Whither Joseki ...

    Keywords: Joseki, SL description

See also the Where is Go Going and Future Use of SL discussions.

Charles Matthews I'm not convinced that an excessive interest in joseki is a Good Thing. Excessive interest in ratings is worse - Go's equivalent of Original Sin (deplorable but human nature). I still find it depressing that so many players I see know and play a very limited range of corner openings - often from a narrow selection to be found in books now old.

On Sensei's Library, however, there are joseki pages in abundance. I find things there to disagree with, which is more to do with the quarter-century that has passed since Ishida's Joseki Dictionary was first compiled, than any enormous attention to this area on my part.

There seem to be a number of issues here.

  1. Is the basic SL format good for joseki discussion[01] (compared, say, with branching SGF files)?
  2. Should one aim of SL be to provide a good reference for joseki, or just to select a suitable set of variations for learners?
  3. Should joseki material be indexed more systematically here[02]?
  4. Would it be better to assume that people with a real interest have access to the basic books, and concentrate on 'errata', recent developments and innovations, Korean jeongseok and so on?

The database search method very quickly turns up material. I think there is some agreement that this is only a first step.

So, is there a common sense of direction for the deshis here? Obviously everyone will continue to look into what most interests them. But some structure helps too, not least because the site is growing quite fast.

[01] See also: Links in diagrams.

[02] See also: Opening systematic classification.

Dieter: Wonderful though this Wiki is, it has its limits. For instance, I find Go servers far more suitable for playing games. Storing professional games is done by Jan Van der steen and many other sites. As for joseki, I'm in doubt. Kogo's joseki dictionary is an initiative in the same spirit as SL: you can send variations you found (in pro games, in new books) to Gary Odom and he will update the file. BUT, the final editing remains with one person so that it takes more time for variations to settle into the dictionary. Also, SL Recent Changes have the advantage of showing which joseki is under discussion.

I think SL is the best imaginable way to discuss joseki and why not settle the results into a structured set of pages, more or less like we are doing now.

Charles Matthews I have to say I have reservations about Kogo's as a model of how to do things. Simply adding variations from books as one finds them - that can easily overstep any idea of fair usage. Much better to do one's own research from a database and add one's own comments.

One of the problems we have now is that (for purely commercial reasons) Ishida has been re-issued unchanged, so that it is full of old information (compared to the current version in Japanese). I'm not happy with any way of proceeding that will deter future books being written.

I would hope that Sensei's Library would become a distillation of knowledge rather than a chat room, a newsgroup, or a game server. The distilling process may involve some discussion, but before long the discussion can itself be distilled out of a page or the pages and the knowledge remain. -- Hu of KGS

ArnoHollosi: I agree with Hu that SL is not a Go server, a chatroom or newsgroup. Well, the latter might actually not be true :o) (actually, SL was born out of the idea to make a "best of rec.games.go"). I don't mind unstructured discussion. But if you have a look at SL, you will see that most pages have sort of a "permanent" character, more like a book. Therefore I think SL has qualities which make it suitable to discuss joseki or make it a suitable resource for information about joseki.

True, we are all amateurs, so the content of SL has to be taken with a grain of salt. But, everyone is free to correct mistakes or add information on newer variations etc. I think that is the real strength of the wiki concept. Eventually, all information will be correct and complete - well, at least in theory :o)

Now, I reckon that finding information on SL is getting ever more complicated. We rely on visitors to create meaningful index pages, paths and structures. Somewhere in my todo list there's also a pattern search for all positions in SL. That should help as well (when it is added eventually).

As for Stefan's comment: I don't think that everyone has Kogo's dictionary. Furthermore, even if you discuss many josekis, eventually you should have accessible index pages pointing to your discussion, otherwise noone will find the page (and information) once it leaves Recent Changes.

Charles Matthews The ultimate indexing method is just a notation for trees: say jA means 4-4 point joseki, jAA 4-4 point joseki with the usual approach at 6-3, jAAC might be the large knight's move response to that, and so on. Names of this kind are incredibly dull and unreadable, but perfectly systematic. They might be suitable as alias names on pages. Or it might be enough to mention them in the text of any page discussing the joseki starting there, so that one could find all relevant pages through Full Search. Actually, that's quite attractive as a thought - it means that anyone who cares can add references, but that those who don't want to don't have to master any system. More wiki-like?

Anyway, it's a big type of step, to do once-for-all.

Tim Brent: It's also a bit like Chess openings. You may try to learn several, but end up playing what you feel comfortable with for joseki as well as fuseki. I read in The Magic Of Go from Yomiuri that a lot of people play 4-4 basically because they do not wish to learn the myriad varieties of joseki. The last few times at IGS I've seen a lot of players using the Avalanche joseki.

It would be a mistake to think that only joseki in Ishida's time could become obsolete. This will inevitably be the fate of, say, jAC14, leaving a saddening gap between the fully accepted joseki jAC13 and jAC15.

I have been naming pages so that about 5 to maximum 10 joseki are on the same page, typically with the first three corner plays fixed: see 3-4 point high approach one-space low pincer.

And Stefan, if you call Kogo a reference you seem to ignore Charles' comment which I think is a pertinent one. --Dieter

Charles Matthews I don't believe anyone who says they understand the intellectual property position. I have reasons for this. But it is also simply stupid to offend people, independent of the legal position. Also, quite unnecessary. Books should only be quoted only up to the point that the reader says 'now I want to buy the book'.

