Efficiency

Knotwilg · #1

Click Here To Show Diagram Code: [go]$$ Position at move 50 $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . X . . | $$ | . . X , . X . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . 2 . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . , . . . . . , X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . O . . . . . . . . . O . O O . | $$ | . O X . . . . . . . . . . . . . O X . | $$ | . O . O . X . . O X . . . . X X O X . | $$ | . . O , X . X . X O . O . O . O X . . | $$ | . O X X . 1 O O X O . X . O . O X . . | $$ | . . . . . . . . X O . . . . O X X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

In Ian Butler's "Murdered by a 2d" KataGo makes a big deal of this position. Black has played the :b1:

to capture the 2 stones. White left to attack the right side.

Click Here To Show Diagram Code: [go]$$ Better play by Black $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . X . . | $$ | . . X , . X . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . , . . . . . , X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . O . . . . . . . . . O . O O . | $$ | . O X . . . . . 1 . . . . . . . O X . | $$ | . O . O . X . 2 O X . . . . X X O X . | $$ | . . O , X . X 3 X O . O . O . O X . . | $$ | . O X X . . O O X O . X . O . O X . . | $$ | . . . . . . . . X O . . . . O X X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Black should have played this way to take a firm grip on White's stones and make it difficult for him to take profit in the center.

Click Here To Show Diagram Code: [go]$$W White's opportunity $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . X . . | $$ | . . X , . X . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . , . . . . . , X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . O . . . . . . . . . O . O O . | $$ | . O X . . . . 5 3 . . . . . . . O X . | $$ | . O . O . X . 2 O X . . . . X X O X . | $$ | . . O , X . X 1 X O . O . O . O X . . | $$ | . O X X . B O O X O . X . O . O X . . | $$ | . . . . . . 4 6 X O . . . . O X X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

White can force this way. Black uses 3 stones to capture 3 stones, while White is strong on the outside.

This is a fairly easy sequence to see. However, what is more insightful is that KG prefers this by far to playing elsewhere. The whole sequence is almost forced.

Click Here To Show Diagram Code: [go]$$W Black's alternative $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . X . . | $$ | . . X , . X . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . 6 . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . , . . . . . , X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . O . . . . . . . . . O . O O . | $$ | . O X . . . . 5 3 . . . . . . . O X . | $$ | . O . O . X 7 2 O X . . . . X X O X . | $$ | . . O , X 9 X 1 X O . O . O . O X . . | $$ | . O X X . B O O X O . X . O . O X . . | $$ | . . . . . . 4 8 X O . . . . O X X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Black can play away at this point. KataGo evaluates this approximately even but less "natural", taking the previous diagram as the main one to explore.

kvasir · #2

Click Here To Show Diagram Code: [go]$$ Better play by Black $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . X . . | $$ | . . X , . X . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . , . . . . . , X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . O . . . . . . . . . O . O O . | $$ | . O X . . . . . 1 . . . . . . . O X . | $$ | . O . O . X . 2 O X . . . . X X O X . | $$ | . . O , X . X 3 X O . O . O . O X . . | $$ | . O X X . . O O X O . X . O . O X . . | $$ | . . . . . . . . X O . . . . O X X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

is also remarkable for the impact it has on the game. Black gets to defend the shape and also starts looking at capturing the white cutting stones as well as moving out the black cutting stones. I'd argue it is not efficiency that matters here but rather effectivity (or should one say effectiveness or impact?).

In my experience efficiency matters a lot more for points and in endgame-ish games than it matters for fights. Effectivity matters the most in fights and can take precedence over efficiency in many situations but the most effective move is often not the best move because in the end it is a game decided by points.

Really good moves are both efficient and effective, like :b1:

, but I think it is interesting to make a distinction.

Btw my b40 KataGo slightly prefers to play this way. I think this is because the power of :b1:

is is really that it handles the cutting stone and white now needs to mind his own weakness. It is a complete reversal from the game move that gave white opportunity to become strong and begin to capture the cutting stones on a large scale which is also efficient.

