Go Dreams

This page on Sensei's Library is an interesting and hilarious collection of Go dreams and other involuntary appearances of Go into the mind. Some highlights:

And finally, a dream from someone who has taken to heart the lesson that you shouldn't let your groups be surrounded, even if they are alive:


Myself, I've had Go dreams, I know. But I can't really remember them very well. Certainly I can't remember the specifics of any "technical" Go dreams, where the focus is on the actual moves made. I had one dream where a friend of mine became 2 dan, and another where I beat a 4 dan player on the Kiseido Go Server. In yet another dream, I imagined that I, as white, caused a go fight to extend "into the atmosphere", which seemed to be an analogy for "towards the center", to make the fight more difficult for black. And finally, I once had a dream where a decision I made in real life, but later screwed up, was like making a brilliant move in Go, but mis-reading the life or death situation afterwards.

Weird conflagrations between Go concepts and other concepts, like the thing with the alarm clock or the person waking up in atari, do happen to me, when I'm very sleepy, or have just woken up. But generally speaking, when I'm fully awake I don't think about everyday objects in terms of tactical Go concepts unintentionally. Certainly, a lot of high-level, abstract, "strategic"-level concepts will often find relations to Go in my mind (e.g. here and here), but stuff like the alarm clock feels more "tactical", and that only happens when I'm very sleepy.

Also, similar to the Sensei's Library contributor, I'll find that during extended periods when I'm spending a lot of time playing or learning about Go, Go problems will spontaneously appear in my head, seemingly out of nowhere, and not, as far as I can tell, based on positions I've ever seen. I'll often find myself unconsciously trying to solve these problems, in the background of whatever else I'm doing, as if a song were stuck in my head but I don't have to really consciously think about it to keep it there.

David No Insane

The most incredible thing just happened to me. I am NOT making this up.

There I am outside on my porch, trying to solve a nefarious problem in Life & Death, when I could swear I was hearing the theme song from Hikaru No Go, an anime about Go which I had downloaded and watched a month ago. I tried to listen closely, and I couldn't really tell if that's what I was hearing, so I just assumed that it was something on TV which sounded vaguely like it, or maybe someone was actually listening to the song, but not in a Hikaru No Go context (after all, it is a Japanese pop song in its own right). So I ignored it.

But then, about a half hour later (just enough time for one episode to finish), I hear the same thing again! This time I walk off my porch, and walk towards the sound. I hear, emanating from a neighboring house, the sound of Japanese dialogue, and then the unmistakable background music of a Hikaru No Go episode! So I grab my goban, and walk around to the front door. This Asian woman in her 20's greets me, and I explain what happened, and ask her, tentatively, "Is someone watching Hikaru No Go?" I guess my pronounciation is bad, because she doesn't understand me at first, but then she realizes what I'm asking and answers that she is in fact watching Hikaru No Go DVDs which she ordered from Hong Kong!

I tell her that I play Go, showing her my goban as evidence. She, it turns out, doesn't play Go, but she's just really into Japanese anime. So I offered to teach her if she ever wants to learn.

Kind of disappointing, but pretty amazing anyways! I mean, what are the chances of that?

Now excuse me while I try to collect myself.

Dominique vs. David Part 2: A Calm Opening, and then a Probe

« Part 1 | DGS Game Page

Ultimately, I decided to respond to Dominique's approach with the simple one-point jump at 4. We shared the corner with 5 and 6, and Dominique stabilized his group with 7. I then took the big side star point at 8.

What followed were two more approach patterns similar to the one that occured in the topright corner, the first one consisting of moves 9 through 13 (identical to the topright corner pattern except for the two-space jump at 10, which I felt worked well with 8). 14 is what's called a "checking extension", expanding the black group's base while limiting the white group's ability to extend at the same time. Normally, a two-space extension from an already settled group is considered too slow a move this early in the game, but because of its dual purpose, a checking extension is particularly valuable. Finally, white approached the bottomleft corner with 15, and after 19, the ball was in my court.

The question is: what to do? Well, before I get into that, first some comments about the opening so far. This is what you might call an incredibly calm opening. I'd even go so far as to call it a boring opening. To an observer, it looks as though there has been no struggle whatsoever; just a mechanical performance of standard patterns. Now, the truth of the matter is that on just about every move, I considered several possibilities, and spent a good deal of time reaching the decision I reached. I cannot speak for Dominique; I know his moves came quicker than mine, but I do not know how much total time he spent considering each one (it could just be that Dominique devotes all his energies for several minutes to decide on a move, whereas I think casually on it over several hours while I perform other tasks).