Andrew Walkingshaw Getting back vaguely towards the subject, I suspect for a lot of players who read SL - certainly for myself - comprehensive joseki dictionaries aren't that useful. What'd be really good for me would be something a bit more discursive - describing the ideas behind various joseki (direction, taking sente, emphasising the corner, side or centre, etcetera) rather than just listing move after move. At the moment, joseki selection is pretty much a black art for this reason, though admittedly I've not invested time in really studying joseki seriously.

I'm thinking in terms of Charles's On Your Side series (which you can get to from [ext] gobase, or Pieter Mioch's "Gentle Joseki" at the same site: these try and put joseki in context, which gives them a much more approachable/comprehensible feel.

Charles Matthews What SL begins to look like is a reference 'back end', and a 'front end' of current discussions. That's already something useful. It is made more useful by exploiting hypertext. Those who have the time can add hyperlinks, adding value to discussions. So, joseki get discussed in the BQM strand in the context of a question of interest to someone. It is obvious that value can be added if there is an existing page to link to, with the relevant mainstream joseki knowledge on it. Same comment for fuseki patterns, same comment and of particular interest to me about side patterns.

I think, to pick up Andrew's point, that SL discussions are going in general to fall short of the polish of a magazine article; but with the advantages that you'll get the perspective of several points of view, and updates over time.

I'd just like to get some idea of what 'enabling' structures could be put in place, fairly much behind the scenes, to make access to joseki (first of all) more convenient.

Arno Hollosi: well, the only enabling structure I can think of is implementing a position search (i.e. search for shape patterns, joseki, fuseki). If done right, those searches could be stored and thus linked to. I assume that would be useful (in general everything that can be linked to is useful :o). Charles, I'm very interested what you mean by "back end" and "front end" of discussions. Could you explain?

Charles Matthews Oh, I'm not a very technical person. But take for example the one-line biographies of pro players. I have added a couple of hundred (?) ... and now it is easy to create name links to them. Anyone who wants to read all these from Names in Go should probably get out more often. But they are there in the background. I feel that the correct way to handle joseki might be the same.

For example, I have added quite a number of pages for enclosures with a systematic naming. In time this should make it easier to have discussions of the middle game, because there is already a page showing a few vital points and standard invasions or probes, etc. Again, I don't suppose anyone would mistake Unusual enclosures for light reading.

Hu: I think Charles is right and Arno's earliest comment about the permanent structure of SL is right too. A "dictionary" of joseki can evolve easily and naturally from the ongoing interests of users, in much the same way as the names. Josekis that will tend to get written uphere are important and interesting ones. The databases and book dictionaries can go for completionism.

Well established josekis will tend to evolve very tight detailed expositions and will spawn pages as their variations get fully explored. Cutting edge josekis will have sprawling discussions. Discussions will be carved off into their own pages as the cutting edge becomes less bleeding edge and more part of the established canon.

A goal is to try to centralize information and link to it rather than duplicating it in multiple pages. That suggests that pages should not become too large and should be chopped up when they do. Joseki form DAGs (directed acyclic graphs). But mostly they can be treated as trees. So a complicated joseki like Taisha has already spawned Taisha Database Search.

Hu: I would like to heartily endorse Andrew's point above. I have found SL to be a wonderful resource to teach myself Go. Guidance is what is most lacking in books and on the Web. Let other resources strive for completeness for reference. Guidance is what Sensei's can easily provide and how it can differentiate itself from other resources. -- Hu

Charles I begin to get a picture of what joseki coverage should be aimed for here. Something like

If footnotes are systematically used, they solve one of the link problems, as in 'see this page note #8'. I still like the sort of lossless notation discussed above, and am not convinced that the 'B12 subsection c' notation is flexible enough for go. But perhaps it would be OK to have joseki A01-A13 for 4-4, etc. - which really isn't so different.

Andrew: A historical note: the systematic naming of chess openings (the Encyclopedia of Chess Openings (ECO) code, A01-E99) came about with the dawning of the database age in chess. Chess Informant, a three-monthly periodical of the games of the top players in the world (and we're talking "telephone-book" thickness here, well over a thousand games an issue) were its' progenitors: nowadays, the ECO code system is utterly invaluable for searching things like Chessbase, and is completely entrenched. For example, when I was studying chess much more seriously than I am now, I wanted all the games I could get in one obscure and sharp opening I played as Black; it was a trivial operation to get this very high-class data to study.

Given Go has always had, more or less, a caste system for the top players, with a small and difficult-to-search set of data being all that was available to the general public until very recently, the dawning of the "database age" hence seems likely to lead to fairly rapid changes to research/study methodology among amateurs. Perhaps, therefore, it's not surprising that the desire for a systematic naming scheme hasn't arisen until the wide availability of databases - GoGoD and Gobase, for example.

Charles It is arguably better for most players' go not to look at fuseki this way. The analogy definitely is with fuseki: the Big Fuseki Dictionary approach is close enough ordered pairs of the notations I used at side patterns to describe parallel fuseki, which is the more important case. There are around 100 patterns to recognise in my private system for indexing fuseki. On the whole I think one should approach joseki with an open mind, for SL's purposes: it is hard to be really original about such a hackneyed subject, but on the other hand the joseki dictionaries do perpetuate some variations better left to die a natural death.

Whither Joseki ... last edited by Hu on June 17, 2003 - 06:21
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