Learning Go

I got to play another game of go at our local game club last Friday. It’s pretty surprising how many people had no idea what the game is, despite it being just about the most popular boardgame in the world. Regardless, the opportunities to play are not far and wide, even if I do like the game: most gamers who even know what it is about just look at how abstract go is and give it a pass, preferring anything with some more structure and more players. Still, from the few games I’ve got to play I have to say that I like go: it’s very deep and surprisingly interpretive for all of its abstractness: the stones have strong relationships to each other and there’s tremendous variety in the possibilities. Playing go is very entertaining and not at all heavy and difficult, as one might assume with experience from worse abstract games.

It might be useful if I write some words on the difficult topic of learning go. While most games are played for fun, players of go are often not very fun for us normal people: they are very serious, they outright expect you to not have any fun with the game for the first dozen games, and they insist on teaching the game by playing it instead of explaining it. The latter I find especially frustrating, as I’m not getting any wiser by putting random stones on the board, but apparently it has something to do with the difficulty of teaching: a person who knows how to do something won’t necessarily be very good at putting it to words.

Also, it doesn’t frankly help the situation that many layman sources on the game mystify the thought processes involved. Usually this involves stories about how difficult go is for artificial intelligences, or how playing go cures Altzheimer’s disease. Much of this is probably because go is too difficult for a dull-minded journalist to actually learn, so he’s left mumbling vaguely about aesthetics of the board of something like that, instead of just saying straight out what the players are trying to do.

The following assumes that you know the basic rules of go, but like me, have never really grasped enough of the procedure to really make reasonable decisions in the game. I’ve only played four sessions of go since Christmas, when I got a board myself, so I’m far from an expert. (My previous experience is even less; while I’ve known the rules and the culture around the game for ages, I never got around to playing it, secluded as I am here in Sonkajärvi.) The following reflects the understanding I have of the game as it stands now:

Living and dead groups are key

As you probably know if you’re at all familiar with the rules of go, a “living” group of playing pieces is one that the opponent cannot capture because you can defend against any such efforts. Specifically, the surefire way to protect your group is to create two “eyes”, which are empty spaces surrounded by your playing pieces. (There are other ways, such as connecting the group to a stronger group.) A “dead” group, on the other hand, is one that, despite your efforts, will succumb and be captured by the opponent in the long run.

This is not outright said anywhere usually, but I started to understand what go was about when I realized that the strategic play depends on correctly predicting the mortal status of your and opponent’s groups: if you can predict a group’s status before your opponent can, you can conserve your moves for the important places while the opponent wastes his in trying to attack a living position or defending a dead one. A basic understanding of the purpose of the game can be attained when you realize that your goal is to create living groups and try to prevent your opponent from creating them.

Stones extend influence

This is a bit abstract realization, but I think it’s necessary for understanding why go players play the way they do. How do you attack in go? This needs to be understood to play the game at all, I find. It’s fucking stupid to just play randomly against an experienced opponent when he could just tell me how this stuff works, but for some reason it’s apparently considered somehow better for people to figure this stuff out for themselves. This isn’t working for me because without some simple, basic principles to follow I simply have nothing to do in the game. I’m just putting pieces down randomly, and that’s not fun at all.

So, to put it simply: you don’t want to play your pieces directly next to the opponent’s pieces right away, because your opponent has initiative in that kind of encounter, if only because you just used your move to put your piece next to his, which allows him to put another one there, resulting in a 2-1 advantage for him. That’s why you “attack” an opponent’s position by playing your first stone a couple of places away from the opponent’s. After this you have a stronger position to go into direct contact, as you can connect your contacting stone with the stone farther back if necessary, making it a stronger group.

A related point is that you don’t generally want to play pieces next to each other at the start, either: the long chains of stones that you see develop in the game only happen when players are competing actively for an area. When your opponent is not playing in a given area, you’ll want to just put your stones into influential positions close, but not touching, each other. The idea is to build a kind of a frame which you can then fill later on to defend the area when the opponent starts pressing there. The initial stones are enough to give you a defensive advantage, so it’s a waste of moves to play your pieces next to each other when just developing the position.

Go is about optimization, not tactics

I could discuss liminal profit analysis here, because that’s very similar, but I’m saying it the way a boardgamer would say it: go, even if it seems like a strategy game, is more easily understood as an optimization game. When you and me as boardgamers look at go maneuvers and understand nothing of what’s going on, that’s because nobody told us yet that the game is about optimizing your score gain for each individual move. I figured this out after three games or so, at which point the game started to make some sense.