Regardless, this is where we ended up. How come? I wonder if this placidity is more common to play-by-email games, rather than real-time games. If so, I could think of an explanation for that phenomenon: when you have a lot more time to consider the possibilities, you are less likely to A, be overly aggressive, or, B, create vulnerablities in your own formations which invite your opponent to attack.

In any event, to decide move 20, I had to consider the actual state of the board. Black has about 40 points of potential territory on the left side (though it's not completely safe from invasion or reduction just yet), and about 15 points on the topright, for a total of 65. White has no more than 25 points of secure territory, but has total dominance over the bottom and right sides of the board, plus 4 points of komi. It seemed like my objective should therefore be to try to establish a presence somewhere in white's framework in the most delicate way possible, so that I don't create an easy target for attack which will allow white to solidify large swaths of territory in the process.

Diagram A

Partly to achieve that goal, and also to try to stir up some excitement in this otherwise boring game, I played the tsuke (attachment) at 20, as a "probe". Probes (of which attachments are often an example) are moves which force your opponent to respond, but where it's not clear how he should respond, and where that response can have a relatively large impact on the future course of the game. Sometimes the player who played the probe will immediately follow up, or sometimes he will leave the situation as it is, to serve as a source of complication and worry for his opponent later on.

Any decent Go player's first instinct against a tsuke is to play a hane (pronounced "hah-nay"; meaning diagonal wrap-around), at 2 (or a) in Diagram A. Occasionally, it is better for white to respond to the tsuke with a simple stretch at b or c, to avoid complications, but that is usually a submissive way to play, giving black an easy time. The hane is more severe, taking away a liberty from black's stone.

Diagram B

In the actual game, Dominique mainly had to choose between a and b in Diagram B. In this case, a seems inefficient, given the proximity of the white stones on the left. Such a play would allow black to easily stake a presence in the wide stretch to the right, supported by 20. After a, black could play b, and now white has a cutting point above a to worry about, while black can create a base for himself.

So Dominique played 21 at b. Yet, there's something awkward about the shape doing this makes. Certainly 20 is now very weak, but it can be treated as a throwaway stone by black, and white's formation has something to worry about as the game progresses.

Diagram C

The question now arises for me: do I leave this situation as it is, and invade white's framework elsewhere, or should I immediately follow up? If I do follow up, how? One possibility is to make life on the side. This is undesirable, because I will end up with a very small amount of actual territory, much of it on the second line, in exchange for giving white a powerful, thick wall on the third and fourth lines. Another possibility is to force myself through to the left, cutting off two white stones in the process, such as in Diagram C. However, careful reading reveals that such an attack would be doomed to failure; after white plays the double-hane at 29, black's three stones have no escape.

A remaining possibility is to treat 20 "lightly" — i.e., to not be so worried about the survival of that stone per se, but to use its presence to achieve something else in the area. There are many potential ways to do so, and I'll spend some time thinking about them, but for now I haven't decided.

« Part 1 | DGS Game Page

DavidSJ’s Favorite Go Resources

For those just getting into Go, or those already playing, here are my favorite Go resources:

Introduction

• Go: Life Itself — This article does a great job of explaining why you should be interested in Go. It doesn't delve at all into the tactics or strategy, but it's an engaging primer on Go history and the current Go scene. Reading this is what lured me into the game.
  
• The Interactive Way To Go — This step-by-step online tutorial is probably the best way to learn the rules quickly and concisely.
  
• The Book of Go — This book is a nice introduction to the game, including basic opening theory and full-board strategy, and it also contains a portable cardboard 9x9/13x13 goban.
  
• The Master of Go — A classic novel, calm and contemplative in tone, which does a terrific job at conveying the Go feeling, as well as the clash between the old and the modern in Japan, as manifested in a famous 1938 title match.
  

Getting Strong

• Graded Go Problems for Beginners — Each book in this four-volume series contains hundreds of problems, mostly on (but not limited to) life and death. It is absolutely vital to solve problems like these to build up your tactical reading ability.
  
