Merry Christmas everyone,
it’s been over 4 months since I posted my review of Endgame 3. I want to apologize to Robert for being so long. I am somewhat surprised nobody else picked up his offer to receive and review one of his endgame books. That or people have been even slower than me .
I am Arnaud Boucherie. As I am typing this review I am ranked 1 or 2 dan EGF but 2019 has not been an active year. I received a physical copy of the book in exchange for the review.

What’s in the book
As the title says, it is a problem book. The book contains 150 problems, 130 of which are about getting the correct evaluation of a local endgame (sente/gote, count and move value). The first 20 problems are about getting the correct endgame on a 11x11 board. They are about spotting tesuji and reading life and death correctly and not much about calculation.
Even if it is a problem book, there are many pages about the theory. It is obviously much shorter than what you get in Engame 2 and 3 but it is enough to get you going if you are a good learner, and a great reminder if you have already read Endgame 2 and 3. The topics dealt are :
- counting gote and sente endgames
- distinguishing gote and sente positions
- kos
- theory of long sequences (should you play the sequence in one go or should play be interrupted)
Each problem comes with a detailled solution.

How did it go ?
As I mentionned in my previous review, I actually started the book in June before reading Endgame 3. I made some mistakes in the first 20 problems from reading too fast, but most of them went fine. In my opinion all of them are doable for players at least 5k. The main lesson is that at some point you need to grab your sente moves before it’s too late.
I then did the first 25 calculation problems. I got « only » 4 of them wrong but I was very slow at the time.
At that point I switched to Endgame 3, learned some more theory, got a bit better at calculation from doing the examples and exercises. I was faster during my second trial.
Here are the results, problems are seperated the way they are in the book :
Problems 21~59 (basic problems) : I cannot find my results (edit:I found them! I made 6 mistakes) but I remember being much faster the second time
Problems 60~67 (not as basic) : I made 1 mistake
Problems 68~79 (ko) : I made 3 mistakes
Problems 80~99 (mixed problems) : I made 1 mistake
Problems 100~118 (some more mixed problems) : I made 7 mistakes
Problems 119~150 (problems for long sequences) : I made 7 mistakes. Those are also about indentifying long sequences worth playing, which I have not done (yet). I focused on getting the correct counts and move values.
My mistakes were of various types. Sometimes I didn’t count points correctly. Sometimes I overlooked a move was sente. Sometimes I overlooked a move tactically. Sometimes I didn’t understand the problem correctly, usually when it was in the center.
Edit: I forgot to mention that usually mistakes came in groups, which suggests that there has been days where I was sharp and days where I was off.

What did I learn ?
Obviously I didn’t learn anything about theory, as the other books already covered it. Still, reading about long sequences the second time was easier. I lack the courage and confidence to do the related problems, but the examples were completly clear.
Practice makes perfect. My speed has increased a lot over the last 6 months. Calculations to check if a move is sente or gote became second nature. In that regard, the book helped a lot.
I still cannot do the problems mentally, unless the position is very easy. I tried when I picked up the book the second time, but I soon gave up. I use pen and paper to keep track of values. I wonder how well I could do without it in a real game.


Would I recommend it ?
Yes ! I have not read Anti Tormanen’s Rational Endgame so I can’t compare. If you already know a bit about miai counting, Endgame problems 1 is a very good stand alone book. A year ago, in my review of Endgame 2, I complained about the first 50 pages being hard to follow, in my opinion. This time there are not many theory pages, but they are very clear and to the point. It is hard to judge if it is enough for most new readers, as I have already read about the theory before. If you started with Endgame problems 1 I would like to read your opinion. Anyways, if you are considering getting only one of Endgame 2/Endgame 3/Endgame problems 1 this is the one I would recommend.
Lastly, I was very impressed by the proof reading of the book. Whenever I disagreed with the solution, I always identified the mistake on my side, and it was never an issue because there is a lot of detail.


Since I learned go in 2004, 2019 is by far the year I played the least. Hopefully that will change in 2020. I will update my post next year to tell how it goes in real games and do a longer review of problems about long sequences.
If you have any question about Endgame 2 & 3 or Endgame problems 1, don’t be afraid to ask. I will happily answer.

