goproblems.com

Weiqi problems (also called go problems, baduk problems, or tsumego) are short, self-contained board positions with a correct sequence of moves to find. You train with them by solving sets matched to your rank, on a schedule that mixes untimed reading practice with timed drills, and by tracking which problem types you keep failing. On GoProblems.com, for example, problems are organized by difficulty from 30 kyu up to 1 dan and beyond, grouped into themed collections, and served through daily sets, time trials, and tournaments.

What counts as a weiqi problem

A weiqi problem gives you a position and a goal: live, kill, connect, cut, capture a race, or find the best local move. The solution is usually a short forced sequence, so the whole exercise is a reading drill rather than a full game.

The main categories you will see in a problem database:

GoProblems.com's collections reflect this spread — for instance, Semeai / Capturing Race, Connect, Cutting, Snapback, Ladder, Squeeze Tesuji, Under the Stones, and an Encyclopedia of Life and Death series.

Why solving problems improves your game

Problem solving builds three things that show up directly in real games:

• Reading depth. You calculate a sequence to a conclusion instead of stopping at "this looks fine."
• Shape recognition. After enough repetitions, weak and strong shapes become visible without conscious counting.
• Calculation speed. Timed practice compresses the time you need to verify a sequence, which matters under clock pressure.

The mechanism is repetition with feedback: you commit to a move, the problem tells you whether it works, and the correction sticks. Guessing defeats this, because a lucky guess teaches nothing about why the other move failed.

How to pick problems at your level

Difficulty is expressed in kyu/dan ranks, and the right zone is problems you can solve but not instantly. If you solve everything on sight, the set is too easy; if you fail most attempts, it is too hard to learn from.

GoProblems.com labels its daily set by rank, for example:

• 30 kyu — Beginner
• 19 kyu — Easy
• 9 kyu — Medium
• 1 dan — Hard

A practical rule: work mostly at or one step above your current rank, and keep a smaller share of clearly easier problems for speed drills. The site also lists recommended problems by rank (30 kyu, 29 kyu, 26 kyu, 23 kyu, 22 kyu, and so on), which gives you a starting ladder if you are unsure where you sit.

Training modes and how to use them

GoProblems.com exposes several modes, each suited to a different purpose.

Problems of the day

A daily set split by rank. Use it as your baseline habit — solve the set for your level, then check whether your accuracy or speed changed from previous days.

Time Trial

Timed solving at three settings: 3 minutes, 5 minutes, and untimed. The page shows a "today's best" figure per mode (for example, 3 min: 28, 5 min: 21, untimed: 22), so you can compare your own count against the day's benchmark.

• 3 min / 5 min — trains speed and pattern recognition. Accept lower accuracy here.
• Untimed — trains depth. Take as long as needed and read to the end.

Recommended problems

A rank-ordered list that fills gaps in your level. Good for a structured session when you do not want to choose.

Collections

Themed sets with visible progress. Collections progress on the site includes large sets such as Semeai / Capturing Race (3091 problems), Connect (973), Encyclopedia of Life and Death - Easy (741), Cutting (707), Local Choice (583), Snapback (509), Ladder (487), Squeeze Tesuji (487), GuanZiPu (432), and Under the Stones (373). Pick one theme at a time when you notice a repeated weakness.

Tournaments

Scheduled competitive solving, listed with status, time, ruleset, and entrant count. The site shows formats such as "Solving royale" with a countdown and a winner recorded for completed events. Use these if you want external pressure and a score to compare.

Adding and discussing problems

The database is community-driven: you can add new problems, search, and comment. The newest-problems list shows submissions with an estimated rank and the contributor's name, so you can see how fresh material enters the pool.

A simple routine you can keep

1. Warm up (5–10 min). One timed set — 3 or 5 minutes — to wake up pattern recognition.
2. Main work (15–30 min). Untimed problems at your rank or one above. Read each to a definite result before answering.
3. Theme block (10–15 min). One collection that matches a weakness, for example Ladder or Snapback.
4. Review. Note the problems you missed and why: wrong first move, stopped reading too early, or miscounted liberties.
5. Track. Record your daily counts and accuracy so you can tell whether a change in routine helped.

Consistency beats volume. A short daily session with honest reading produces more improvement than a long session of guessing.

Common sticking points

• Guessing instead of reading. If you answer in a few seconds on an untimed problem, you are probably pattern-matching, not calculating.
• Only easy problems. Comfortable sets feel productive but stop teaching.
• Only hard problems. Constant failure without understanding the refutation does not build skill.
• Ignoring the failure reason. The value is in the correction, not the score.
• No timed work at all. Reading depth without speed leaves you short of time in real games.

FAQ

Are weiqi, go, and baduk problems the same thing? Yes. Weiqi, go, and baduk are names for the same game, and problem sets are labeled interchangeably — GoProblems.com uses all three terms.

