I got caught being wrong on rec.games.go. I hate that. Usually, I know when I'm confused. The situation involved a double ko, a concept which gives me trouble.
Double kos are funny, because they are not really kos. A garden variety ko occurs when the outcome of a local struggle becomes undefined, through the occurence of a loop in the local move tree. If the battle is important, the game would not end without some kind of move restriction (and if the battle was not important, would the players be fighting it?). The usual restriction is that immediately recapturing a single stone which captured a single stone is prohibited -- the simple ko rule. The prototypical outcome is a ko fight: you'd lift the prohibition by making a move elsewhere, giving your opponent the option of answering your move or resolving the ko.
Double kos are different. There are at least two different kinds, both of which are locally resolved situations: the struggle is over, in the sense that they do not couple a local status to a global ko fight. There are at least two kinds of double ko. One is the double ko seki.
In the double ko seki, both sides have two liberties but not two eyes, and one liberty for each group is in a ko. The two kos act as flip-flops: capture one, your opponent will capture the other. Both sides live, but both sides can use the situation for an infinite number of unremovable ko threats.
There are two variants of the double ko seki: in one case, the extra liberty comes from an eye on each side, in the other, the liberty is shared.
The other kind of double ko occurs when one side has an eye on the other does not. The side without an eye (black in this example) cannot win, or even live; but he can use the situation for an infinite number of ko threats. White can remove black's source of ko threats by spending a move capturing.
Funny things happen when multiple double kos occur on the board; and the outcome may depend on the rules. I won't get into that, I've made enough mistakes already.
Is go an art, or just an arbitrary competitive game? In most arenas, "is X art" is an uninteresting question: art is whatever artists do, and even posing the question is dreadfully retro. There are go players who regard themselves as artists, but competitive sports and skilled trades are also valid analogies. Pros get paid for results.
The association of go and art is old: the Chinese considered it one of the four gentlemanly and scholarly accomplishments, along with music, calligraphy, and painting, things we would normally consider art. And certainly there is an esthetic sense associated with the patterns of stones on the board, although such "good shape" is highly coupled to what moves are effective in the context of the competitive game. And there is a rich and ongoing tradition of go in art, ranging from traditional paintings from China and Japan to Blake Haber's pop-art videos of marching stones. One might complain that such works do not accurately represent the game as it is played, but this is misdirected: these works are art using elements of the game, not illustrations in an instruction manual. Analogously, consider Susan Rankaitis's drawing on scientific imagery in her works. Susan's works are not science, or even scientific illustration; they are (just?) art. The game of go itself is separate, its own entity, not needing to be an art, or a science, or nonverbal communication, or anything other than what it is. "Chess is chess" say the chess players, and so it is.
Who is surrounding who is the first question when looking at a capturing race situation. The general case will result in a graph with nodes of two colors, no two adjacent nodes the same color. (Not every such graph will be realizable on the go board. The set of realizable graphs might be the same as maps which can be colored by two colors.)
After spending some effort (in my May 23 entry, really need to set up some kind of archiving and linking) trying to understand who is surrounding who in a relatively straightforward situation from a real game, I got it wrong. (And no one corrected me, confirming my suspicions about my lack of readers.) The bottom white group and the top black group are in fact adjacent, right at the crosscut. (Duh!) Deleting inactive groups, here's the correct structure:
- | | -
Four active groups is the lowest number where two structures are possible, five has three structures. Here are the structures for less than six groups:
- - -
- | | -
- - - -
- | | - -
----- | \ | | | | \ | -----
Arno: your suspicion about "few readers" is groundless :o) I did a quick check in the database and there are only two pages created after May 1st that have more hits than this page (for reference: over 300 pages have been created since May 1st). Currently you have about 30 hits (readers) / day.
Fujisawa Hideyuki, the hard-drinking self-proclaimed greatest artist of the game of go, had a reputation for winning the first game of major titles. He became the first Kisei in 1977, and defended it five times, a period when he was famous for only winning four games a year, about a week of work, and partying the other 95% of the time. This is enough, if they are the right games. You might call it efficiency, you might call it laziness. I'd call him a slacker, in the very best possible sense of the word.
