![]() StartingPoints Referenced by
|
Systematic joseki
Difficulty: Intermediate Keywords: Joseki
mgoetze: Being not entirely content with the nonuniformity of current joseki pages, I am considering starting afresh. I will outline my plans below, please comment whether you like the idea, dislike the idea, or have some suggestions for what might be done differently.
In any case, this project will not start until I have my brand new GoGoD. My main guide will always be what the pros play (and how often). Of course, people will be allowed to add discussion about why the pros play it, but I would like to make it clear where the facts stop and where the discussion starts. (This introduction will probably be deleted or move lower on the page once the project starts, if it does.)
DiscussionVelobici: Names such as SystematicJoseki34548447? are very hard to understand. SystematicJoseki34?-54-84-47 is marginally easier to parse. (Hmmm...our wiki software splits words on -. Brackets fix that issue: SystematicJoseki34-54-84-47?) Still one has to perform subtraction to distinguish one space from two space from three space approaches or pincers. One reason to prefer names over numbers. mgoetze: Well the actual page name would be Systematic Joseki 3-4 5-4 8-4 4-7, but the url would of course be as I showed, because the wiki software drops spaces and dashes. Anyway, my idea is, instead of typing in some long sentence in your browser, you just come to the Systematic Joseki start page, click on the navigation diagram at the top 4 times, and you're there. There really isn't any need to understand the page name. JohnAspinall: Seems like an opportunity for links in diagrams. Not only do I not care what the link is named, I don't even care what the page is named. As long as I can find it with a few clicks. GoBase does this. Can't SL? kokiri wouldn't it be possible to use the alias property of pages in order to have an underlying systematic structure but retain easier to understand names as well?
Velobici: What I find I consider in games is not a what's the next move reference but rather I want to make thickness facing left what move gets me that from the set of possible moves: approaching high vs low, near vs far etc. Given that move, what can the other person do to prevent or deflect me from that goal. Its not the path to the pattern but rather the pattern and wanting to know what path gets me there and what detours are possible. Almost as if we where to have a komoku, high approach results in these diagrams type of Joseki reference. the present idea seems to concenrate on the path rather than the possible results. mgoetze: I could of course give a "main line" for each of the continuations. But I don't believe it's possible to make it quite as easy as you would like to, since you also have to consider at what point the opponent can choose another line. I don't see why there shouldn't also be "How to choose Joseki" pages for kyu players but, indeed, the method I am planning to follow achieves a different goal. mgoetze: I started some sample pages so people get a better idea of what I have in mind for the 6-4 Point. Dieter: before you go full steam, I'd drop my sympathetic dislike of the term systematic joseki. We have had a couple of attempts and suggestions for what you have in mind and discussed them at length a while ago. I'll see if I can find them. In any case, preceding every page with systematic joseki is rather heavy. BobMcGuigan: I'm not sure it is worth re-creating joseki dictionaries on line. However, I think it would be nice if joseki "reference" diagrams could be made to switch stone colors and orientations. I think some people might have difficulty mentally changing colors or board orientation when trying to compare what happened in a game with what is on the reference page. mgoetze: I have no plans to re-create any joseki dictionary, they are not good enough for me. ;) If you have too much trouble switching colors or board orientation, you might have to actually use Kombilo yourself, instead of seeing what happened when I used it. Dieter, do you have a better suggestion for a name? I do want to keep a prefix so as not to pollute the namespace. Dieter: In fact I am pretty happy with the existing mixture of pagenames like 3-4 point, high approach and Magic Sword. I think we should gradually improve on that, creating aliases, splitting up pages, raising new discussion, giving examples from pro play, instead of rebuilding stuff from the ground. A question of effective use of energy, IMO. Rafael: I don't like the Black plays next: 20 matches (14 at a, 5 at z) thing. For several reasons. First, it will get outdated fairly soon. Also, I think it means getting too much information from whatever source you're using (Gobase or GoGoD). And while it's fair to remark when a joseki is not played anymore, or when it has never been much played at all (as the MagicSword you mention), what's the use of these stats? The page should explain what are the benefits of each key move, how they relate to the