Click Here To Show Diagram Code: [go]$$ Better play by Black $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . X . . | $$ | . . X , . X . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . , . . . . . , X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . O . . . . . . . . . O . O O . | $$ | . O X . . . . . 1 . . . . . . . O X . | $$ | . O . O . X . 3 O X 2 . . . X X O X . | $$ | . . O , X . X . X O . O . O . O X . . | $$ | . O X X . . O O X O . X . O . O X . . | $$ | . . . . . . . . X O . . . . O X X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

I think

really stands out for being effective, it would be a turning point in the game. The efficiency takes secondary role in this position. That is my opinion or observation.

Knotwilg · #3

Thanks. To dwell on terminology: with effective you mean "purposeful" (regardless of the way the purpose is achieved) while efficient means "in a good way" (regardless of the purpose)?

That useful distinction is a meaningful comment on the intention of my post. I was indeed suggesting that the reduced efficiency of Black's stones was enough to overcome the possibly higher purpose of a tenuki. You refute that argument by saying it's really the purpose of the moves at the bottom that make them preferable over what happens on the right side.

Still, it matters if you can achieve the same purpose with few moves or more moves. In between Black's atari and the game move is the simple solid connection, which would reduce the temperature to zero.

I find it interesting that this remains a hot place, even if Black has played a move to capture the stones.

(Just reflecting here, need to think more). Appreciate your thoughts and diagrams.

jlt · #4

I'd say that move A is more efficient than move B if both moves achieve the same purpose(s), but move A allows to make more points (for instance because it leaves fewer weaknesses, or exposes a weakness of the opponent, or expands further, etc.). Whereas in your example, move A has a double purpose and move B a single purpose.

kvasir · #5

I think what I mean is close to the dictionary definitions in Merriam-Webster:

Effective: "producing a result that is wanted"

Efficient: "capable of producing desired results without wasting materials, time, or energy

In the game position there is a single move that stands out as most efficient and most effective, but I think in many positions it is not clear.

I tried to come up with a situation when black would play differently in the game position, but I realized that this won't work because black is practically capturing cutting stones. I say "practically" because white can escape but is not going to do that. If I change the position it is likely that black will prefer to capture the cutting stone or just not play locally at all.

Capturing the cutting stone changes the balance of power in the center even if white replies locally. White defends locally because the follow-ups are painful. Efficiency is also important, I just wanted to point out that this might be secondary in this case that, I think, is more about the huge impact (or effect) that this play has.

When black connects, just for comparison with the diagram (in previous post) when black captures.

Click Here To Show Diagram Code: [go]$$ -3 pts error $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . X . . | $$ | . . X , . X . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . 3 . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . , . . . . . , X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . O . . . . . . . . . O . O O . | $$ | . O X . . . . . . . . . . . . . O X . | $$ | . O . O . X . . O X 2 . . . X X O X . | $$ | . . O , X . X 1 X O . O . O . O X . . | $$ | . O X X . . O O X O . X . O . O X . . | $$ | . . . . . . . . X O . . . . O X X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Knotwilg · #6

jlt gives a good angle to my skewed analysis: Black's correct move is double purposed; when he captures the wrong way, White can still squeeze out some purpose out of the captured stones

Click Here To Show Diagram Code: [go]$$ $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . , . . . . . , X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . O . . . . . . . . . O . O O . | $$ | . O X . . . . . B B . . . . . . O X . | $$ | . O . O . X . W O X W . . . X X O X . | $$ | . . O , X . X B X O . O . O . O X . . | $$ | . O X X . . O O X O . X . O . O X . . | $$ | . . . . . . . . X O . . . . O X X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Click Here To Show Diagram Code: [go]$$ $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . , . . . . . , X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . O . . . . . . . . . O . O O . | $$ | . O X . . . . W W . . . . . . . O X . | $$ | . O . O . X . B O X . . . . X X O X . | $$ | . . O , X . X W X O . O . O . O X . . | $$ | . O X X . B O O X O . X . O . O X . . | $$ | . . . . . . B B X O . . . . O X X . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

If you look at the results side by side, the difference in the usefulness of the stones becomes really apparent.

kvasir · #7

I think one needs to use three things to judge the efficiency:

What was achieved, as in the impact or effect.
The difference in the the number of of moves.
The total number of moves used for each player.

