Life In 19x19

Mynavi sell Go books in Japanese as PDF - which can be bought and downloaded. Note they won't ship printed copy books to the UK. The PDF is licenced to you and there is a page at the back with your details.

https://book.mynavi.jp/ec/products/?top ... 1&pageID=1

The available PDF contain some exciting books - please don't get carried away!

I was pleasantly suprised to find that in many cases the text in the book can be picked up with select/copy and pasted into DeepL for a reasonable translation. This is not the case for all PDF, in some cases the text is not in the PDF as unicode it's an image - which can't be selected.

The website contains some 300 books but the search above on 碁 also turns up some Shogi books and magazines.

I started to look at books using Adobe Acrobat and the first book was "All About Sonoda's Proverbs" by 苑田　勇一 it was recommended by John Fairbairn some years ago.

https://book.mynavi.jp/ec/products/?top ... 7%E4%B8%80 will find you three books by Sonoda. "All About" is the first one and the PDF costs Yen 1135 - or some UK £7.50 which is not bad.

Here is the text on page 234 - japanese text pasted into DeepL for translation.

馬暁春九段との一戦です。私の黒番。最初は定
義の基本問題から。辺が四つありますが、条件を
考え、大きい順番を答えてください。

"It was a game against Ma Xiaochun 9dan. It was my black turn. At first, we started
I started with a basic problem. There are four sides.
Think about the conditions and answer in order of size."

I haven't discovered anything that indicates if the text from the PDF can be extracted or not - so that is a bit hit and miss.

I have successfuly extracted all the text from Sonoda into a single file using Foxit PDF Editor - Adobe have an add-on for Reader but I had no success with it. I successfully manged to OCR the text in four books so that it could be selected/copied into DeepL. I used Read Iris V17 (I have used Abbyyy Fine Reader in the past which I will revisit).

DeepL proves quite good machine translation - it picked up Ma Xiaochun's name but it might struggle with Japanese Go terms in katakana - "sabaki" becomes "mackerel". I have found it best to give DeepL one sentence at a time.

Please feel free to PM me if this is an area of interest - PDF text extraction is full of pitfalls and finding what works and what doesn't is quite "exciting".

John Tilley

You could entice us with more, uhm, critical information. What tesuji did Mr. I-know-more-tesuji-then-that-kid play? Was he fluid like water?

If mackerel is sabaki what is cod?

Good find. I didn't know MyNavi sold PDFs. I have a few books that they've published but they are mostly secondhand. I think many of the MyNavi books are on Amazon.co.jp if someone is looking for the actual book. Amazon JP is pretty good with shipping. Still, the PDF have the huge benefit of being able to grab the text.

I find that DeepL works good enough when translating normal writing, like the history/introduction for a game record, or journalistic articles. But for me, DeepL starts to breaks down when reading go terminology in the game commentary. Like, I'm look at the Japanese book and the DeepL translation and it's just dropped whole statements. Google translate often provides a worse translation but it doesn't seem to suffer from the dropped statement issue. In the end, game records are so formulaic and descriptive that I end up just using the Google Translate app for OCR and a quick summary and then copying the character to look up in a dictionary as needed.

I wish the Kindle version of Go World had text that could be highlighted and copied.

John Tilley wrote:Mynavi sell Go books in Japanese as PDF - which can be bought and downloaded. Note they won't ship printed copy books to the UK. The PDF is licenced to you and there is a page at the back with your details.....

You can:
search for books + go category

If mackerel is sabaki what is cod?

Actually mackerel is saba (and cod is tara).

You may mock, but there is a nice phrase you can use in go (saba wo yomu - counting the mackerel), which is to cheat by giving a wrong number. A well-known amateur counting-up tesuji? The posh name for mackerel is the Scombroids. That's perhaps a nice euphemism you can use about such people (i.e. instead of scumbags)?

And there is also a mimetic phrase sabasaba suru, which means to be relieved. Which is what you feel after you have achieved sabaki (= coped in a difficult situation).

You can start using these phrases at the London Open and impress all your friends. Just make sure you don't let a scombroid get you in the cod-piece lest you end up the opposite of sabasaba (i.e. basabasa, making whimpering rusty sounds on the floor).