whole board, and maybe mention something like: The usual follow-up is a, or The hane at b is uncommon but can be played when White is such and such. Stats are misleading. mgoetze: The idea is to have an at-a-glance summary with statistical information, and then a commentary on the moves further down. I would probably copy much of the material that Charles has posted here on SL, for instance... myself, I'm too weak to comment, so I let what pros play speak for itself. And I don't plan to limit myself to only one information source... (i.e. I would use GoGoD, go4go, Friday Night files...) Charles You do realise that, with one page per node, you are proposing about 20 times the number of systematic pages that are here now? (Joseki dictionaries have about 3500-5000 lines.) Well, perhaps I misunderstand your intentions. I have no wish to discourage (I know that people always say this, before being discouraging ...). A good systematic system would be one that actually encouraged others to contribute and edit - so needs to be somewhat flexible. crux: Doing this manually seems tedious - this should be an ideal task for a computer program. I don't see why you couldn't automatically create a rough but comprehensive set of corner sequences from a database and emit it in an sgf file like Kogo's dictionary. If you extend the program to read an existing file and keep the comments it finds, you can keep it up-to-date whenever your database of games grows in size. (Sebastian:) Your proposal contains two features that make it systematic:
I don't think that these improvements require the creation of a parallel universe in addition to the existing Joseki pages. Once we agree on your proposals, we can work on this together at a leisurely pace, adapting layout and name of existing pages one page at a time. John F. Slightly random comments, neither for nor against. Just grist for whoever wants to run a mill. I'm probably one of the few who has experience of working in a large-scale way on josekis via a large database. The dominant conclusion is that it is BORING. I have kept it within bounds by just doing the Nadare (1310 games, 604 different lines). The extra hands on SL may make lighter work, but they are also taking on a wider range of josekis. I'd be surprised if a major project like this didn't run into the sand pretty soon. Worse, the exercise is probably also pointless. I say this for two reasons. One is that I have never made any use of my JBase, or any of the various joseki dictionaries. I'm not convinced by those who say they find them useful. Second reason is that no-one has written to us (GoGoD) about josekis. I infer, therefore, a relative lack of genuine interest. Against that, I have found stats useful, and so probably disagree with Rafael (his word was "misleading"). They can point to interesting nooks to explore, they can confirm or demolish hypotheses. I generally feel that there is a badly misplaced obsession with josekis among kyu players. They may think they are a quick way to mastery. They may be influenced by a prior chess career. Either way, they should be disabused. To crux's point about getting a machine to do the work, Michael Reiss has done this for the next issue of his Go++, using GoGoD games (with permission). What I think would be much more valuable is a sort of reverse joseki dictionary. A list of final position types (wall on the right; quick small territory but in sente; centre thickness; etc) and examples of josekis that will achieve such goals. Incidentally, I see a lot of hair-splitting about what is and what isn't a joseki. I'm of the school that regards it as nothing more than a tactical exchange in the corner or the side in an open area of the board. The point about an even result is so often ignored it's otiose (Dieter: What do you mean?. Joshual000 Why not begin on a small scale; start with the 3-3 point, build various front ends (aiming for the left side territory, aiming for influence, all acceptable moves, etc..) I believe this can be achieved using the current naming convention, and possibly splitting position pages. Obviously attempting all joseki is an impossibility, so start with something achievable. uxs: The thing that I like most about the proposal are the links that would be in the diagrams so you would be able to find the variation you would like to know about more easily. Compare 44PointJosekis (=good) to 33PointJosekis (=not so good) to see what I mean. Starting a whole new series of pages for this however would be bad, better to keep the information about 1 joseki on 1 page. Bill: Technical footnote: It is possible, and perhaps desirable, to organize joseki material on a single page or a few pages, by using footnotes as bookmarks. If you put the number for the material at both the beginning and the end, it makes it easy to return to the previous place on the page by clicking the number. (That does not work if you go to a bookmark on a different page, however.) This is a copy of the living page "Systematic joseki" at Sensei's Library. ![]() |