Do I need an account to train? The site describes interactive solving, search, commenting, and adding problems. Whether an account is required for specific features is not stated in the available material, so check on the site.

How long should a session be? Long enough to complete a focused set with real reading. The modes above are built around 3-minute, 5-minute, and untimed blocks, which is a reasonable frame to build from.

Which problem type should a beginner start with? Life and death and basic capturing races, at beginner ranks (the site's daily set starts at 30 kyu). Add tesuji and cutting once basic reading is reliable.

A Go problem database is a searchable, ranked collection of Go problems (tsumego, tesuji, and more) that you can filter by difficulty, category, and skill type, rather than working through a fixed sequence in a book. GoProblems.com is one example: it organizes problems by kyu/dan level, groups them into themed collections, and adds training modes like daily problems, time trials, and tournaments. Use a database when you want volume, variety, and the ability to target a specific weakness; use a book or fixed set when you want a single linear curriculum.

What makes it a database rather than a problem set

A book or a single PDF gives you one path through a fixed number of problems. A database gives you:

• Filtering by rank — problems labeled from 30 kyu up to dan level, so you can pick material near your strength.
• Filtering by theme — life and death, tesuji, capturing races, joseki, fuseki, endgame, and more.
• Search and community contribution — users can add new problems, comment, and discuss solutions.
• Progress tracking — collections show how many problems you've completed.

That combination is what turns a pile of problems into a training tool: you can decide what to practice today, not just how many.

The main problem categories

If you're not sure where to start, life and death is the highest-return category for most players below dan level, because it converts directly into won games.

How to pick problems at the right level

Use the rank labels as a starting filter, then adjust based on how you actually solve:

• If you solve most problems instantly, the level is too easy — move up.
• If you can't find the first move after a few minutes, the level is too hard — move down.
• The productive zone is where you sometimes fail but can usually find the answer with real reading.

The site's "Recommended problems" section lists levels from 30 kyu upward, and "Problems of the day" is split into bands (30 kyu Beginner, 19 kyu Easy, 9 kyu Medium, 1 dan Hard), which gives you a ready-made difficulty ladder.

Training modes you can build a routine around

Problems of the day

A daily set at your chosen difficulty. Good as a fixed warm-up: solve the day's problems before anything else.

Time Trial

Timed solving in 3-minute, 5-minute, and untimed modes, with a daily best score shown (for example, "Today's best: 3 min 28 · 5 min 21 · Untimed 22"). Use this to measure speed and accuracy under pressure, not to learn new shapes.

Collections

Themed sets with progress counters, such as Semeai / Capturing Race (3091 problems), Connect (973), Encyclopedia of Life and Death - Easy (741), Cutting (707), Local Choice (583), Snapback (509), Ladder (487), Squeeze Tesuji (487), GuanZiPu (432), and Under the Stones (373). Pick one collection and work through it over weeks rather than jumping between themes.

Tournaments

Live solving events ("Solving royale") with entrants and a winner, plus a countdown to the next one. Useful if competition keeps you motivated.

Tracking progress and reviewing mistakes

• Use the collection progress counters to see how far through a theme you are, and finish one collection before starting another.
• Re-solve problems you got wrong a few days later. A problem you failed once and then solved from memory hasn't been learned.
• Keep a short error log: note the type of mistake (missed a snapback, miscounted liberties, wrong order of moves) rather than the specific problem. Patterns in your errors tell you which collection to open next.
• Use the time trial best scores as a rough speed benchmark over time, not as your main measure of improvement.

A simple weekly routine

1. Solve the daily problems at your level.
2. Work through one themed collection for 15–30 minutes.
3. Run one time trial to test speed.
4. Once a week, re-solve the problems you missed.

Adjust the difficulty band whenever your solve rate drifts too far in either direction.

Go problems are short, solvable positions—usually life-and-death, tesuji, or capturing-race situations—that you read out to a definite result. Solving them regularly trains the two skills that decide most games: reading ahead and recognizing shapes. GoProblems.com organizes them into a searchable database with daily problems, timed modes, and collections, so you can practice at your own rank from 30 kyu up to dan level.

What counts as a "go problem"

A go problem (also called tsumego, or baduk/weiqi problems) is a fixed board position with a best sequence of moves. You don't play a full game—you find the key move and the follow-up. The site's collections show the main families:

The largest collections on the site include Semeai / Capturing Race (3091 problems), Connect (973), Encyclopedia of Life and Death – Easy (741), Cutting (707), Local Choice (583), Snapback (509), Ladder (487), and Squeeze Tesuji (487).

Why solving them improves your play

Reading is a skill you build by repetition, not by talent. Each problem forces you to:

1. Read to a conclusion — you can't guess; the position has a right answer.
2. Recognize patterns — after enough snapbacks and ladders, you see them before you calculate.
3. Count liberties and space — essential in capturing races and life-and-death.

