How to Get Better at Go: A Practical Plan for Beginners
To get better at Go: play a lot of games, review the one or two big mistakes in each of them, solve life-and-death problems regularly, and move up to a bigger board only once the smaller one has stopped surprising you. That is the whole plan. Everything below is detail on how to run it, in the order it pays off, because the common failure is not laziness. It is studying the wrong thing at the wrong time, usually opening theory, years too early.
The short version
- Learn the rules properly. An evening at most.
- Play 9x9 against the computer until you can beat the medium level regularly.
- Move to 13x13, then to 19x19 when 13x13 feels small.
- At every stage, solve life-and-death problems and review your losses.
- Start playing people once you can finish a game without losing a big group by accident.
Step 1: learn the rules
There are about five of them and they fit on an index card. The interactive tutorial teaches them on a real board, one move at a time, and the rules of Go is the written version with diagrams for liberties, capture, ko, eyes and scoring.
You do not need to understand the rules deeply before you start playing. You need to know them well enough that you are not confused about why a stone disappeared. That takes most people ten to twenty minutes.
Step 2: play 9x9 until it is boring
Play 9x9 against the computer and expect to lose a lot. That is the point. A 9x9 game takes fifteen or twenty minutes, so you can fit a dozen into a week, and every one of them puts you in a capturing race and an eye-shape decision within thirty moves. A 19x19 game gives you the same lessons once every two hours.
Beat the easy level consistently, then the medium level. Somewhere in there you will notice you have stopped losing groups by accident and started losing on points instead, which is real progress. 9x9 strategy covers what to do on that board.
Step 3: move up in stages
When medium no longer challenges you, go to 13x13. The fights are the same, but there is now a genuine opening and enough room for a group to run, so the strategic ideas start to have somewhere to happen.
After that, 19x19, which is the full game and where most serious play happens. Do not rush this step. A beginner on 19x19 spends most of the game lost, and lost is a bad state to learn in. Which board size to play goes through what changes at each size.
What to study at each stage
Study in roughly this order, and ignore the next thing until the current one is comfortable.
At 9x9: liberties and capture, then eyes and life and death. Almost every beginner game is decided by a group living or dying.
Early 13x13: strategy basics and the ten common mistakes. This is the stage where you stop answering every move and start choosing.
Later 13x13 and 19x19: good shape, then the opening, then the endgame. The endgame is last on this list and first in points gained per hour of study, so do not skip it out of boredom.
Joseki, the memorised corner sequences, can wait a long time. Learning them early tends to make you worse, which is common enough that there is a proverb about it.
Tsumego: the highest-value study habit
Tsumego are life-and-death problems: a small position, one side to play, and one move that lives or kills. They are the closest thing Go has to a workout, because the skill they train, reading a sequence accurately before you play it, is the skill that decides games at every level.
The answer is the middle point of the three. Playing it splits the space into two separate eyes, and a group with two eyes can never be captured.
That is the whole idea, scaled up. A problem collection starts here and works up to positions that take strong players several minutes. Do a handful a day rather than an hour a week. Books and phone apps exist at every level, and it is worth trying a few rather than trusting anyone who calls one of them the best.
Play people, not just the computer
A computer opponent is patient, always available, and never plays the weird move that teaches you something. Once you can finish a game without losing a group by accident, start playing humans. You can play a friend on the same screen from the play page, and online Go servers covers the places to find opponents at your level, most of which handle matching and ranking for you.
Humans also give you handicap games. Starting with a few extra stones against a stronger player is the traditional way to improve, and it works: you get a real game against someone who can explain it afterwards.
Review your own games
Reviewing does not mean analysing every move. It means finding the one or two moments that decided the result. Two questions cover most of it.
Did a group die, and if so, when did it become unsavable? Usually several moves before it was actually captured. Find the last move where it could still have been defended.
Where did the score flip? If you were ahead at move forty and behind at move sixty, twenty moves contain the answer. Count roughly at a few points through the game and look at the stretch where the count changed.
That is ten minutes per game and it is worth more than ten games played without it.
How ranks work, briefly
Amateur Go ranks run downward in kyu and then upward in dan. A complete beginner is somewhere around 25 to 30 kyu. You improve toward 1 kyu, and the next step above that is 1 dan, then up to around 7 dan for very strong amateurs. Professional ranks are a separate scale entirely.
One rank is roughly one handicap stone, which makes the system useful: a 10 kyu and a 7 kyu can play an even-feeling game if the weaker player starts with three stones.
Progress is fast at the start and slows down. Going from 25 kyu to 15 kyu happens quickly with regular play, and every rank after that costs more games than the one before. How long single-digit kyu takes varies enormously, so treat any specific timeline you read as one person’s experience rather than a schedule.
How long does it take?
The rules take ten minutes. Feeling like you know what you are doing in a game takes a few dozen games. Getting genuinely good takes years, and that is the appeal rather than the problem: people play this game for decades and keep finding new things in it. A game whose depth ran out after a year would not have lasted two and a half thousand years.
The best next step is a short game rather than more reading. Start one on the home page and play it to the end, including the endgame, then count the result yourself.
Frequently asked questions
Is Go hard to learn?
The rules are not. There are about five of them, there are no special pieces, and most people can play a legal game after ten minutes. The game itself is hard, because the board is large, everything is connected to everything else, and good judgement comes from pattern recognition you can only build by playing. That gap between easy rules and a difficult game is why people stay with Go for decades.
How long does it take to get good at Go?
You can play a reasonable game within a few dozen games, which for most people is a few weeks. Reaching single-digit kyu, where you are clearly a competent player, takes regular play and review over a long period, and the timeline varies far too much between people to give a useful number. Improvement is fast at first and then slows steadily.
What is the fastest way to improve at Go?
Play many short games and review each one for the single move that decided it, while doing a few life-and-death problems most days. Playing without reviewing plateaus quickly, and studying without playing does not stick. The combination is what moves people up ranks.