Tsumego: Go Life-and-Death Puzzles That Build Reading Skill

Read ~7 min · Updated July 2026 · Part of: Learn Go

Tsumego (詰棋, also called life-and-death problems or 死活題) are the puzzles serious Go players solve daily. Each one asks a single sharp question: can this group make two eyes and live — or can it be killed? The answer is unique and provable — not anyone's opinion, but the inevitable consequence of one rule: a group with no liberties dies, and a group with two eyes lives.

This guide explains plainly what tsumego are, why every Go school treats them as daily fundamentals, demonstrates what "reading" actually is using a 4×4 puzzle verified by the brainGO algorithmic engine, and lists the vital points of the common life-and-death shapes.

English中文Plain meaning
tsumego / life-and-death problem詰棋/死活題a life-and-death puzzle — can this group live?
life and death死活whether a group can make two eyes
two eyes兩眼the condition for living — two real eyes can never be captured
reading推算/閱讀seeing a local fight several moves ahead in your head
vital point (急所)要點/急所the single point both sides must fight for

1. What a tsumego is

You're given a local position and a clear goal — Black to kill (Black plays first and kills the white group) or White to live (White plays first and makes two eyes). There's usually one correct first move, and the solution often requires reading several moves deep, because the move that looks right is frequently a trap — the opponent has a unique reply that refutes it.

🧩 Analogy: a tsumego is to Go's life-and-death what a "mate in 3" problem is to chess. Compact, forced, and the answer is provable. The difference is that in Go, the "forcing" nature often only reveals itself on the second or third move.

Tsumego span two extremes. The most elementary are "one-move captures" — find the move that removes the enemy's last liberty (brainGO's 4×4 capture puzzles start here). The deepest are ten-move life-and-death brawls on the big board. But underneath, they are the same thing: read a local fight forward until it resolves.

2. Why players solve them daily

Every Go skill — defense, attack, judging whether a group is safe, deciding when to abandon a group — rests on one ability: reading a local fight several moves deep in your head. Tsumego train exactly that one ability, in isolation, with no time pressure.

That's why every Go school on earth treats "a few tsumego a day" as basic training — it's the weightlifting of Go, and the purest form of calculation training.

🎯 The skill transfers everywhere. Once you can read "can this group live?", you can read captures, ladders, and capturing races too — because they're all different faces of "read a few moves ahead and see who runs out of liberties first."

3. Solve one: what reading actually is (brainGO engine demo)

"Reading" is abstract until you solve a puzzle. Here's an entry-level tsumego computed by the brainGO engine — Black to play, capture in 1 move:

The reading process (step by step, like code):

  1. Count liberties. The white stone (top-left) — its only empty neighbor is (1,0); the other side (0,1) is a black stone. So white has exactly 1 liberty, at (1,0).
  2. Find the move. Since white is down to one liberty, Black simply plays (1,0) to fill it — white drops to 0 liberties and is captured.
  3. Check the opponent has no reply. Once captured, the stone is removed; there's no "next move" for white. Solved in 1 move.

That's the whole skeleton — count liberties → find the filling move → confirm the opponent can't escape. This three-step loop is the backbone of reading, from a 1-move beginner puzzle to a 10-move deep calculation. Only the number of loops grows.

🔬 That "1 move" isn't someone's say-so. The brainGO algorithmic engine re-runs and verifies it at build time — Black can indeed capture the (0,0) white stone within 1 move, and the solution is unique. What you see is a reproducible computation, not an opinion.

Of course, a 1-move puzzle is just the entry. Real tsumego often require several moves — "if I play here and he blocks, then I play here… until he can't escape." brainGO's one-stone variations are the advanced version of this: they don't just compute "can it be killed," but "how much each stone weighs toward the result."

4. Common shapes and their vital points

Beginner tsumego run on "count liberties," but the moment two-eye life is involved, you need to know a handful of classic shapes — their vital points (the point both sides must fight for) are fixed, and memorizing them lets you solve huge numbers of puzzles instantly.

