Forum for Joseki

A Most helpful thing [#172]

Back to forum     Back to page

New reply

jared: A Most helpful thing (2005-11-21 06:49) [#518]

Recently, criticism has been leveled at the amateur level of the SL deshi. We can all agree, a little more professional approach to the concrete areas of SL would go far in deflecting such criticism.

I suggest we make a collective effort to 'cite' any information we add to SL joseki. What does a joseki citation entail? How about:

  1. The first professional game the joseki appears in
  2. Further pro games which show the joseki in context

I don't know about everyone else, but I learn by example.

I think this kind of 'citation' would take our joseki reference to a whole new level of awesomeness. :)

X Re: A Most helpful thing (2005-11-21 19:48) [#520]

The most recent known pro game with an online game record in which a joseki appeared would be interesting, I think. It would not only give a sense of current pro opinion and frequency (rare joseki would tend to have older citations) but would provide context for the chosen sequence of play, as you indicate. I think that context is very important. Once there are a few stones on the board, some joseki are more equal than others.

Calvin: Re: A Most helpful thing (2005-11-23 18:43) [#537]

It's important to be realistic about what can be done here. The repetoire of joseki is immense, so comprehensive coverage of variations is probably not going to happen. What I've found in the Systematic Joseki sections so far is that they do not add much to the material in existing dictionaries.

One option is that SL, because it is dynamic, should trace some recent developments in joseki, but this would require participants who have regular contact with active professionals.

One thing I'd like to see is an analysis of common joseki mistakes. There used to be a database of variations built from amateur games of players of all stengths on KGS, tagged with the frequency of strength range of the players. It was interesting for me to go through it and find variations, for example, that were common but not played by players stronger than 10kyu, or 2dan, or whatever. Usually these are patterns where the normal, intuitive move isn't optimal. Unfortunately, the database had a Korean name that I don't remember, so I can't find it on the internet anymore. The was a python program that generated SGF.

Calvin: Re: A Most helpful thing (2005-11-23 18:59) [#538]

I found it. It's called "soksoo." [ext]

tapir: Re: A Most helpful thing (2010-02-02 15:26) [#7154]

I disagree to base discussion on citations. Though mentioning sources where available is good, and referring to actual games is even better. But I fear that the whole approach to the opening on SL would be even more antiquarian with an "citation needed" policy for often the citation would be from joseki dictionaries.

reply ((no subject)) (2005-11-21 18:46) [#519]

Bob McGuigan: References to sources are a good thing, but many, if not most, people could not provide the kind of references you describe. Even if such were given it would not mean the information is correct, though, since joseki undergo re-evaluation. Some sequences that were considered joseki 30 years ago are not considered joseki any longer, and there are many older pro games in which they were used. My impression is that many of the joseki pages depend on material more-or-less copied from published joseki dictionaries, sometimes including sequences that even the dictionary says are not joseki. What counts, anyhow, is the reasoning behind the moves, not just the move sequence.

Nachtrabe: ((no subject)) (2005-11-21 20:18) [#521]

I argued strongly against doing this kind of thing on the Systematic Joseki pages, but here's my rough argument again.

The points I made then, addressing both when it was last played and the relative frequency of a given jeongseok:

  • Joseki are Overrated: Knowing which is the most common reply is good, but is it really important to know that this particular joseki is played in 2 percent of pro games with the prior situation without knowing what the rest of the board looks like?
  • Database Biasing: Is that 2 percent of games in the last year, five years, ten years, or fifty years? Each of these numbers has a different value and none of them are particular useful unless you are examining a particular pro's style compared to other pros. Does the database cover all pros, or is there a bias away from less known pros? These biases matter to anyone doing serious research, and they undermine the value of statistics that are collected.__
  • Updating: This is not some database collected long in the past. This is an evolving data set. Joseki are a living set--particularly with the Koreans on the scene--and the pros have trends. I've made it a personal project to keep a small set of professional Korean players updated (about 12 of them) because some of them hadn't been updated in years. That's relatively trivial and anyone can do it with a little practice. Joseki require access to specific databases and the software to scan them, then it requires updating every page in the library, which leads me to the next point:
  • The Size of the Database: There are tens of thousands of joseki. For creating the database this isn't so much a problem--we can do it one page at a time and the page content only changes when a variation gets added. But keeping them updated? That's another question entirely and there's really no way to cleanly keep things updated.

The first time a joseki was played is kind of nice to know and, unless there is a correction, it never changes. But the last time? I lack the ambition to even keep track of that for my favorite players (and the expertise to do so even if I had the time), much less for several hundred players playing games all of the time with tens of thousands of possible joseki.

I'd like to add to that a few additional comments:

  • Just because it occurs in a pro game does not mean that it is jeongseok. Sequences are constantly being re-evaluated and reviewed. Sometimes a new move will become jeongseok, but sometimes a refutation is found.
  • If something is older and less frequently played today, it does not imply that there is anything (necessarily) wrong with it or that it is not still considered jeongseok--simply that the modern style has moved away from it for the moment (e.g., playing a knight's move instead of a one-space extension when there's a knight's approach to a 4-4). Sometimes that 30-year-old sequence has been refuted, sometimes it simply gives away too much influence or is too calm for the modern style, or sometimes it has simply fallen out of favor because the winds have changed direction.
  • As you say, context is very important. One game is often insufficient to gain that context though, so if you are interested in studying a particular jeongseok in context the base way to do it is to go over many many games that it has occurred in, rather than just looking at the last few.
reply thanks for your thoughts (2005-11-22 22:58) [#533]

thank you for your thoughts.

Bob, I agree that many people will not be able to include a 'citation' when they add a new joseki. I suggest that if a deshi finds such information that they add such to SL. I would really enjoy it.

I am most interested in the first use of the joseki -- the 'inventor' of the move.

Much harder (and time-consuming) to provide (and update) are the uses of a joseki in context. Despite the difficulty of providing this information, I still urge motivated deshi to try. :)

Back to forum     Back to page

New reply

[Welcome to Sensei's Library!]
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Login / Prefs
Sensei's Library