The payoff is that in a real game you spend less time on shapes you've already solved and more time on the genuinely new situation.

Choosing problems by rank

The site labels problems by kyu/dan level, so you can start where you actually are rather than where you'd like to be. Its daily set is split into four tiers:

• 30 kyu — Beginner (119 problems)
• 19 kyu — Easy (97)
• 9 kyu — Medium (56)
• 1 dan — Hard (26)

A practical rule: pick problems you can solve in a few minutes but not instantly. If you're getting nearly all of them right, move up a tier; if you're stuck on most, move down. The "Recommended problems" list (shown from 30 kyu downward) and "Newest problems" feed give you fresh material at your level.

How to practice on GoProblems.com

The site gives you several modes, each with a different purpose:

• Problems of the day — a fixed daily set across the four difficulty tiers. Good for a consistent habit.
• Time Trial — solve as many as you can in 3 minutes, 5 minutes, or untimed. The page tracks a daily best (for example, "Today's best: 3 min 28 · 5 min 21 · Untimed 22"), which turns practice into a measurable score.
• Collections — themed sets (ladder, snapback, semeai, etc.) so you can drill one weakness at a time. Each shows a progress count.
• Tournaments — live "Solving royale" events where entrants compete; past events list a winner.

A simple routine

1. Warm up with the daily problems at your tier.
2. Run one Time Trial to measure speed and accuracy.
3. Open one collection that matches a weakness (e.g., Ladder if you keep misreading them).
4. Track your tier and your Time Trial best over a week—if both improve, you're ready for the next level.

Common sticking points

• Guessing instead of reading. If you click a move hoping it works, you're not training reading. Read the whole sequence first.
• Practicing too far above your rank. Hard dan problems won't help a 20 kyu build fundamentals.
• Only doing one category. Life-and-death is the base, but tesuji, capturing races, and cutting/connecting all show up in real games.

Start at your honest rank, do a small set daily, and let the Time Trial score tell you when to move up.

Tsumego is the Japanese term for Go problems that ask you to read out the status of a group — typically life-and-death situations where you must find the sequence that kills, lives, or connects. You practice by solving problems at your rank, reading every variation before you play, and reviewing the ones you get wrong. This works for any player from beginner to dan level, and the main requirement is a problem source organized by difficulty, such as the collections on GoProblems.com.

What counts as tsumego

Tsumego focuses on local fights over life and death. The question is usually narrow: can this group make two eyes, or can the surrounding stones kill it? That makes it different from other problem types you'll see in a Go problem database:

The line between tsumego and tesuji blurs in practice, because the killing move is often a tesuji. The distinction is intent: tsumego is about the life-and-death result, while a tesuji problem is about finding the sharpest move in a shape.

What tsumego actually trains

Solving life-and-death problems builds one core skill: reading. You learn to hold a sequence of moves in your head, branch into alternatives, and judge which branch ends with the group alive or dead. That skill transfers directly to real games, where the same shapes appear in corners and along the side.

GoProblems.com organizes its collections around exactly these recurring shapes, which tells you what you'll be drilling:

• Semeai / Capturing Race — 3091 problems
• Connect — 973
• Encyclopedia of Life and Death – Easy — 741
• Cutting — 707
• Local Choice — 583
• Snapback — 509
• Ladder — 487
• Squeeze Tesuji — 487
• GuanZiPu — 432
• Under the Stones — 373

Each collection isolates a pattern family, so you can target the shapes that keep costing you games rather than solving at random.

A practical practice routine

1. Start at your own level

GoProblems.com labels problems by rank, from 30 kyu up through dan level. The "Problems of the day" section shows a set per level — for example a 30 kyu beginner set, a 19 kyu easy set, a 9 kyu medium set, and a 1 dan hard set. Pick the level where you solve most problems but still miss some. If you solve everything instantly, move up; if you fail most attempts, move down.

2. Read before you play

The point of tsumego is the reading, not the answer. Before placing a stone, work out the full sequence in your head: your move, the opponent's best reply, your follow-up. Only then play. If you guess and check, you're training pattern recognition at best and memorization at worst.

3. Review every mistake

When you get a problem wrong, don't just look at the answer. Re-read the position and find where your reading broke down — did you miss the opponent's strongest reply, or stop one move too early? That gap is the thing worth fixing.

4. Use timed modes to test recall

GoProblems.com offers a Time Trial with 3-minute, 5-minute, and untimed modes, and shows a daily best for each. Timed solving pressures you to recognize shapes quickly, which is closer to how you need to read in a real game. Use untimed mode while learning a new pattern family and timed mode once you're comfortable.

Tracking progress and raising difficulty

Two signals tell you it's time to move up:

• Your accuracy at the current level is high and your solving is fast.
• The collections you've been working through feel automatic.

