Latest haul

[Jujube]

Apart from ordering the excellent Genjo Chitoku a few weeks ago, a package dropped through the letterbox earlier, containing Direction of Play, and Positional Judgement High Speed Game Analysis, which alongside Rational Endgame, complete the latest Go binge.

I'm around 5k EGF. What's your impression of Direction of Play and Positional Judgement High Speed Game Analysis, what were your thoughts, are they still relevant in the AI era, what do you think of them for my level? Which one should I read first?

That brings my book collection to around 45. None of the books come close to the progress I made from Go lessons I had last year. But y'know, they're books. Who doesn't love a good book!

[Calvin Clark]

What's your impression of Direction of Play and Positional Judgement High Speed Game Evaluation, what were your thoughts, are they still relevant in the AI era, what do you think of them for my level? Which one should I read first?

I think those are good questions. As for level, you're fine. That's not the issue. The issue, as you point out, is would you be better off with more modern treatments?

I have both books, but I haven't looked at them in years and certainly not since AlphaGo. My suspicion that Cho Chikun will hold up better as the idea is to teach a way of estimation that's fairly practical. Just flipping through the Direction of Play, though, it's chock full of old patterns, so I'm less convinced of its utility. Most of my ideas about direction are influenced by Yilun Yang than anything an that book, anyway.

It would be interesting to see how these texts hold up against modern AI. I may try checking a few diagrams just to see if there any howlers in there.

[gowan]

It has often been said that it is good to understand the moves you play. From that point of view the books you bought will be good for you. Trying to use the AI moves would be difficult, I think.

[Jujube]

What's your impression of Direction of Play and Positional Judgement High Speed Game Evaluation, what were your thoughts, are they still relevant in the AI era, what do you think of them for my level? Which one should I read first?

I think those are good questions. As for level, you're fine. That's not the issue. The issue, as you point out, is would you be better off with more modern treatments?

I have both books, but I haven't looked at them in years and certainly not since AlphaGo. My suspicion that Cho Chikun will hold up better as the idea is to teach a way of estimation that's fairly practical. Just flipping through the Direction of Play, though, it's chock full of old patterns, so I'm less convinced of its utility. Most of my ideas about direction are influenced by Yilun Yang than anything an that book, anyway.

It would be interesting to see how these texts hold up against modern AI. I may try checking a few diagrams just to see if there any howlers in there.

Running a few diagrams through Leela Zero gives some interesting counter arguments, e.g. p.23 "Why has White ended up at such a grave disadvantage?" - Leela Zero gives 55% to White. Or, p.27, there are moves in that joseki sequence that Leela Zero doesn't like, one move is -12%. I like the book, Kajiwara certainly has an entertaining writing style, like he's in the groove, showing variations hither and thither with colourful, slightly comical prose. It's like his treaty, or how he himself understands the game, and he's going to give you the message without holding back!

[Kirby]

analyzing joseki with ai is a little weird, because ai typically evaluates based on the whole board. the placement of other stones on the board could significantly impact the ai's judgment.

[Jujube]

analyzing joseki with ai is a little weird, because ai typically evaluates based on the whole board. the placement of other stones on the board could significantly impact the ai's judgment.

I'm not saying it was an isolated joseki, I'm saying it was a joseki. There were other stones on the board.

[RobertJasiek]

I'm around 5k EGF. What's your impression of Direction of Play and Positional Judgement High Speed Game Analysis, what were your thoughts, are they still relevant in the AI era, what do you think of them for my level?

Direction of Play:

It gives a useful idea, but go theory is richer than the book might make you believe: there is more than one concept of direction of play: in which part of the board to play next (IIRC, this is the kind explained in this book); in which directions to develop stones and prevent development of opposing stones; how stones at different places work together (haengma). Then, there is the fighting Korean's view that tactical reading dominates "old-fashioned" directions. Furthermore, one can evaluate moves and thereby determine important places. So learn from the book but open your mind very much more broadly than the book reveals, i.e., also look beyond the book.

Positional Judgement High Speed Game Analysis (Cho Chikun):

It is useful but only scratches the topic. Remove the advertisement tag High Speed Game Analysis from the title for the same reason. If you want a much more detailed discussion of territorial positional judgement (the book's topic), read Positional Judgement 1 - Territory. (Afterwards, you might consider volume 2 to also evaluate influence, options, aji and fights.) Furthermore, there is the Japanese tewari kind of positional judgement and absence / balance of positional inconsistencies (such as "one inefficient stone here is balanced by one such opposing stone there"), which is not discussed in the aforementioned books but which you might learn about from other sources. Positional judgement is one of the central go skills so it is good that you start studying it.

AI programs have their own kind of positional judgement: estimated winning probabilities. Except for wild approximative guesses, such is impossible for human players. We do need positional judgement in terms of strategic concepts. Our judgement may not be useful for AI programs, but I have found it very useful also for judging about AI-created positions.

Level:

5k is good for reading all mentioned books, but for PJ2 this is the lower rank.

[Bill Spight]

I'm around 5k EGF. What's your impression of Direction of Play and Positional Judgement High Speed Game Analysis, what were your thoughts, are they still relevant in the AI era, what do you think of them for my level? Which one should I read first?

Both are dated. You can learn from both, but if you can check out their suggestions with Leela, that would be good.

I loved Direction of Play when I read it, but I was aware that Kajiwara was a brilliant eccentric. He is good for stimulating thought, but you can't take him too seriously. Especially now, when even standard ideas about direction of play have been brought into question by the top bots.

As for Positional Judgement, I threw it in the trash can almost immediately. (Well, not exactly the trash can, it's still on a shelf somewhere.) The idea of counting almost sure territory is a good one, but the real question for positional judgement is how to evaluate influence. Again, the bots are much better than humans in terms of positional judgement. Checking Cho's assessment of the positions in the book against a top bot's assessment could be quite valuable.

Edit: One false impression that I got from the idea of counting "secure" territory is that it is necessary to respond to all those apparent sente moves. (This was from other sources, not that book.) As the bots know, tenuki is often right.

As for which to read first, Kajiwara is entertaining.

[RobertJasiek]

Bill, judgement play is not meant to repeat regular play but is a model for the evaluation purpose. Local play would be interrupted in real play but sente only does not explain gote, of course. Nevertheless, we use judgement play because it is reasonably accurate and very much faster than absolutely accurate endgame play, which distinguishes gote etc.

[Bill Spight]

Bill, judgement play is not meant to repeat regular play but is a model for the evaluation purpose. Local play would be interrupted in real play but sente only does not explain gote, of course.

True, but when I was 4 kyu I got the implicit message that the "sure" territory belonged to the player who defended it. When, OC, that was not the case, nor were those defensive plays correct.

By the time I got the book, I was aware that the keima enclosure was worth around 25 pts. But the books evaluated it as 11 pts. Tilt!

[RobertJasiek]

11 points of current territory plus the value of influence.

[Bill Spight]

11 points of current territory plus the value of influence.

Of course. But how do you figure the value of influence if you didn't know it already?

[RobertJasiek]

The simplest is to count the influence stone difference (B minus W influence stones).

[zermelo]

My recollection of the 'Direction of Play' is that it contained a lot of very subtle, high level examples and ideas that are not very meaningful for amateur play. And probably not professional play either, when bots have shown that many subtle high level pro ideas are just wrong.