Sub-page of Sh

Table of contents

Thoughts on Improvement - a Guide for Beginners

This section is intended as assistance for beginners, to give them some guidance, motivation and spirit, but I am just writing my own experience, so probably YMWV.

Since it is now one year ago that I decided to learn go more seriously, I thought it would be a good idea to write a retrospect to let beginners know how much one can improve with what efforts.

1. Where to Start?

I know from my own experience that when first playing on a 19x19 board it looks overwhelmingly big, and you have no idea where to start. Also during the game there is a feeling of being lost since there are soooooooooo many points to set the stone. Some guidance is needed.

1.1 Getting the very basic knowledge about the game

The first I did was to delve into the theory, especially the unlimited resources of SL. A good place to start from are the pages for beginners. The Improvement page gives links to guides on improvement, such as that from Benjamin or Dieter. Both state the importance of doing tsumego, so I will spend an own section for that.

1.2 Playing online

Since I am a programmer and was looking for information about writing go engines, it probably took me too long to get into actual play with real people, online as well as on a real board. So I signed up at KGS eleven days later. Though it is repeatedly reported that there are rude people in the beginners' ranks (around 15-30k), I recommend it to beginners. Just don't let you put off by that. Other servers like IGS and wBaduk are more friendly, but the beginner ranks are considerably stronger than those on KGS. Beside the mentioned real time go servers there are also turn based go servers, the most famous being DGS and OGS.

1.3 Fast or slow games?

This is a question that is often asked. In order to get experience and to learn the shapes, fast games on small boards are better. A proverb says: "Lose your first 50 games as quickly as possible". On the other hand slower games are good for developing reading strength. So maybe a mix of both is the best. Personally I don't like playing fast/blitz games and almost all of my games are at least 30 min plus byoyomi. The dislike of blitz games has not stopped me from improving to the SDK ranks. For further improvement I will probably need more blitz games though. I heard that at the IBA, which is one of the leading go schools, they recommend playing many blitz games for training. Slow games are just for tournaments.

1.4 Playing on a real board

If you have a go club, go game evening or tournament nearby, don't miss the opportunity. Playing on a real board is a completely different experience. Some people have reported that they have difficulties to play on a real board after they have played online only. The best is to do both at the same time. If you have no one nearby to play, use your own board and lay out problems, difficult situations or replay online games on your board. One important difference between online play and playing on a real board is the scoring of the result. On the computer you can count the result with a few clicks. On the board you have to do it yourself (together with your opponent). The best is if you ask a stronger player to explain the scoring method.

You can do this in your local go club or at go game evenings. When I went there the first time as a newbie, I was around 18k/26k (KGS/EGF), and the players there were between 5k and 4d (EGF). Even though they were patient with me, I felt that it was no fun, neither for them nor for me, so after four times I stopped going there again. Maybe I will check back there after I have advanced to EGF-SDK. Playing against players that are a few stones stronger surely helps to improve the game, but if they are 20 or more stones stronger, it's pretty useless for everyone involved.

An alternative that I chose was playing tournaments. Since go in the Western Hemisphere is just a niche game, players of all strength participate in tournaments, i.e. also beginners. The pairing method used at these tournaments ensures that players would play against other players of similar strength. That is much fun! Besides you can learn a lot. Due to the longer pondering time and exchanging experience with other players I have gained a stone each tournament.

1.5 Playing go engines.

Some go players discourage beginners to play go engines (also known as go programs or go bots), since one can get bad habits from that. But I think as a beginner it is not wrong to play programs. My favourite program was Aya, but I also played GnuGo. There are also stronger engines available, e.g. Fuego and MoGo, but I haven't played more than two games against each of them.

The advantage of programs is that you can play them when you have no internet connection, and programs never complain if you have to interrupt or stop the game.

My recommendation is to play a go programs as long as the program still wins. If you start to win regularly, then the time has come to stop playing it.

1.6 Tsumego

"Tsumego" is the technical term for go problems whose purpose it is to train life and death or tesuji situations. Since I consider this as the most important point to improve, I have dedicated an own section for this (see below).

1.7 Go books

Once you have acquired some basic knowledge, you'll probably want to get some more knowledge and ideas from go books, but don't know which book is approriate for which strength. According to my experience you will get most of the books when being at the listed ranks (which all refer to KGS).

Regarding the problem books it depends on how you work with them. If you just want to learn and hardwire the problems and want to understand them, you need less skills than if you want to solve them routinely. Hence these books are listed twice, once for "learn and hardwire" and once for "solve routinely".