As I understand your reviews of Endgame 2 / Values, Endgame 3 / Accurate Local Evaluation and Endgame Problems 1, you have read the last chapter each of Endgame 3 and Endgame Problems 1 only roughly. Otherwise, you seem to have checked everything carefully with paper and pen without finding any evaluation mistake in these books. You attribute this to my proofreading but I think that first working out and understanding well the theory has also been necessary.

Meanwhile, you have learned about modern endgame theory. Have you developed an opinion how it compares to traditional endgame theory?

Most books about endgame values only or mostly calculate the move value in the initial position. The three books (but also Bill Spight in his explanations) put equal importance to the count of the initial position, and the move values and counts of follow-up positions. Have you learnt what uses considering more than one value might have?

You have accelerated your value calculations for easy problems. In my experience, further practice enables acceleration also for intermediate to difficult problems. Like practising the equally important tactical reading, endgame calculations are a life-long project of improving speed and correctness.

Using paper and pen are tools for homework at school or study at university:) However, I think that you must practise solving problems while only using your brain. This should further improve your speed, correctness and confidence. Wouldn't you agree?

Tricky question! The biggest difference for me so far is that mordern theory shows me a clear path to get a move value. Traditional endgame I feel was usually about showing easy examples but often disregarding calculations for complex situations.


I don't know if it showed in my reviews but I also paid more attention to move values than to counts. I don't know where you want to lead me with your question so the answer is probably no. While their calculations are connected, to me counts and move values don't serve the same purpose. Counts are useful to know who's ahead and by how much. Move values are useful to decide where to play next.


I don't think using pen and paper was a wrong decision for Endgame 2 and 3, as it helped focusing on the theory. Now that the review is done, I agree that there is no point in still doing problems with pen and paper, as I won't be able to do it during a real game. I'll need to find some ways to keep track of numbers, as this is the sole reason I use paper.

Thanks both, interesting review and discussion! As usual I'm buying books faster than I can read them, so I'll wait a little before I buy this one, but hoping to get to it by the end of 2020.

I've started this thread for talking about mental calculations.

Well said.



Counts serve these (and other) purposes:
- local positional evaluation
- counts are derived from the counts of the followers and their followers
- move values are derived from the counts of the followers and their followers
- gains are derived from counts
- the values of each follower can be studied as if it were an initial position; thereby we get free extra problems for practising calculations
- counts occur in conditions for identifying and verifying correct calculations
- counts occur in conditions for choosing correct strategy

Therefore, counts are as important as, if not more important than, move values and gains.

In my books, I try to convey this message by studying counts as much as move values. My question has implied to find out whether you have become aware of this importance of counts. They are not just a tool to be forgotten as soon as the move value is calculated, especially since we should always also know some positional evaluation and embracing counts as equally important means helps accelerating, and improving correctness of, our endgame value calculations.

I think there's more to the story here (and wondered if Endgame 2 - Values might explain this, but it didn't; perhaps this is addressed in a later volume). Traditional theory, if explained well, can still give you a clear path to the move value. The problem is that you can't add numbers in an intuitive way -- Bill has given examples elsewhere on this forum to show how traditional endgame theory plus a bit of arithmetic adds up to nonsense. But in modern endgame theory, if you have a count of 12 and then you play a move with a gain of 5, you can be sure that the count has changed to 17. (Remember though that gains are not quite the same as move values! Sente gains nothing, but it still has a value.)

I suspect that modern endgame theory is also better for evaluating kos (although I don't know enough of the traditional theory to be confident of that).

I'm aware of those uses


I noticed my papers were mostly full of counts I was already aware of the importance of counting, though I've been mostly using ballpak numbers than accurate numbers. I don't think the lack of accuracy hurts my game, as mistakes are still quite big. Knowing I lead by 3~4 points rather than 0~5 points would not impact my play at all at my current level. I agree that embracing counts before the endgame will help with speed and correctness. Now that I have the tools, I can use them if the time settings allow it.
On the other hand, if before I thought two endgames moves had similar value (which was maybe even not the case), I clearly see the impact accurate values would have on my games.