The site tracks "Collections progress," so you can see how far through a collection you are and which ones you've finished. The "Recommended problems" section also surfaces problems near your level, which is a low-effort way to keep difficulty matched to your current strength. As you improve, shift from easy life-and-death sets toward harder ones and toward pattern families you haven't drilled, like Under the Stones or Snapback.

Common mistakes

• Memorizing answers instead of reading. If you recognize a problem and recall the solution, you've learned nothing new. Read it out anyway.
• Solving only easy problems. Comfortable problems feel productive but don't stretch your reading. Keep some problems just beyond your level in the mix.
• Skipping the opponent's best reply. Many wrong answers come from assuming the opponent plays passively. Always ask what their strongest response is.
• Treating tsumego as separate from your games. The shapes you drill are the shapes that decide real fights. When a game reaches a life-and-death situation, slow down and read it the same way you read a problem.

If you want a starting point, open the "Problems of the day" set at your rank, solve it untimed while reading every variation, and review anything you miss before moving to the next set.

Baduk problems are short, solvable positions — usually life-and-death or tactical situations — that you read out and answer without playing a full game. They go by several names: go problems, baduk problems, weiqi problems, or tsumego (the Japanese term most often used for life-and-death). The point is not to memorize answers but to train the reading and calculation you need during a real game. You should train with them if you want to improve your ability to judge whether a group lives or dies, spot tesuji (clever tactical moves), and calculate sequences accurately under time pressure.

What counts as a baduk problem

A problem gives you a fixed board position and a goal: kill the opponent's group, save your own, capture a race, or find the best local move. You read out the variations in your head, choose a move, and get feedback. Because the position is small and the goal is clear, you can focus entirely on calculation instead of the whole-board strategy a full game demands.

The main categories you'll encounter:

• Life and death (tsumego) — can a group make two eyes, or can it be killed? This is the core of most problem sets.
• Tesuji — finding the sharp, often surprising move that works where the obvious move fails.
• Capturing race (semeai) — two adjacent groups fight to capture each other; you count liberties and find the winning order of moves.
• Snapback — a sacrifice that lets you recapture a larger group.
• Ladder (shicho) — reading whether a ladder works, including ladder breakers.
• Cutting and connecting — whether stones can be separated or joined.
• Joseki and fuseki — opening and corner patterns, which are more about shape and choice than a single correct kill.

On GoProblems.com these categories appear as collections you can work through, with counts shown for each — for example, Semeai / Capturing Race, Connect, Cutting, Snapback, Ladder, and Squeeze Tesuji are all listed as separate collections, alongside an "Encyclopedia of Life and Death - Easy" set.

How to train so it actually transfers to your games

The trap with problem training is solving hundreds of puzzles while your actual play stays the same. The fix is a loop: read, check, repeat, and apply.

1. Start at your level, not above it

Problems are ranked by kyu/dan level. GoProblems.com shows a "Problems of the day" set split by strength — for example 30 kyu Beginner, 19 kyu Easy, 9 kyu Medium, and 1 dan Hard. If you're a beginner, start at the beginner tier even if it feels easy. Accuracy at an easy level builds the reading habit; struggling at a level too high just teaches you to guess.

2. Read before you move

Commit to a full calculation before you play the first move. If you can't see the whole sequence, that's information — it tells you the problem is at the edge of your ability, which is where learning happens. Only then check the answer.

3. Repeat your mistakes

A problem you got wrong is worth more than ten you got right. Redo it a day later without looking at the solution. The goal is that the shape becomes instantly recognizable, so in a real game you don't have to re-derive it from scratch.

4. Use timed and daily modes to build consistency

Regular short sessions beat occasional long ones. GoProblems.com offers a "Problems of the day" set and a Time Trial with several modes — 3 min, 5 min, and Untimed — and tracks your best score in each (the page shows sample bests like "3 min 28 · 5 min 21 · Untimed 22"). A few minutes daily keeps the reading muscle sharp. The site also runs Tournaments (for example a "Solving royale" format) if you want competitive practice.

5. Apply the shapes in real games

After a session, watch for the same shapes in your own games — a snapback you just drilled, a capturing race you now count correctly. This is the step that converts problem skill into rank. If you never connect the two, you'll solve puzzles well and still lose the same fights.

A simple weekly routine

Common sticking points

• Guessing instead of reading. If you click moves to see what happens, you're not training calculation. Read first.
• Only doing easy problems. Comfortable problems maintain skill but don't raise it; push slightly past your level.
• Skipping review of wrong answers. The correction is where the learning is.
• Never applying it. Problem skill that stays in the problem set won't show up in your rank.

If you want a structured place to start, GoProblems.com organizes problems by level, theme, and collection, with daily sets, timed trials, and tournaments, so you can pick a level and a theme and begin immediately.