In March 1978, though, things were looking grim. He'd gone behind 3-1 against a Killer Kato maturing into his peak. Down 3-1 is a deep hole in a seven game series: your odds of coming back are 7:1 against if you're evenly matched. In practice, it's usually worse, since going down 3-1 may mean you aren't quite evenly matched. I don't know what he was thinking, but the gravity of the situation must have been clear.
I wouldn't be telling the story if it didn't have a happy (or at least a snappy) ending. Here it is, the final position of game five, a spectacular capturing race which won the game and may have turned the series around. Fujisawa won the next two as well, and remained Kisei until Cho Chikun turned the tables on him five years later, coming back from a 3-0 deficit to become the second Kisei of the modern era.
Let's look at the position. The group is surrounded: if W connects at a, something like b should suffice to keep it hemmed in. Six separate groups of black stones neighbor the group, though not all of them are active players in the capturing race. First, illustrations of being surrounded: the two S6 stones, neighboring only the group, have three liberties, but only one second or third order liberty. They are surrounded, eyeless, and dead; played to remove an eye, an example of what Dave Dyer has called an intron. The four black K10 stones, on the other hand, don't have many more liberties: only four, not enough for tactical stability, and less than either of the two white groups which are players in the capturing race. However, they are not surrounded: five second-order liberties, four third-order, six fourth-order, ten fifth-order, the increasing numbers indicating space to run. Higher order liberties aren't quite a cookbook, eg. consider the group, with 2, 3, and 5 liberties of the 2nd, 3rd, and 4th order. If it did get out, it could start to run, but reading is necessary to show that it can't escape.
So, topologically, from bottom to top we have (NB: Wrong! see correction 25 May)
where each group is represented by a stone and adjacency is represented by a hyphen. The active players are divided by a single crosscut, at N/O 11/12. This is about as complicated as it can get with a single crosscut, which by definition involves only four groups. Greater complexity is possible, of course: imagine that white M11 stones were surrounded, bringing the M/N 11/12 crosscut into play. I won't try to enumerate the possible outcomes of a four-group capturing race, suffice it to say that it's more complicated than A dies, B lives; vice versa; seki; or ko possibilities for two groups.
Before I leave this (and yes, I know I haven't counted the liberties) I'd like to show how the players got themselves into this mess.
When Black completes his framework with , White must do something big. To my double-digit kyu eye, White looks way behind, but then kyu players are notorious for overestimating the value of a framework. White goes in very deep with . With the title on the line, he'll not be accused of lack of fighting spirit. It's not necessary that live, of course: as long as something good happens somewhere, whether a reduction or successful invasion elsewhere, capturing a few stones, etc. The outcome of this invasion isn't determined until 87 moves later, in the diagram above.
 Counted using a strict nearest neighbor definition, better might be to include liberties gained by joining up with friendly forces like the H10 stone, but avoiding reading is impossible in any case.
Capturing races play an important role in basic tactics, but they have some subtleties associated with them. The first question has to be one of definition: what is a capturing race. Given the trouble I had understanding a reasonable definition for seki (see my May 3 entry), which is, after all, just a particular outcome of a race, you won't be disappointed that I don't define a capturing race very precisely. (As an aside on seki definition, Berlekamp and Wolfe need to define seki and life for their discussion of the differences between rules. Unfortunately, they screw it up, and end up only considering sekis with two shared liberties per string, missing hane sekis (with one shared liberty) and perhaps other kinds of seki with unusual numbers of shared liberties. Evidently, even understanding the terminal position of a capturing race is tricky.)
Definition aside, we all know what a capturing race is. They happen when you've surrounded and are quite close to killing an opponent's group. You've made sure it hasn't enough eyes, but it has a fair number of liberties, and, while squirming around, manages to cut off some of your stones. No problem, until you realize that your stones are surrounded too, and (oh no!) don't have as many liberties as the group you almost killed. This is usually massive, not some small local loss: almost without realizing it, you staked the game on killing a group, and got killed instead. If you aim at a king, you must aim to kill.
