Japanese tournament records are well organized and widely available. So I did not expect to find many errors. But I just find this Old Meijin game from a Chinese book, which is slightly different.

This is the widely available version:



This appears to be wrong for several reasons: the results is B+12 but the position on the board is B+14; move 162 looks very strange; Move 222 is a smaller yose than J10.

This is my version, which has the correct result and looks more reasonable:

GoGoD (1965-09-18a.sgf) matches yours.

It's interesting that the first version of the game gives the event as "4th Old Meijin Final". This is my personal usage, and as it's not a translation it's hardly likely that someone else would come up with the same wording independently. Yet the moves are different....



I'm intrigued why you think this. Many games of the early Honinbo are actually still missing. As to being well organised, even Kido did not publish Games 2, 5 and 6 of one term's title match, but did publish 1, 3 and 4. But in non-titlematch games, players have long been responsible for providing the game record. The norm was for the winner to write the game down from memory afterwards, but (a) this produced errors and (b) quite a few players were notorious for not putting the records in their dockets for the admin people. A young lady there was quite diligent in chasing laggards up, but had no punishments to enforce the rule.

For quite a few of the Old Meijin and other early league games, GoGoD had to get the game records direct from the players themselves (it'll be interesting to see where else these turn up ), but there were cases where we failed - players had died, etc. Typically the missing games were by players from the Kansai or Nagoya branches of the Nihon Ki-in, and so not played in Tokyo. This is the likeliest explanation of the reason why game records were not deposited in the Tokyo pigeon-holes.

Another major problem of this era was that Kido did not publish dates with the game records. In many cases I was able to find the dates from other sources but for even more I had to just resort to marking the date as "Publication" date. People who have omitted this word to disguise their copying have of course spread wrong dates around the internet.

One practice of some early compilers of databases (but not GoGoD) was to switch round a couple of trivial endgame moves, to act as a marker. These games still surface and have to be counted as errors. (I know many of them because we early compilers used to correspond quite often and we traded "secrets".)

So, beyond Kido (and Kido only when it could be bothered) , where else are tournament game records of this era "well organised and widely available"? The only other major sources I can think of, which are far from complete, are hard-to-get newspapers and very expensive game collections of individuals.

My own experience is simply that the Japanese records have a much higher standard than Chinese and (worst of all by far, Korean when it comes to avoiding typos. They have also been at the business of publishing longer than the Chinese and Koreans. But I have had clumps of hair to prove they were not well organised or widely available.

Well, it is all relatively speaking - good luck if you try to locate a missing Chinese game from 1980's or even 1990's.

John, I am puzzled by this statement. "4th Old Meijin Final 6" is a direct translation from “第4期旧名人战决赛第6局“. Why is this use personal to anyone?

In this context "Final" is a bit abnormal. It has a very strong connotation of a knockout (i.e. an event with semi-finals and quarter-finals), and so is appropriate normally only for things like the Samsung Cup. Other prolific western writers such as John Power and Bob Terry or database compilers such as Jan van der Steen and MasterGo have (correctly) preferred terms such as title, title match, etc.

As to the reason why I opted for such an unusual term, I need to take you back to the antediluvian days of database compilation. Those few of us involved had, as I mentioned above, frequent correspondence, and among the topics were standardisation and how to cope with the inadequacies of the sgf format. For example, Jan van der Steen was very keen to use the format EV[Meijin, 4th] as this made indexing easier for him (sensible enough, though he had problems in the case of the Meijin with the change of sponsor and the renumbering). Others preferred RO[1] in place of RO[Round 1] or RO[Game 1]. There were also diverging views on how to cope with transliterations and dates and connecting kos, or whether final passes should be recorded. There were also things that we all wanted to include but the sgf format didn't have a special rubric for, meaning lots of info got dumped into the GC[] game comment or the GN[] game name. Some of us insisted on komi in the form 5.0. Number lovers such as myself preferred 5.

There were so many points at issue that we all ended up doing our own thing, and the result has been that every database ended up with a sort of fingerprint of the compiler. In the case of GoGoD, for example, our fingerprint includes Final, McCune-Reischauer, a special way of handling lunar dates, and quite a few other distinctive things.

Because these were the dark ages, there were no database search programs. I wrote my own, tailored to information I wanted to cull for the GoGoD tournament database. For reasons I can't remember exactly, but they included something to do with tracking titles won by Sakata, as there were discrepancies in the various numbers quoted in Japan, I wanted to be able to call up all title matches in my search program, but the problem I had was that I could only search on the whole database in one pass on one item. So I could not search on, say, Honinbo, and then search on the results for Final. I had to opt for the format EV[Xth Honinbo Final] RO[Game 1] rather than EV[Xth Honinbo] RO[Final, Game 1].

As to why Final as opposed to Title Match, again I can't remember all the reasons but one was that we were having problems with "Title Match" "Titlematch" and "Title-match", and another was that not all events had a proper title (e.g. Ryusei or NHK Cup) and Final seemed to encompass all events well enough.

So, while not impossible, especially nowadays when people may have been influenced by GoGoD usage, I believe it rather unlikely that anyone else would render 第4期旧名人战决赛第6局 as a "Final", not to mention that Chinese normally add the word for Japan. It's an interesting linguistic nicety that, on the other hand, it is very easy to write a script satisfactorily to convert all games with entries such as EV[4th Old Meijin Final] RO[Game 6] into 第4期旧名人战决赛第6局.

Many problems remain. Some compilers used to prefer Ten Dan to Judan. Not unreasonable, until the Siptan and the Shiduan came along. So some people like to say Japanese Ten Dan, Korean Ten Dan, Chinese Ten Dan. But in the case of the Meijin, it seems very odd to me to refer, as some do (Go World included?), to the Chinese Meijin instead of the Mingren.

Another big problem in China is re-using the same tournament names multiple times: the Qiwang, the Old Qiwang, the Very Old Qiwang and the Even Older Qiwang somehow doesn't sound convincing.

I have coped with all the problems in a purely empirical way, having long ago realised that it's impossible to come up with a watertight protocol for a go database.

Very interesting stories.

I over the years also had to invent lots of terms to describe tournaments. In early years when my English wasn't as well-developed, I had very inaccurate and sometime silly translations. For example, I simply called these bilateral competitions something like '10th jp-cn Agon Cup'. Unfortunately, these terms got copied by other people

Funny you should say so. My experience is that it is very common for old games to be available in several different versions (when available at all).
The Western sources I have seen all follow (or at least agree with) Martin Müller (don't know about GoGoD; gobase lists data consistent with Müller's).

Four Chinese sources, and also Logan's Rin Kaiho and Sakata Eio pdfs
(by the way: it is easy to turn these pdfs into sgfs, and the sgfs are much more useful)
agree with the second version.

Aren't Logan's pdfs made from the GoGoD sgfs?

I don't know. GoGoD is advertised as being high quality. Logan's pdfs have a very large number of flaws.