Because there are not many books about the endgame in western languages it was hard to answer the question without thinking about the books I have read.
If I go back to the examples I posted in my review of Endgame 3:

Click Here To Show Diagram Code

[go]$$Bc Example 1
$$ +-------------
$$ | X X X X O .
$$ | . O O O O .
$$ | X X . O . .
$$ | . O O O O O
$$ | X X X X X O
$$ | . . . . X .[/go]

Click Here To Show Diagram Code

[go]$$Bc Example 2
$$ +---------------
$$ | O O O O . X .
$$ | O O O O . X .
$$ | . X X X X X .
$$ | O O O , X . .
$$ | O O O . X X .
$$ | . X X X X . .
$$ | O O O O X X .
$$ | . O . O O . .[/go]

Neither situation is complicated. Say a ko arives, to kill a surrounded group, so settling the ko would not impact the outside ("simple" situation again). In the two traditionnal books I have (The endgame by Jame Davies and Yose by Motoki Noguchi), unless my memory is faulty, I don't see the tool to know how big the ko is, nor can I see how big of a ko threat it is to play in either examples. My reading of Robert's books answers both questions clearly and even now I would confident to do the calculations without paper while playong online with a 30 second byo yomi. But I can't tell if reading Antti's book would answer these questions.

It answers half of them :-) There's a very clear chapter on working out the count and move value for different types of ko. But he doesn't go as far as comparing the move values in the ko position with ko threats elsewhere. More discussion of Antti's book here.

explo, 1~2d with quite big mistakes (that is, larger than 4 points) is a hint to improve by stop making them...! :)

xela, traditional endgame theory explained well? Endgame 2 provides an explanation, which essentially says: do it as for modern endgame theory, but gote move values are twice as large. While this gives a method for all ordinary problems (but not: ko, ambiguous, double sente etc.), I made it up - it is not what traditionalists would explain. Furthermore, the theory of distinguishing types, correct moments of playing elsewhere and strategy would be borrowed (and maybe transformed) from modern endgame theory but more as an excuse that it could be done for traditional endgame theory rather that what had been done before the theory appeared in modern endgame theory.

From them, I have only seen hints to partial explanations for specific kinds of ordinary problems:
- gote with gote follow-ups,
- gote with one immediate sente follow-up,
- sente with one immediate gote follow-up.
They would expect us to understand on our own how to apply this to more complicated gote / sente follow-up combinations. IIRC, I have never seen them explaining sente with sente follow-ups. Usually, ko was a mess. Initial (and mostly intermediate) counts were beyond their knowledge.

Modern endgame theory is much better for kos. Even for ordinary kos, traditional endgame theory had trouble deciding whether to compare 1/3 or 2/3 or 1/2 of a value of a ko move or ko threat.

If I wasn't making big mistakes I wouldn't be a low dan player
Bill Spight said he expects a dan player to play the small endgame almost perfectly. From my experience of watching 1~3 dan players in tournaments, I notice a lot of mistakes at that stage, as a spectator, in close games. When I was more active, around 2010, my endgame was quite good compared to my peers, and it was more a matter of avoiding dumb mistakes than a matter of superior understanding of the endgame.

IMX, different dan players have different endgame skills. However, consistently bad endgame is unheard of for 5d (EGF) or stronger. This is so because they are aware that ten 1 point mistakes during the endgame phase in games with long thinking time would be too expensive.

Happy New Year!

Just a couple of comments on the discussion so far.



That's a little strong. I think that a dan player has the skills to play endgames at or below temperature 1 almost perfectly. All they need to know is a few simple heuristics.



Yes. Counts are basic for the evaluation of both territory and move values.



Traditional theory has only part of the theory of ko evaluation. It is rudimentary. Berlekamp developed the theory of evaluating hyperactive kos and positions, where the count (and thus move values) depends upon who is komaster. I developed the theory of evaluating multiple kos, where neither player may be komaster. Also, Bill Fraser and I worked with Berlekamp to develop the theory of kos where neither player is komaster.

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

Visualize whirled peas.

Everything with love. Stay safe.

Is it stronger than what you said here

Well, I don't always expect people to do what they should do.

I mean, really, it doesn't take much effort for a skilled player, does it?

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

Visualize whirled peas.

Everything with love. Stay safe.