Thank you very much for the information. I made an account, which book do you recommend? By the way deepl seems very good^^

John Fairbairn wrote:You may mock, but there is a nice phrase you can use in go (saba wo yomu - counting the mackerel), which is to cheat by giving a wrong number. A well-known amateur counting-up tesuji? The posh name for mackerel is the Scombroids. That's perhaps a nice euphemism you can use about such people (i.e. instead of scumbags)?

Such counting tricks stink like a mackerel. I think another phrase for Go is to call a dead group "dead as a mackerel" but I am not sure if that is a real colloquialism, a baseball term or both.

I think another phrase for Go is to call a dead group "dead as a mackerel"

If that's a genuine phrase it's new to me. My own maybe regional phrase for same is "stone dead" (or "stone deid" to be pedantic). Which is proof that go was invented somewhere in Britain. It probably disappeared, and so was forgotten, because it was banned after English go fans tried to break into the London Go Centre without bubonic plague passports.

Lichigo - How strong are you? Let me know and I will offer some suggestions. I will look at translating with Google (again) and you might want to try both Google and DeepL. DeepL does seem to miss phrases and even sentences, as has been mentioned earlier.

The following translation of a book I bought was provided by Google in March 2017:

"Rather than purchased, but is what was hit in the lottery, because I have a casting pearls before swine, I yield to the lovers of someone go. Never used even once, it is what has been kept at home."

I rather liked this, but it put me off Google translate.

Best Wishes - John Tiley

John Tilley wrote:Google in March 2017:

I rather liked this, but it put me off Google translate.

As far as Japanese is concerned, Google translate has improved a lot in the meantime, according to my obversations.

I would like to second the tip using both AI in parallel.

John Tilley wrote:The following translation of a book I bought was provided by Google in March 2017:

"Rather than purchased, but is what was hit in the lottery, because I have a casting pearls before swine, I yield to the lovers of someone go. Never used even once, it is what has been kept at home."

I've seen Google translate through out a couple shocking words multiple times when translating go terms.

I find that it's best to use the translation for a quick overview and then jump to the sentence that is most interesting and use the dictionary from there.

Unfortunately games commentaries include a ton of descriptive fluff that is a waste to translate (e.g., moves A to Z settle the group, so that...). At least it is easy to identify the move numbers and figure references to find what I'm interested in. Sometimes I just want Shuho-level commentary: "Black 21 is good."

Unfortunately games commentaries include a ton of descriptive fluff that is a waste to translate (e.g., moves A to Z settle the group, so that...). At least it is easy to identify the move numbers and figure references to find what I'm interested in. Sometimes I just want Shuho-level commentary: "Black 21 is good."

Obviously it's everyone to his own taste, but I think you may be missing a lot with that approach. When I started doing the Go Wisdom indexes, I was rather shocked to find that the incidence of each term could surprise me. I was particularly asonished to see how often comments came up about probes, forcing moves and tsumes (blocking extensions). I already knew these were big priorities for pros, but their importance was much bigger than I expected. In contrast, invasions, moyos and attacking were generally very low. I think all of this is the reverse of the amateur mindset - at least in my experience.

Another of the big surprises for me was the very type of comment you mention: settling groups. Not only do pros seem to pay more attention to this than we amateurs do, but it turns out to be part of a nexus of prophylactic moves, all of which pros love and we despise.

"Black 21 is good" is funny (in its own way) but what does it teach you? I am fully aware it tells you something about the state of the game - the pro commentator thought this was an important cru=x point in the game" - but AI can tell you that.

I made my comments in reverse order.

John Fairbairn wrote:"Black 21 is good" is funny (in its own way) but what does it teach you? I am fully aware it tells you something about the state of the game - the pro commentator thought this was an important cru=x point in the game" - but AI can tell you that.

For me, something like "black 21 is good" has very high value compared to its translation complexity. When a statement is this simple it will be easy for computer translation and I might even be able to read it myself. Tsumego books often have simple statements like this that I can read or figure out with my limited vocabulary. As for AI, it's not usually helpful for me. I don't need to know about some complex sequence having several moves I'd never play to set up a complex ko for a half point win. I prefer pro commentary with moves I can understand and use.

Plus, the Pros have already selected the more interesting variations in the more interesting games. There is a lot of entertainment to be had from a pro Game Commentary with little effort. I do not enjoy digging though AI suggestions.

John Fairbairn wrote:Another of the big surprises for me was the very type of comment you mention: settling groups. Not only do pros seem to pay more attention to this than we amateurs do, but it turns out to be part of a nexus of prophylactic moves, all of which pros love and we despise.