• Lessons in the Fundamentals of Go — Widely considered (including by yours truly) the essential reading for any Go player. Made a huge impact on my game when I first read it over a year ago (five months after I started playing). Re-reading it now 12 stones stronger, I am uncovering a whole new dimension I could not appreciate the first time around. Useful at any level, from a kyu who has been playing a couple months, to experts in the dan ranks.
  
• Sensei’s Library — An open-source Go encyclopedia; easy to get lost in. Contains a huge volume of information on just about every subject relevant to the game.
  
• GoProblems.com — For those who don’t want to spend money on books, this library of Go problems of all skill levels is almost as good.
  

Places to Play

• Kiseido Go Server — Very sociable, and great teaching tools. The best place to send new players (since it doesn't require installing anything), or to have a lesson. If you play there regularly, you should download the CGoban client to your computer.
  
• Internet Go Server — This server has more players, so it's easier to find a game, and the atmosphere is more serious and competitive. This is where I play most of my games. If you've got a Mac, Goban is a great IGS client. There's also glGo, which is a very good cross-platform IGS client.
  
• Dragon Go Server — This server is designed for “play-by-email” games. You get an email notification when your opponent moves, and you can make your move by visiting the web site. Great for playing a year-long game in the background of your life.
  
• Local Go Clubs — The American Go Association’s list of local Go clubs and chapters in the U.S.
  

Kenny vs. David Part 2: A San-San Invasion, and a Slight Mistake (Maybe?)

« Part 1 | DGS Game Page

The sequence I predicted in Part 1 panned out, and Kenny used sente (the initiative) to approach my bottomright corner with move 11. Taking into account my stone at 10, I felt comfortable playing a tight pincer attack two spaces to the left of 11, at 12. Kenny then used his approach stone to assist in an invasion of the san-san (3-3 point), with move 13.

After a 3-3 invasion, the owner of the 4-4 stone typically has a choice as to which side to block on. If the surrounding situation is symmetrical, then it doesn't matter which side, and in fact it may pay to delay the decision until you have more information. On the other hand, if the situation is assymetrical, it is not always obvious which side you should block on. To understand how to make that decision, it's first important to understand what happens when there are no stones around at all.

Diagram A

After white invades with 1 in Diagram A, black blocks above 1 at 2, and pushes white down to the second line from 3 to 8. White then pushes out with 9, and both players protect their shape with 10 through 12.

You may say that white has taken black's corner, so white is better off, but the reality is anything but. In exchange for white getting roughly 10 points of territory, black has gotten a powerful outer wall which is worth much more than 10 points. It can be hard to determine the value of a wall, and harder still to utilize a wall to get your rightful value from it, but both are vital skills in the game of Go.

So what exactly is a wall like this for? For starters, it's not for making territory. If you try to start building the three other edges to complete an enclosure of some area, you will end up spending many moves for a relatively small profit. It is much more efficient to make your territory against the edges and corners of the board, as you already have natural foundations to build against there. It is rare in Go for a substantial amount of territory to be made above the fourth line.

Diagram B

The best analogy is perhaps to a minefield. If your opponent's weak groups end up near a wall, they will have to run the other direction to live. As I've remarked before, you probably won't kill, but you will have a favorable fight. Power limits your opponent's options, and when your opponent's options are limited, you can profit.

In fact, that is the very purpose of playing on the 4-4 point to begin with. If black wanted the corner as territory, he would play the 3-3 point. A play at the 4-4 point is not intended to make immediate territory, but rather to create a dominant presence in the corner, so that the fights which follow are favorable to your side. If white is going to demand the corner from black, he will get it, but he will pay a high price.

Now let's take a look at another symmetrical situation. The butterfly formation in Diagram B is a great target for a 3-3 invasion, because black already has a substantial amount of outward influence. There is no way in which black can attack the invading white stone to build up a substantial amount of power that it does not already have. Sure, it'll get a little bit of additional outward strength, but much less than the territory white will gain. As such, the 3-3 point in Diagram B is a big point for whoever takes it first.

Diagram C

However, if white were to try the 3-3 invasion in Diagram C, black invites the opportunity to block to the left of 1, at 2, cutting off 1 from the nearby white stones. After white lives in the corner, black has a destructive invasion at a, assisted by the firmness created by 2. Without 2, an invasion at a isn't so appealing, so white has created problems for himself by allowing black to get thickness exactly where it's useful.