  • 25-30k: Graded Go Problems, vol 1 (learn and hardwire)
  • 20-25k: Graded Go Problems, vol 1 (solve routinely)
  • 20-25k: Opening Theory Made Easy (also teaches about basic shapes)
  • 20-25k: Graded Go Problems, vol 2 (learn and hardwire)
  • 15-30k: Learn to Play Go, vol 1-5
  • 15-20k: Graded Go Problems, vol 2 (solve routinely)
  • 15-20k: In the Beginning (theory about the opening; easy to understand and necessary to get rid of the feeling to be lost in the beginning of a game)
  • 15-20k: 1001 Life and Death Problems, Part 1-2 (One move problems - learn and hardwire)
  • 12-18k: Get Strong at Tesuji (learn and hardwire)
  • 10-15k: Tesuji (study the theory and the examples, not the additional problems which are considerably more difficult)
  • 10-15k: 1001 Life and Death Problems, Part 3-4 (Three move problems - learn and hardwire)
  • 10-15k: Making Good Shape (learn the theory part)
  • 10-15k: Graded Go Problems, vol 3 (learn and hardwire)
  • 10-15k: Get Strong at Life and Death (learn and hardwire)
  • 10-15k: Get Strong at Invading (learn and hardwire)
  • 7-15k: 1001 Life and Death Problems, Part 1-2 (Three move problems - solve routinely)
  • 7-12k: 501 Opening Problems (learn and hardwire)
  • 7-12k: 38 Basic Joseki
  • 7-12k: 501 Tesuji Problems (learn and hardwire)
  • 7-10k: 1001 Life and Death Problems, Part 5-6 (Five move problems - learn and hardwire)
  • 5-12k: Get Strong at Tesuji (solve routinely)
  • 5-10k: Tesuji (read it again and now you can do the more difficult problems)
  • 5-10k: 1001 Life and Death Problems, Part 3-4 (Three move problems - solve routinely)
  • 5-10k: Making Good Shape (do the problems)
  • 5-10k: Graded Go Problems, vol 3 (solve routinely)
  • 5-10k: Get Strong at Life and Death (learn and hardwire)
  • 5-10k: Get Strong at Invading (solve routinely)
  • 3-12k: Attack and Defense
  • 3-10k: Lessons In The Fundamentals Of Go (controversial book: some people say it's the best go book, others can't get much out of it; it's a good read nevertheless)
  • 3-7k: 1001 Life and Death Problems, Part 5-6 (Five move problems - solve routinely)
  • 3-7k: 501 Opening Problems (solve routinely)
  • 1-7k: 501 Tesuji Problems (solve routinely)

Books that I will pick up later (again); These books are probably all SDK level:

1.8 Teacher

Some say that having a teacher is important. I haven't tried this, so cannot say much about it. As long as I still see improvement for myself by self-studying, I won't get a teacher.

1.9 Reviewing your games

One important point of improvement is to know your flaws. You can achieve this by reviewing your games. Of course only meaningful games (i.e. slower games that you lost) are approriate for this, since blitz games are prone to obvious errors that decide the game. Find the decisive point where you still could have won the game and/or two or three other points where you could have done better.

Sometimes you will have no idea what you could have done to win a game. In that case get your game reviewed by someone stronger than you. This will give you new ideas for future games. You can either participate in a teaching ladder or request a review at Life in 19x19 which is by the way a good and friendly place to talk about go related stuff.

Don't forget to review games of weaker players once you have achieved a certain strength. It is a give and take, and to make the system work everyone has to contribute one's mite.

2. About Efforts and Improvement

Since I have already mentioned the Improvement page on SL, I am going to write about my own improvement in this section.

2.1 Efforts

Counting my 19x19 games that I've had during the last 365 days I have to state that 300 games seem pretty few, and when counting 13x13 games, it's just one game per day on average. I often see people playing 5-10 games per day (or even more), also those that are around my rank (not to mention the dan players of course). Here are some stats that show where I've played so far and how much (until 2009-07-01):

KGS: 118
IGS: 60
CyberORO: 57
Other servers (turn-based): 5

Learning matches (at go game evening): 4 (and another dozen or so 13x13 games)
Teaching matches (with my kids/nephews/parents/siblings): 24 (and another few dozens on 9x9 and 13x13)
High-Handicap fun matches (with my kids): 12
Tournament matches: 31

The tournament matches only make 10% of my total games, but it seems that they are the games I took most benefit from. I had my first tournament last November (2008-11-01), i.e. four months after I had started with go. In spite of me just being a 20 kyu (EGF, and 16k KGS) I found it very exciting. I got acquainted to many new people, and for two out of my five opponents it was the first tournament like me. I found (and still find) it extremely cool that the pairing of opponents according to McMachon is aimed at finding two equally good people to play (it is different in common chess tournaments where they use the Swiss system). After a while you know most of the people there, since they often come to tournaments. I have already played four times with one of my opponents (out of six tournaments), who is around my strength, though I am slowly overtaking him and my rank/rating is already ahead of him. I have taken him (secretly) as one of my "rivals".

2.2 Improvement

I know ranks are just arbitrary numbers, but new players might be interested in what is possible to achieve with just one game per day for a whole year:

KGS: 8k (finally SDK, yeah!)
IGS: 12k
CyberORO: 16k
EGF: 15k (at the time of my last tournament at the end of March, so now probably around 13k)

Ok, "just one game per day" is not the complete truth. During that year a bunch of books have gathered on my bookshelf. From them I have acquired my theoretical knowledge and done lots of tsumego. I think with tsumego I spend the main part of the time that I dedicate to go. It's something I can do during lunch time at work or at the weekend when my wife and kids are still sleeping - usually I get up at 6 o'clock in the morning. That's the time to read a good book or do some tsumego.