Let's think about the individual steps involved. If a group is surrounded, that means that it has no where to run away. Running away has multiple effects: it's harder to surround; it has more liberties, so it will take more moves to remove from the board; and it has more tactical possibilities -- somewhere in the combinatorial explosion of possible moves and responses may be a path to connecting, making life, or capturing some opponent stones. So surrounding the group is the first step. Intuitively we know what it means for there to be room for a group to run. Those who want to quantify it (no doubt the same types fond of rules like "Five liberties for tactical stability") may find the concept of higher order liberties useful. Normal liberties are empty nearest neighbors of your group. Second order liberties are empty next-nearest neighbors, or the neighbors of your first order liberties. Third order liberties are empty neighbors of second order liberties, and so on. While normal liberties count how many stones it would take to capture you, higher order liberties track how many more liberties could be gained by running away. They've been used in computer go, but the concept, at least, seems more generally applicable.
Not all surrounded groups are dead, of course. Eyes are the next question. Making and destroying eyes for surrounded groups defines the very large topic of life and death. If the group has two eyes, it's not an active player in a capturing race, though it might be helping surround an active player.
To write down my weak definition, a capturing race occurs when surrounded groups without two eyes are adjacent, so that one might live by capturing another. The simplest case is two opposing adjacent groups along a boundary, with a single crosscut separating them from friendly forces, but any number of distinct groups may be involved. Analyzing multigroup capturing races can get pretty complicated. Even classifying possibilities of who's surrounding whom gets complicated: phase boundaries between groups extend to edges or crosscuts, delineating groups which may or may not be active participants in the race. And you really need to understand who the players are and their relationships before going into counting liberties, real vs. virtual, internal vs. external, big eyes, small eyes, and no eyes, single point kos, etc.
This is all pretty abstract, I'll admit. I do have an example in mind, a 1978 Kato-Fujisawa game which ended in a capturing race taking up half the board involving three or four active groups. Hopefully I'll get to it soon.
I spent a summer playing ping-pong (or table tennis, if you prefer). My boss wanted a sparring partner, and he was my boss, so... we played everyday at lunch. He was fairly serious, and I was a novice, so he only taught me to block. Hold the paddle here, tilt it forward a bit, and try to get it in front of the ball. It was much harder than it sounds, but by the end of the summer, I was getting pretty good. Sometimes we'd have a rally several long: hit, block, hit, block, etc. Thrust and parry, kind of similar to sente and gote.
Being on the attack in ping pong is not easy either. You're trying to put the ball away, so you're reaching for the limit. It's pretty easy to miss the table. And if you hit the table, most of the time, with some skill on the other side, the ball with come back to you, 1/10 of a second after you hit it, with 95% of the speed and spin you put on it. You can't just block, or you're dead: you have to put it away again. And again. And again. Attack and defense.
When I play through pro games, that's what I see predominate: threats (usually threats to cut), and responses. Both sides must keep up the pressure. It reminds me of ping pong. Sente and gote.
Nomenclature is a topic of ongoing interest here, which plays out in many ways. Sometimes, it takes the form of discussing which language to steal go terms from, or how pages should be named. Other times, people are interested in how to (and not to) classify and organize. Nomenclature is important: if it has a name, know it says the proverb, with the corollary if it has no name, christen it. It's probably overreaching to claim, as some linguists once did, that one cannot conceptualize what one cannot name, but names certainly help. Debates over nomenclature are sometimes signatures of substantial disagreements about concepts or philosophy: consider, for instance, recent debates in biology as to whether to redo all taxonomy based on new information, not available to Linnaeus, coming from large scale genomics. The question is whether the term "E. coli" is simply an arbitrary convenient term for a common bug or whether the term encodes semantic content and must therefore be subject to alteration based on new information. Both positions are true, which makes for an interesting debate. Which is more important, the descriptive, semantic content of the label, which must change with new discoveries, or the past and current usage of the label, which is by definition unchangable?
In the case of joseki, those sequences of moves which are repeated in game after game, the question of systematic classification is straightforward: move sequence choices have a natural tree structure. The leaves are the resulting pattern, which (modulo move reordering and symmetries) uniquely identifies the sequence. Any way of describing this paths or leaves of this tree, whether by name, SGF file, coordinate list, or agglomeration of Japanese derived words, is mathematically equivalent. So the choice might seem to be purely a question of context and taste, without semantic consequence.
And so it is, but there are still those questions of context and taste. To introduce another analogy, consider drug names. Molecular structures of chemicals are (mostly) well described by the valence model, which is equivalent to the standard cartoons of atoms and bonds. A mathematician would see an undirected graph, with an atom name at each node. All such structures can be systematically represented through the drawing itself, formal IUPAC names, or other formats such as SMILES strings.