On the other hand, if white already had a stone at b, then its group to the left would be strong enough not to have to worry about a black invasion at a. In that case, if white plays 1, black would do better to block at c, letting white connect his groups by playing at 2 himself. After this exchange, black feels somewhat clumped up and inefficient, with his two-stone wall so close to the marked black stone above, but at least he gets something out of it. If he blocks to the left, at 2, then white gets more of the corner, and black gets a completely useless wall, because of white's stone at b.

Kenny vs. David (14 - 20)

Thus was my reasoning when I played 14. My intention was to separate white's two unsettled stones. Because of my stone at 12, 11 was already weakened somewhat, and could be treated by white as a sacrifice stone (with a bit of leftover potential even after it's sacrificed). So, after 14, it's perfectly reasonable for white to push upward with 15 at 16, instead of playing 15 where he did. However, white chose to first threaten the connection to 11, by playing 15. At this point, I decided that a connection to 11 is rather unimportant, so I blocked at 16.

This is where I believe Kenny made a mistake. The normal response to 16 is to play at 17 in Diagram D, which serves to both connect to 11, and to threaten to move into the right side somewhat. If black were to try to cut at a now, white could safely play b, which would leave black's cutting stone no room to escape. Without 17 in Diagram D, black could play c after white b, capturing 15, and white would be devastated.

Diagram D

Instead, Kenny played 17 directly to the right of 13, allowing me to cut off 11 with 18. White had no choice but to protect 15 with 19 (otherwise his corner would die), allowing me to link up to 12 by playing 20. White's 17 and 19 are an inefficient combination, serving somewhat redundant purposes, whereas a play at 17 in Diagram D would have most of the same benefits, in addition to linking up 11.

As a result of this exchange, I got the best of both worlds. I effectively managed to block white off on both sides, and connect to 12, leading to a position qualitatively similar to Diagram A, which heavily favors black. In this case, white's territory in the corner is a bit smaller than in Diagram A, but the power black receives is a bit looser. On the whole, I am very happy with this result, and I believe this game is looking up for me.

« Part 1 | DGS Game Page

Kenny vs. David Part 1: Shimari

DGS Game Page | Part 2 »

Well, Steven Den Beste declined my offer for a game, and Vladimir changed his mind about playing me. However, I did get a game with a guy named Kenny. Kenny's one stone stronger than me, so I'll get two handicap stones, like with Dominique, and the komi will be 6.5 (normal komi).

For some variation, I placed one of my handicap stones at the 3-4 point, instead of the 4-4 hoshi point. Kenny took one of the remaining hoshis, and I took the last one.

The purpose of playing the 3-4 point is to build a shimari, or corner enclosure, with a second move. Here are some examples of shimari:

A

B

C

Shimari A, the knight's move enclosure, is by far the most common, and it completely secures the corner as territory, as well as providing a bit of power for extensions along the edges or for fights with opposing stones.

Shimari B, using the 6-3 point to make a long knight's move, captures a slightly larger corner as territory, but much more loosely. If later, white plays a stone in the vicinity, black may have to play a followup move to protect his corner. What this enclosure lacks in firmness, it somewhat makes up for in flexibility. Black may during the course of the game be presented with an opportunity to trade the corner to white, in exchange for power on the outside.

Shimari C, the one-space jump, leaves the corner open for invasion by white, but emphasizes power along the left edge. This may work well if black is building a framework along that edge.

There are many other kinds of shimari, some of which do not involve the 3-4 point at all, but these three are perhaps the most common. Making a shimari is a very big move, and it is urgent for black to do so as soon as possible, or for white to prevent it if he can. Thus, once all the corner star points were taken, Kenny immediately played a kakari against my 3-4 stone, at 3, to prevent a black shimari. This is the most common approach to the 3-4 point, though there are several other reasonable options. Below is the sequence that normally follows such an approach:

First black plays an attachment with 4, to claim the corner for himself. This is important because corners are particularly inexpensive areas to build territory and eyes, so black wants to make sure that white does not steal it. White plays 5, to protect 3, which causes black to pull back to 6. White must then protect his cutting point with 7. This move is absolutely vital; if omitted, black will cut at 7, tearing white's formation to shreds. Finally, black plays a stabilizing extension at 8, and white does the same with 9. Both players have gotten a fair result here, given what they put in. If white wishes, he may play 7 at a, which also protects the cutting point, and he may play 9 at b, emphasizing central influence over edge territory.