Consider: go is a victory point -based game where you gain victory points by capturing opposing pieces and conquering territory (in the Japanese scoring system, that is) . When the skilled go players start the game by playing a piece here and another there they’re claiming territory for later in an effort to optimize their point gains. That’s why play jumps all over the board in go: often you are not so much trying to counter or develop a local situation, but looking to find the place where you can make the most points with that one stone. This might well develop into a sacrifice situation where you lose some points in one place to gain more points in another.

Go players, like chess, have special terms for this process: a player is said to have initiative or sente if his move was such that the point-optimal answer for his opponent is to tactically respond, trying to counter-act the move. When a countermove also has initiative we get the chainful play that seems more easily understood: players play stones next to each other in an effort to overwhelm the opponent locally. The corresponding term for losing the initiative is gote: if your move was such that the opponent finds a better liminal value in ignoring it and playing somewhere else on the board, then your move was gote. This might be the case if you make a mistake, disagree with your opponent about the value of the different situations on the board, or finish a sequence of moves in a way that makes it fruitless for the opponent to continue struggling against you. The key realization is that this process depends on score optimization analysis: your each move will either react to counter the opponent’s last move or develop your own situation independently depending on how many points you assume each option to gain you in the long run.


After figuring out the above simple principles go seems much more sensible to my eyes. If you, like me, have had trouble understanding what is actually happening, then perhaps the above considerations help with actually playing a functional game of go where you know what you’re doing. To finish, I’ll give a short and very crude checklist that reflects my current understanding of the thought processes needed for getting things done in go:

  1. Check to see if your opponent’s last move threatens to cost you points. Counteract its influence if necessary. Ideally your stones provide each other with some support, as you’ve played them close enough to each other to connect them when the opponent threatens you. You won’t actually do the stone connecting and area capturing in other situations: you only actualize the territory you have marked for yourself when you defend it against the opponent’s pressure.
  2. If you have the initiative, then you can act to improve your position: check to see if there are any corners of the board still empty. The corners are yummy and pointful, as it’s easy to defend a territory limited by two board edges. You can also check the four edges to see if there’s developmental possibilities there. What you want is to play your pieces into positions that make it possible for you to claim territory near the board edges, where it’s easiest to gain. If there is an empty corner, the move is pretty simple: play into the 3-3 point (three spaces away from the corner on both axes) to have a good basis for capturing territory in the corner. Alternatively, play at 4-4 to have more room for operating, but be aware that the opponent might decide to play closer to the corner to force you into attacking there instead of defending.
  3. If there are no obvious gains to be made in the above manner, then look into threatening and curtailing the opponent’s position: play some stones that threaten to capture some of his. This keeps the opponent from developing new positions. Ideally it will also allow you to improve your own position, as the stones you lay down not only threaten the opponent, but also provide you with bulk that can be used to capture area in the opposite direction.

I should note that the above is not some super-strategy. I’m pretty sure that a good player will balance all the above factors in a much more situational manner. But the above is good enough for actually playing in a purposeful manner, which is more than can be said about our first, stumbling tries at playing go. Which is a shame, as the game is quite fun after you figure out what you’re even supposed to be doing.


4 Responses to “Learning Go”

  1. Linnaeus Says:

    Useful stuff, Eero. Go just doesn’t suit the shape of my brain, but I have acres of admiration for it as a design. I’ve mostly settled for the lesser, but still great, GIPF series as my abstracts of choice, especially YINSH and the eponymous GIPF.

    One decent web resource I came across when I still thought I might become competent at Go was the Sensei’s Library. It’s a well organized wiki of Go information that seems to avoid the mysticism that you talked about at the start of your piece, and focuses on concrete tips.

  2. Kaj Sotala Says:

    http://www.wingsgoclub.org/books.php has some free Go books up for download. Way back, I downloaded a book that wasn’t on that list, about opening games. I wish I’d still have the link, because I seem to recall that it was a good one, and I’d like to refresh my mind about some of the principles I read from it.

  3. Eero Tuovinen Says:

    GIPF games are pretty fun, too. But not only do I lack motivation to own them myself, I would also have constant difficulties in finding players. Much more so than even with go, I suspect.

    Sensei’s Library is quite interesting, I found it yesterday and spent a long time reading it. There’s plenty of interesting advice on playing the game, albeit scattered.

  4. Go on a large board « Game Design is about Structure Says:

    […] in following Mikko’s 4-4 with my own 3-3, but I wanted to test out a premise from my earlier study: I should be able to build a viable corner group against a slightly inferior opponent even going […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: