays b, perhaps because he or she has seen it looking like joseki in the teaching tool, Elf now considers this a pretty much game loosing mistake at 29.9%. I can post-hoc rationalise this a little.. but the swing is huge!

Go is hard.

Elf thinks that Black should begin sealing out his or her victory like this..

Click Here To Show Diagram Code

[go]$$B Punishing the blunder?
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . 2 3 . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . . . . . . . . . . . . . O b . . . |
$$ | . . . . . . . . . . . . 0 . . O X . . |
$$ | . . . O . . . . . , . . . X . O X a . |
$$ | . . . . . X . . . . . . . . O X O O . |
$$ | . . . . . . . . . . . . . 1 4 X 6 . . |
$$ | . . . . . . . . . . . . . 9 5 7 8 . . |
$$ ---------------------------------------[/go]

Followed by a or b presumably to remove that ko aji.

@MikeKyle: same network, yes depends on whole board position

My mistake.

Click Here To Show Diagram Code

[go]$$W Punishing the blunder?
$$ . . , . . . . . , . . . |
$$ . . . . . . . . . . . . |
$$ . . . . . . . . . . . . |
$$ . . . . . . . . . X . . |
$$ . . . . . . . O . . . . |
$$ . . . . . . . . O X . . |
$$ . . , . . . X . O X . . |
$$ . . . . . . . O X O 1 . |
$$ . . . . . . 2 . X . . . |
$$ . . . . . . . . . . . . |
$$ -----------------------[/go]

Elf seems to think that this 2 is a brilliant (and difficult to spot) move that punishes 1 in all cases.

Seems reasonable - black gets the corner without all that pushing and the white wall (or if white takes the corner, it's small and the centre stones start looking weak)

The only reason Elf agreed with master in the teaching tool position is because Elf failed to spot the move (2 above) on that occasion. Show Elf the move and you get the same huge swing.

Does this mean that Master missed the move? Or does Master have some incredible sequence to refute the refutation?

That move is known to humans (but not me!), and josekipedia (citing some joseki dictionary) says it's good for white to then live in the corner: http://josekipedia.com/#path:qdodmcnepf ... rcsdsbsfra. Is that judgement usually correct or usually wrong according to super-strong bots? Or maybe this whole board situation changes things...

"Show Elf the move and you get the same huge swing."

This observation underscores the fact that all players are fallible, superhuman bots included. We need to maintain a healthy skepticism about their evaluations, even while recognizing that they are better than ours. Errors in evaluations are greatest, as a rule, in the opening.

From a scientific point of view, one problem with current bot evaluations is that they do not come with an error estimate. We would like to say that if play A is evaluated as 2% better than play B, that play B is a mistake, but we do not have enough evidence to say that. We are guessing.

Currently the focus in go bot research seems to be on producing the strongest bots, not on producing the most accurate evaluations, even though the two are related. As a human, I would find the latter goal more useful. Even so, I would like evaluations to come with error estimates.

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

Visualize whirled peas.

Everything with love. Stay safe.

Move a is the best if you calculate this deep.

Move b is the best if you calculate this deep.

Calculation depth is approaching infinity.

-> There is no absolute best move.



(If you have difficulties to imagine the infinite calculation depth, consider ko and an increasing number of unstable stone groups on the board all related to the status of the whole board and the game.)

And for all practical purposes, which of the good moves is accessible to you? It may be a very strong move, but can you handle it?

I love go, finest game discovered yet.