I played 4, expecting a sequence like this one. Since the sequence ends with a white move, my follow-up plan is to use sente (initiative) to take the marked hoshi on the right side of the full-board diagram, building a large, flexible moyo (framework) along the right edge of the board. This point is bigger than any of the other hoshi points, because of the way it works with my two existing stones on that edge. A play there by white would be just as large.

Will the next few moves be as I predicted? The coming days will tell.

DGS Game Page | Part 2 »

Dominique vs. David Part 1: A Kakari

DGS Game Page | Part 2 »

As expected, Dominique (white) took one of the remaining corner hoshis. I took the last one.

With move 3, white then played a kakari (approach) against my stone in the topright corner. The question is: how should I respond to this? Points a through h in the diagram below show many of the common responses to this move.

a through e are what are called pincer attacks, assaulting the enemy's stone from the opposing side. White would really like to play an extension at b or c, to build a base against the edge of the board which can be used for making territory, as well as making the eyes necessary for life. The purpose of a pincer attack is to deprive your enemy's stones of such an extension. Now, this does not mean that a pincer attack is an attempt to kill. It is very hard to kill groups of stones in go, and though it does happen, even in professional games, it usually happens because a player decided that it is too expensive to make a group survive, not because they were incapable of doing so. A pincer attack, therefore, is an attempt to increase the cost of survival for your opponent, as well as to limit its territorial potential along the edge.

Of the available pincer attacks, a and b are somewhat calm, whereas something like d or e are quite severe, and immediately start a serious fight, and c is perhaps somewhere in the middle. If black is going to play a pincer here, he must temper his aggressiveness with knowledge of the marked white stone below, which may have an influence on a fight on the right edge of the board, so he should play one of the calmer moves.

Alternatively, black can play a simple extension at f or g. These are the most common responses to the kakari white has played. f, an ikken tobi (one space jump) extension on the fourth line, emphasizes influence and power over the rest of the board, whereas g, a keima (knight's move, as in chess) extension on the third line, emphasizes building territory along the top edge.

A move like h is best reserved for when black has supporting stones in the area. It is the most severe attack on white. White will almost invariably respond with a simple extension one point to the left of 3, and then black must follow up with an extension at f or g, to stabilize his group. If he does not, white can play a very severe attack on black one point to the left of his hoshi stone. While black will survive in the corner, his group will be shut in from the rest of the board, and white will gain power which can be used in subsequent battles. After black stabilizes his group with f or g, white can now play a longer extension than before, thanks to the additional stone black invited him to play to the left of 3. As such, this is a better result for white than he deserved, unless there was already a black pincer stone at, say a (preventing such a long extension by white), or a black stone near f or g, which would make stabilizing the black group less urgent.

DGS Game Page | Part 2 »

Playing Go

I have just started two games on the Dragon Go Server, a web-based Go interface that allows you to play games with very long time limits. The games I'm playing allow roughly one day per turn. So these will be games I'm playing in the background of my life. I'll be able to stop every now and then to think about my next move. I have also challenged Steven Den Beste to such a game. Hey, I can dream.

I am going to provide running commentary here on the games I play on DGS. I will try to make the commentary accessible to people who know little to nothing about Go, while at the same time provide interesting analysis for people who are somewhat skilled at the game. Hope you guys enjoy it!

In my game against Vladimir, I am playing someone who is two stones weaker than me, so he gets two handicap stones. I am waiting to see where he places them.

In my game against Dominique, the player is slightly more than one stone stronger than me, so we can use the great "proper komi/handicap" feature of DGS to make a fine-tuned handicap. I get two handicap stones, but he gets 4 points of komi. If this were an even game, black would go first, and white would get 6.5 points of komi (compensation for the disadvantage of going second). With two handicap stones, I get two moves at first, instead of one, but his komi is reduced by 2.5 points, roughly 1/5th of a move, so it works out to me getting roughly 1.2 handicap stones.

I have placed my two stones on opposite corner hoshi (star) points on the goban (board). Generally speaking, you want to stake out large areas at the beginning of the game. By taking the stars, I have made a dominant presence in that corner, but I have by no means captured it all for myself. Usually it is best to play on or near the corner star points in the beginning of the game. Most likely, Dominique will play near one of the two remaining corner star points, but it is possible he will play a kakari (approach) on one of my stones instead. We'll see what happens.