I think the first item may be overlooked sometimes because it is too obvious, but moves do not all have the same value (if single moves really have a value per-se) and some judgment is needed on what happened overall. The first item is probably the most important because a player that can judge this accurately is likely to also be able to tally up the judgment of different groups and make an accurate whole board judgment.

The second item is what gets all the talk, I think, and it is hard to understand or explain efficient play without noting the difference in number of moves.

The third item often seems to be overlooked like the first one is. I think that is for a different reason because this seems less obvious than the first item. You'd use this to estimate if you are developing quickly enough. For example, take a position (maybe like the game position) with half of the board having many stones of both color but the other half is mostly just a few stone of one color. What we want to know is if either color is behind in development. The importance of this point as opposite to using the difference in number of moves is that the achievable development (territory, potential and so on) does depend on how crowded the board is. In the game position, black is doing very well with three stones on the top-side but with a billion stones played on the bottom side, I'd say, white is not that good there. Depending on how black plays the lower side is at best OK for white (case of black plays the horrible move) or bad (case of black capturing in the center). The value of sente is large because of the possibility to pincher a stone, in that case OK on lower side plus sente might be better than black, well the computer says so. Compare the bottom side to the top side and it is clear white is behind when black plays the good move, right?

I have heard pros refer to #3 in terms like "black/white is making good amount of territory with the number of moves played in that area". The reason it is not same as the difference in the number of moves is that having +n out of N stones in an area is much more valuable when N is a small number than when N is a large number. Eventually the player that started out under-investing(?) in an area will accept to play much lower valued moves in that area, even if this equalizes the move difference. If I am to boil it down I'd say it is necessary to look at it this way to understand the middle game, while the difference in number of moves is necessary to understand the opening (a gross oversimplification maybe).

Just some confusing thoughts :study:

==EDIT corrected some grammar and related things that might make the original harder to understand

John Fairbairn · #8

My impression of the main strand of understanding of this topic among Japanese pros who study with AI is that they turn it around and say AI play is largely about creating overconcentration in the opponent's positions. In practical terms, this means the difference between pre-AI and post-AI play is often to do with the timing of such plays. Bots, presumably seeing deeper, are more willing to play overconcentrating moves earlier, even if it means losing aji. Humans, unable to see deeply, have hitherto prefer to keep the tension alive and the options open (which, ironically, keeps the already unknowable even more unknowable!).

I came across a superb example last week or so of two young AI-driven players turning almost the entire first half of the game into a mutual overconcentration fest. It was Sumire versus Hoshiai. It is clear that Sumire's judgement of the eventual outcome (kvasir's first point) was superior, and so the efficiency of her overconcentrating moves must also have been superior. It's worth looking at this game.

The use of the actual term 'overconcentration' seem to have waned in pro talk, however,. I'm still hazy about what is replacing it. But 'aji' is being "cancelled" too, it seems, and I think the mood is shifting towards an evaluation of the position expected at the end of the opening. This is easier to predict if you play lots of forcing moves and eschew aji, and so they are no longer fussing so much about individual moves.

This is my feeling based almost entirely to sensitivity to pro language. In that connection, I think I need to say something about forcing moves. In western usage such moves are like a drug - getting high by playing a sente move that simply forces a response. The traditional Japanese has been much more nuanced. A kikashi move is not so much a forcing move but one that has an effect, and so is to be played only when that effect makes sense. I'm getting the feeling, though, that pros (ostensibly imitating AI) are playing forcing moves closer to the western sense, i.e. earlier, cashing in their chips at once.

Efficiency

Who is online