I know, 8k in one year is not that fast, and there are others that develop quicklier but there are also persons that do not develop any more at that stage. One of my opponents of my third tournament has had no development for more than one year. He got stuck at 16k EGF / 8-9k KGS. Another person that I got acquainted to at the same tournament has developed much quicklier. At the tournament I could still beat him giving him three stones, but now his rating is already better than mine. I'm going to take him as my new rival.

3. Tsumego

3.1 About reading and intuition

Many beginners reported that they have difficulties at reading and that they found it very frustrating. I myself did. Be assured that this is normal, so don't be worried or frustrated or put off by that. I will explain about it with an example and show you a way to get out of this problem.

Imagine you want to read out a local situation with a depth of 3 moves for each side, i.e. 6 moves in total. Now as a beginner assume that there are (on average) five possible points to set on for every move. That means that you would have to read out about 5*5*5*5*5*5 moves in total, i.e. more than 10000 positions. That is of course an impossible mission. So what's the way a more experienced player handles this? The solution is that for an experienced player only one or two points are eligible, because he knows that the others don't work. So he probably needs to read out just about 2*1*2*1*2*1 moves - only 8! That's not a too difficult job. The branching factor is the key to the problem. So the solution for you as a beginner is to learn shapes and hardwire them into your brain.

A few years ago I heard about a research that wanted to find out what makes the top chess players so good at chess. Particularly they wanted to know which part of the brain is more active than in amateur chess players. The interesting result was, that it was the long-term memory. So the top chess players just had to recall the patterns from their long-term memory in order to play at top level. Since go is - much more than chess - a game of shapes and patterns, how much more would this result of the research apply to go?

So now you have learned that knowing many shapes is essential in go, and that is one reason why tsumego exist. By solving and repeating them you hardwire the shapes and develop a better intuition for good moves. Until you are about 15k you will probably improve just because your shape knowledge increases, and most of your games will be based on intuition. When you are approaching the SDK levels, reading will become more and more important. And that's the other reason why tsumego exist: to practise reading.

If you would like to start practising reading as a beginner, you can do so. You just need something where the branching factor is very low. Ideal for this purpose are ladders, so practise reading them. This is a fundamental technique that you will need later as well. Recently I had an even game as 9k against an unranked player who turned out to be a 4k later. In the game there was a ladder situation. I read it out and found out that it worked, but my opponent was obviously convinced that it did not work, but he hadn't read it out. So the ladder was played out. When he suddenly noticed that the ladder would work, he resigned. I had a similar situation in a tournament as well where the ladder was played out and my opponent resigned after she noticed the ladder would work. Note that in games of advanced players ladders are almost never played out. Nevertheless it is essential to know in a special situation whether the ladder works or not.

3.2 Many easy or a few difficult tsumego?

This is pretty much the same problem as whether to play many fast games or a few long games as discussed above. The IBA again recommends to do many easy problems. The improvement pages on SL mentioned above suggest to do both. It actually depends on what you want to achieve. If you want to improve your reading, do some more difficult ones, but not spending more than 10 minutes on each. If you want to develop your intuition and shape knowledge, do (repeatedly) many easy ones. Especially for beginners I recommend just to do the latter. Later when you are in the advanced SDK ranges, you can do more difficult ones, but I'd still suggest to spend less then a quarter of the tsumego solving time on these.

3.3 Look at the solution?

Some consider it as important not to look at the solution. They say, it is necessary to read it out completely until you are sure that you have the correct solution. I checked this out and compared my solution to the given one, but in some cases my solution was still wrong. My conclusion is that not looking at the solution is not a good idea, because you memorize wrong ideas. This is especially bad when you are still in the phase of hardwiring shapes and patterns. My suggestion is to read a problem out, but not too long. Then compare your solution with the given one. If you got the problem wrong, read it out more thoroughly until you are sure why your solution did not work. Doing it this way you practise reading and hardwiring at the same time without the danger of hardwiring wrong things.

3.4 Online Tsumego

There are several sources of tsumego on the internet, but here I will focus on two:

  • This is probably the best known resource for interactive tsumego on the web. You can select the kind of tsumego you want to do - e.g. life and death problems or tesuji problems, and also the difficulty can be selected. As a beginner you can start off with problems between 30k and 25k.
  • [ext] This is a jewel that not many go players know. It contains an online course in combination with about 3100 tsumego (2009-08-01). It starts from the very beginning, e.g. counting liberties, atari etc. It claims to take the beginner to about 20k, and that is indeed what it does. Very recommendable.

3.5 My intensified tsumego training program

This is described in my blog and will be updated there.

Sh/Improvements last edited by PJTraill on September 12, 2018 - 22:38
RecentChanges · StartingPoints · About
Edit page ·Search · Related · Page info · Latest diff
[Welcome to Sensei's Library!]
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Login / Prefs
Sensei's Library