Though chemical names are systematic, and any chemical with a known structure can be named, they can be lengthy, and are not very convenient for talking about drugs. Nor is it easy to draw the structure everytime a doctor wants to write a prescription. The pharmaceutical industry deals with this by naming sufficiently important new chemicals. Names must follow a number of rules (short, pronouncable, not too similar to any existing name, etc.) and are vetted by a committee. There's a bit of structure, so names aren't entirely arbitrary -- suffixes usually indicate the class of drug; i.e. if a drug ends in -olol it's probably a beta-blocker, and -icillin means something related to penicillin. However, lots of traditional names are grandfathered in too, so the scheme is certainly not systematic. And there are certain differences between countries.
Such a scheme could certainly work for joseki. Many joseki already have short, pronouncable, often evocative names: taisha, avalanche, magic sword, drooping lotus, and so on. These names can be kept. Naming the rest is entirely doable. For kyu players there are perhaps only dozens of joseki, if that, and extending to all sequences of interest to dan players would probably not extend beyond hundreds. Even naming every joseki in print isn't unreasonable: there are only thousands (perhaps ten or twenty) which is the same order as the number of drugs or the number of genes, all of which are named. True, some of the names get rather silly ( Sonic hedgehog? What does that gene do? When it turns on, do you turn blue, spin around, and start flashing?) but that's okay -- it just makes them easier to remember. No one said nomenclature has to be boring.
Tamsin I agree that naming things is useful for learning them. Labels help to tie concepts together in your brain.
It would be really helpful if joseki all had names, as you point out. The drawback is the difficulty of getting people to agree on a new name for any nameless joseki. As a result, these joseki tend to be referred to as a sequence of moves (e.g., 3-3 point invasion under star point with small knight's move extension). This is not ideal, but what can you do ? This is perhaps one area in which chess has the edge of go: chess openings are usually named after the people who invented/pioneered them or after places (e.g., "Ruy Lopez" and "French Defence"), and their variations are similarly labelled (e.g., "Yugoslav Attack in the Sicilian Defence"). Some variations have special appellations, such as the "Fried Liver Attack in the Giuoco Piano". Because all these names are evocative, they do aid the memory considerably. Perhaps you could make up a private name for each joseki you learn? I sometimes think of certain sequences using names too colourful (actually too blue) to print here :-)
Absolutely. Now if we start sharing our private names and using them in public, pretty soon we'll have named them all. See Named Joseki.
Charles The problem is in a sense bigger - but less forbidding. Get go players to articulate what it is they're up to, and the names will follow, somehow or other.
A few weeks ago, I visited a local museum which had an exhibit on ancient Rome. One exhibit in particular caught my eye. There were, in the corner of a glass case, a small collection of what to me were obviously go stones. They were made of colored glass, and had the flat side common to Chinese stones. This style is easy to make, just drip molten glass onto a flat surface and let it harden. They were no more irregular than the stones of the same style sold by craft stores, intended for making flower arrangements but widely abused by go players for another purpose.
They were labelled as ludum latrunculatum (approximately), and described as "pawns for playing a game like chess". Like chess? How much like chess? So where are the rest of the pieces? And what rules, exactly? Sometimes historians use the word chess for games that no modern chessplayer would call chess. Go was certainly being played elsewhere in the world in that time period, but I'd thought it wasn't introduced to the West until the 19th century. The history books might need some rewriting if go was played in Rome but lost during the dark ages.
There was another, interactive exhibit for the museumgoers to play the game, with a largish square grid, 9 lines by 9 lines (8 squares x 8 squares), and a collection of two colors of stones. The posted rules claimed to describe the same game, but were the rules for five-in-a-row, with pieces placed in the squares. So it was clear that the museum didn't have a consistent view of the game, which wouldn't be surprising if all they had to work with was a few pieces of glass and perhaps a picture or two.