Shape (enclosed empty space) Where the vital point is Result
Straight-three (3 points in a line) the center point whoever plays center wins: you play it → live; opponent plays it → dead
Bent-three (3 points in an L) the bend's corner same — the vital point decides life or death
T-four (4 points in a T) the T's center you play center → live; opponent → dead
Knife-five (5 points, knife-shaped) the middle of the blade opponent plays center → kills (beginners often misread this as alive)
Square-four (a 2×2 of empty points) no point needed always alive (wherever the opponent plays, you can make two eyes)

One rule of thumb to remember: in most shapes, life or death hinges on the center point. If you take the vital point, you live; if the opponent takes it, you die — which is why strong players solving tsumego look first for the vital point rather than testing moves one by one. For the full catalog of shapes and how to judge them, see basic life-and-death shapes and the eye.

5. The 4 most common mistakes when solving tsumego

  1. Playing without reading (guessing). The value of a tsumego is the reading; guessing correctly still trains nothing. Always trace the main line in your head before placing the stone.
  2. Missing the vital point (急所). Those "center points" above are exactly the vital points — the correct move is very often right there. With practice you develop an instinct for them.
  3. Not checking the wrong answers. Why did your move fail? What was the opponent's refuting move? The learning lives in the refutation of the wrong move, not in the solution. Failing without reviewing is the same as not solving.
  4. Reading only one move deep. A beginner puzzle might resolve in one move, but most tsumego demand you read until "the opponent can't escape no matter how they reply." Read two, three moves deep — until it's sealed.

6. How brainGO distills tsumego to 4×4

Full tsumego can be hard, and even teachers sometimes aren't sure of the correct answer. brainGO starts at the entry level, with one advantage no one else has: a 4×4 board has only 16 points, a finite number of states, so it can be completely solved by exhaustive computation. This changes three things:

It's the on-ramp to harder life-and-death later: the 4×4 capture puzzle is the "killing" half of tsumego. Build the reading skeleton here, and the big-board shapes (straight-three, knife-five, and friends) later are just "more of the same."

How to solve them well

  1. Read before you place. Don't guess — trace the main line in your head.
  2. Find the vital point. The key move is usually the point both sides want (the center of a shape).
  3. Check the wrong answers. If your move fails, understand why — that's where the learning is.
  4. Daily and short. Five puzzles in two minutes beats thirty minutes once a weekend.

The fastest way to understand tsumego is to solve one and feel the click.

👉 Play brainGO — your first capture puzzle

FAQ

How many tsumego a day to improve?

Frequency beats volume. Three to five a day, sustained for months, beats fifty once a weekend. Reading skill is a reflex built by "reading a little forward every day," not by a single big session.

How does tsumego difficulty progress?

From "1-move capture" (find the move that removes the last liberty, like cap-001 above) → multi-move reading (read 2-3 moves to seal) → life-and-death shapes (making or destroying two eyes — straight-three, knife-five, etc.). brainGO's one-stone variations are the advanced step: not just "can it be killed" but "how much each stone weighs."

Are 4×4 tsumego the same as 19×19 ones?

The rules and the "reading" are identical — only the board is smaller. The loop you learn on 4×4 ("count liberties → find the filling move → confirm the opponent can't escape") is the same muscle on 19×19 — just more space and more moves to read. The difference is that a 4×4 puzzle can be exhaustively solved by an engine, so the answer is provable; on 19×19, correct answers traditionally relied on authority, though engines can verify those too now.

Can tsumego prevent dementia?

No game has been proven to prevent cognitive decline. The real benefit of tsumego is that it gets your mind engaged and focused, feels rewarding, and is a worthwhile daily habit. For concerns about memory or cognitive decline, see a clinician.

Are tsumego and "life-and-death problems" the same thing?

Yes. The Japanese term tsumego (詰碁) is the standard name for life-and-death problems; in English they're often called life-and-death problems. They mean exactly the same thing: given a local position, ask whether this group can live.

Related guides