At that point I decided that I had enough basis for an amusing crackpot theory: the Romans played go, either through cultural diffusion or independent discovery. In writing this, though, I did a quick web search, and found much more evidence. Take a look at this description of Latrunculi: sets of identical pieces are of two colors, popular thinking man's game, played on various sizes of board (evidently not affecting the basic nature of the game). As to the rules, here's what the Onomasticon says: The game played with many pieces is a board with spaces disposed among lines: the board is called the 'city' and each piece is called a 'dog'; the pieces are of two colors, and the art of the game consists in taking a piece of one color by enclosing it between two of the other color. Dude, that sounds like go. It's not clear if the capturing rule is identical, but the same site mentions other writers referring to multiple stone captures: You win and both your hands rattle with the captured group -- not something that could describe any form of chess, checkers, or even Pente. And, as final proof: look at the pictures on the vases, on that page and the Petteia page: they hold the stones exactly right. Sai would be proud.
They could well be playing go.
I have a few go books in chinese. Most are relics of a visit to Hong Kong, where I found a bookstore with a section of go books and bought a few. (I also visited the go club, a story for another time.) More recently, I mail ordered a Chinese translation of the Fujisawa tesuji dictionary from Yutopian.
Before getting these books, I could say about two words in Chinese ('ni hao') and read none. The game record books are easiest to work with, especially if you don't care about details like who's playing, when, or what the commentator thinks about the game. Still, I learned the characters for Black (黒, which looks like a TV set on feet, or maybe one of those aliens in Space Invaders) and White (白, which looks like an accented present). These came in handy in the problem books. You can treat each problem as a status problem, and you'll probably get stronger that way, but it's easier if you at least know which side is to move. I still can't figure out which diagrams are right and wrong from the Chinese, for me it's still easier to look at the moves in the diagram itself, and guess.
The tesuji dictionary is a much different story. There are no problems. It's a little more than 50% text by space, and by skipping the text you're skipping much of the content: the categorization and commentary of the examples. I wouldn't have bought it at all had not John Fairbairn posted a translation of the table of contents of Volume I to rgg, plus extensive notes on the first few sections, and spoken glowingly of it. I've been laboriously working on understanding a bit. It's an entirely different challenge than playing go, more akin to decryption. I've been taking notes in the book, referring to the various chinese and go term dictionaries running around, bothering chinese friends, and maintaining my own list of characters. I've learned enough to understand the chapter titles of Volume II (all four of them, requiring a whopping eight characters of chinese) and the subsections of the shortest chapter. At this point, I can probably recognize a good half-dozen characters on sight, and think "That one's in my list somewhere" for a couple dozen more.
Is it a good way to get strong at go? Um, I don't think so, not at my level. Is it a good way to learn another language? Probably not. My wife commented, "You'd be real useful on a trip to China. We'd need to find somewhere to sleep, and all you could say is 'Your stones are in atari!'" But I'm having fun, and fun is good.
I updated my web site, and now there's a picture of my shelf of go books. What such a lousy player is doing with so many books is beyond me. Apparently, I prefer reading about go to actually playing and getting better. It's a hobby, so I can do what I want, but it does show well that just reading about the game makes little difference to strength of play.
There's a story associated with each of these books. I find go books have a high information density, in the sense that the amount of time I spend staring at any given page is higher than just about anything else I read. Some of the books I've only leafed through, others I spent so many hours with over such a long time that they are firmly linked to certain periods of my life. There are other books like that too: some academic texts, and A Suitable Boy, which I read aloud, all X,000 pages, to my wife when we were first married. The story, which revolves around a year in which a girl is selecting among three suitors, unfolded in pretty much real time, the seasons changing outside the story at the same rate as they changed inside. It was wonderful.
A bubble tea house opened up in my part of town not long ago. A coworker reported that he'd seen a bunch of people in there, playing go, chinese chess, etc., so I had to check it out. I had visions of a secretive smokey club like the ones in Hikaru No Go or First Kyu, filled with old men, gangsters and players, a weak public echo of some cryptic go-playing substratum hiding beneath the surface of suburban life, far beyond the cozy little club I've visited once or twice.
It is nothing of the kind, of course. I went there for lunch with another coworker yesterday. It was clean, well lit, tastefully decorated out of an Ikea catalogue, and completely empty until we arrived. The menu read like a syllabus for an upper division seminar in Asian popular beverage culture: China, Hong Kong, Vietnam, Thailand; all sweet, often with little floating tapioca beads (the bubbles in "bubble tea"), all nonalcoholic, all expensive. In addition to six pages of drinks, you can get noodles. We selected our drinks and ordered noodles.
As advertised, on a shelf near other amusements was a go set, with a plastic mat, flat plastic Chinese stones in faux wicker bowls. My friend seemed game enough, so we played, his first game. I tried to not drop the stones in the soup too many times, but these things happen. Suddenly, plastic paraphenalia makes sense.
The whole experience was a bit weird. I think I'm not the target market for that place: too old, wrong time of day (it's open til 4 am). And I like my tea hot and simple, no sugar or fruit juice, no floating slimy objects of questionable organic origin. Still, any place that has go sets gets my vote.
Unlike some games, go allows (and even encourages) resignations. Professionals believe in the art of resignation, which calls for a resignation at just the right point: after all reasonable options for reversal are exhausted, but before many meaningless moves have been played. In other words, resign when the game is over, not sooner, not later. Figuring out which side is ahead in a position where a pro resigns is often a good puzzle for double digit kyu players like me.
Even for professionals, the timing of resignation is a matter of taste. In The Treasure Chest Enigma, Nakayama relates a story of a student harshly critized by the master for not resigning after a particular move. The student fought back, and turned the game around, and won -- doesn't that show that resigning would have been wrong? No matter, you should have resigned here. At the other extreme is Sakata, who was famous for not resigning. After playing one game out to a substantial loss, he is supposed to have said "I wanted to see how much I would lose by after making six mistakes in a row." And, since pros play for their livelihood, they may play a game out just to require their opponent to prove the win. I suspect small losses are rarely resigned, no matter how certain the outcome.
Weak amateurs have their own set of problems. Most of us are familiar, from one side of the table or the other, with the beginner getting crushed at nine stones and still so confused at the end as to play out the multihundred point loss. For those who know better, continuing a lost game is indeed a bad habit. But the opposite problem also happens. I've read of one teacher who complains that students resigning too late is fixable -- better counting and positional judgement is what needs to be taught, but resigning too early, sometimes while the student in a handicap game is still ahead? This betrays a lack of fighting spirit, which is hard to engender if it's not there already.
My most recent couple of games were both decided by resignation, perhaps early ones. Both were fast 9x9 games. Since offending the other person by wasting their time is irrelevant in fast play, when playing out the game would take perhaps a couple of minutes, my standard for resigning is simple embarassment. I'd rather have a loss +R in my record than loss+(some embarassingly large number). As Janice Kim says, "resign while you still can". In the first game, I'd given two stones to a lower rated player. Playing fast, I'd forgotten to keep track of the liberties on an important group of my stones, and they could be captured in a ladder. Oops. My opponent missed the ladder too, but his moves were good enough to ensure my group would die and win the game. I resigned. My opponent was confused -- "Why did you resign?" I showed him the moves, but I can't help but wonder if he wouldn't have preferred to find them himself over the board. For me, the game was over, definitively decided. All moves failed, and I find it hard to decide between equivalent failure options. However, the game was not over for my opponent, who was still thinking about how to answer the various threats I might still haved posed.
The same thing happened in my next game, but the other way around. I had black with no komi against someone higher rated. I played perhaps too meekly early on, and was looking at a fairly certain loss. I played a cut-atari followed by descending torwards the edge, leaving a two stone group with four liberties in a capturing race with a three liberty group closer to the corner. I had more liberties, but he had the move, and both groups had plenty of ways of making tactical complications. I had no idea what would happen. A dozen or so moves later, things were looking good, and I was busy thinking about a reply to his latest threat when he resigned. But wait! Are we done already? I thought it was my move. Indeed, he was more interested in going back a few moves to see if he could have done something different than in watching me prove what was, at least to him, already a definitive loss.
Thinking about these, I can't help but wonder if there isn't, in addition to everything else, perhaps a bit of one-upmanship at play. Yeah, you may have won the game, congrats. But still, I read out my loss before you could read out a victory. So who's really stronger? And whose taste is more refined? Resigning will never win the game of go, but playing the game is always inside of a larger social context.
One of the differences between Japanese rules and some other rulesets is the treatment of seki. Points surrounded by groups in seki do not count under Japanese rules, so it's necessary to be able to distinguish sekis from other forms of being alive. And sekis can be quite complicated, so on the surface, coming up with a universal definition might be thought difficult.
The 1989 rules do have a complete and simple definition however. Stones are alive if they can't be captured. Live stones are in seki if they share liberties with the opponent. That's it. If you want to score those points you've surrounded as territory, you better fill those dame. Don't want to fill? Must be a seki. In accordance with tradition, you can fill the dame after the game stop (ie. after both players have passed) but fill them you must. This captures all the variations of multigroup sekis, with and without eyes, even things like hane-sekis, where the groups are in atari, yet cannot be captured without a loss.
There's been a great deal of discussion of positions which can lead to repetition. The standard varieties of ko could, were it not for the rule which prevents the repetition from actually occurring. This restriction rule, which introduces a history dependence into the determination of legal moves in a particular position, leads to one of the major strategic areas of the game. If you don't like ko, don't play go. The traditional ko rule only deals with the most common type of recurrence, with single stones recapturing each other in a two move cycle. Other cyclical positions, such as triple ko, chosei, etc., can arise, and have been extensively discussed. In some rules, these positions lead to a void game, which are rare enough to take on numinous significance. Others have introduced rules, such as the Ing ko rule and superko to specifically eliminate them entirely. In any case, you never see infinite cycles actually played out. What never? Well, hardly ever...
It's kind of embarassing to admit, but I sometimes like to play quick games against Igowin. You can do things against a computer that you never would with a real person on the other side of the screen. Also, computers sometimes make interesting mistakes. In this game, I was White, and thought I had the game pretty much sewn up, until the computer turned my entire territory into a big seki. Oops, oh well. That's how it goes, just punishment for not bothering to make even a single eye. At this point, it's Black's move. He should pass, as should White, and Black wins by a healthy margin.
That's not what he did, though. Perhaps he saw the chain of ataris, and thought he could kill White (who after all, doesn't have any eyes, and each of his groups has one). He read that connecting at a didn't help his cause, and also that although White could capture , he could capture back and still be okay. In any case, he played . White must capture, Black must recapture, White has nothing better to do than pass, and we're back to the original position, with one net prisoner to white -- a classic example of sending two returning one.
Igowin follows Japanese rules, which allow repetition, as long as it's not a simple ko. Because he's a computer, he made the same mistake over and over. It's to White's advantage to cycle, so I did. (Yeah, I wanted to beat the computer, but mostly, I wanted to see what would happen.) The net result is that we played the cycle out, over and over. My plan was to amass enough prisoners to compensate for losing my group, enough extra to make up for the territory I lost, then pass twice, losing every stone on the board but winning the game. It didn't work out that way, though. After a few dozen repetitions (I kept losing count, it was 3 in the morning, and I was supposed to be doing taxes) Igowin threw out a strange error (something about being out of memory) then acted strangely until it got shut down.
Because the cycle includes a pass by one side, the net result is different under Territory and Area Scoring. Re-cycling (as Mr. Ing would have put it) costs Black under territory scoring, but is free under area scoring. This may be one of the reasons that rulesets that use area scoring tend to be more careful about prohibiting cycles.
Why a go blog? Well, I have a fair number of thoughts that I want to record, and a blog seems like a reasonable way to organize getting them out. You might argue that, as a double digit kyu, I have little worth reading on go tactics or strategy, and that I know little about culture or history either. This is entirely correct, of course, but it misunderstands the purpose of a blog. Blogging is about the writer, not the readers. Some diaries are worth reading, but mostly, they are interesting only to the original author, and sometimes, not even them. If you choose to read this, don't expect it to be more than occasionally interesting, correct, or useful...
Also, I don't know of any other go blogs -- perhaps this will inspire others to start.
But why do it here on Sensei's, rather than (say) on my own website? Well, two reasons: traffic (why bother posting stuff if no one reads it), and, more important, nice interface. SL has a great interface for updating things from anywhere and a great interface for working with go positions. It's not built specifically for blogging, but it should work well enough. And the culture here seems to be fairly tolerant to new uses of the Wiki.
I will try and follow the convention of regular updates, at least as long as I keep it going. A couple times per week seems minimal, though I may do more than that. And I will also try to keep it to material which bears some relationship to go.
Arno: good luck, I'll sure have a look every now and then :o)