Well, this is a big assumption on my part, but I've noticed that there are usually diagrams when a move settles a group or not (unless it is obvious, even to me). If it is a tricky situation, then I'll see the move numbers and 図. If the variation is interesting then I'll translate that part. So translating the one or two sentences introducing the diagram is not very helpful.

John Fairbairn wrote:Obviously it's everyone to his own taste, but I think you may be missing a lot with that approach. When I started doing the Go Wisdom indexes, I was rather shocked to find that the incidence of each term could surprise me. I was particularly asonished to see how often comments came up about probes, forcing moves and tsumes (blocking extensions). I already knew these were big priorities for pros, but their importance was much bigger than I expected. In contrast, invasions, moyos and attacking were generally very low. I think all of this is the reverse of the amateur mindset - at least in my experience.

I'm sure it would be nice but for now it's too much hassle. Just to see, I went and translated a paragraph of commentary that I skipped in the last game I reviewed:

黒7あたりから作戦の分れるところ。白9なら黒いなので、自8、10と利かす。白12、黒13、黒17、白18はコスミが四つ現われたが、とりわけ白18はのがすことはできない。逆に黒18と掛けられるのとの差である。白20から22と、幻庵の序盤作戦はスケールが大きい。

Even using the Google Translate phone app to take a picture and OCR, it still took me a few minutes to correct all of the numbers (a common failure in OCR), force quotes on the "iroha" (or letters as those are messed up too), and review the text (this text is probably close). And for my efforts I get this:

[Google Translate] The place where the strategy can be understood from around Black 7. White 9 is black "" i "", so use your own 8 and 10. Four cosumi appeared in White 12, Black 13, Black 17, and White 18, but White 18 cannot be removed. On the contrary, it is the difference between being multiplied by Black 18. White 20 to 22, the early stage operation of Gen-an is large in scale.

[DeepL] Black 7 is where the strategy starts to diverge. White 9 is Black's "I", so Black 8 and 10 should be used. White 12, Black 13, Black 17, and White 18 have four kosumi, but White 18 in particular cannot be left out. White 12, Black 13, Black 17, and White 18 showed four kosmis (but I can read , but White 18 in particular could not be removed [[sometimes DeepL duplicates sentences, sometimes it drops them]], and the difference was that Black 18 could be multiplied . White 20-22, Genan's strategy in the beginning of the game is very large.

At least I didn't have to translate it by hand like in the old days. But when I just skim the Japanese with my limited vocabulary, I feel like I got 80% of the value from the translations. I noticed Black 7 when setting the board and then I saw it mentioned in the text. But the translation doesn't say much about 7. Then I see 白9 (instead of 黒9 on the board) and 黒い and can wonder about how that would play. Then I see 12, 13, 17, and 18, which I already played on the board, and yup, they are all kosumi (but I had read コスミ anyway). Now, I could not read about move 18, but then the book says 黒18 so I knew to consider that move. And I already thought that White 20 and 22 were interesting moves when playing the stones. So, I can't say that I got too much more value out of the translation compared to just fumbling through the Japanese text -- at least, not in the main description. I do find the diagram explanations to be helpful and worth translating.

For game commentaries, I've gotten more value from time spent learning Go vocabulary and grappling with grammar than trying to use machine translation. Though, I definitely use machine translation for the introduction and conclusion statements. These are written in normal language and the translations work well enough (without any need to fix text or review).

天保十年には幻鹿、秀和戦は三局あり、一勝一敗一打掛。ことでは、秀和が幻庵に押しまくられながらも、二枚腰を発揮してよく一目余した一局を取り上げる。丈和隠退の半歳まえ、幻庵、秀和決戦の予兆が感じられるころの前哨戦である。因碩八段、秀和六段であった。なお、秀和の跡目は翌天保十一年五月に聴き届けられたから、それまでは土屋姓を名乗っている。久貝因幡守は五千五百石の旗本、大番頭（江戸城警備の役頭）などを勤め、当時しばしば高手を自宅に招いて碁会を催した囲一碁数寄者であった。

NOTE: The translations think that Shuwa is "Hidekazu." But if you put "本因坊秀和" then it can guess Shuwa.

[Google Translate] In the 10th year of Tenpo, there were three games against Genka and Hidekazu, with one win, one loss and one hit. By the way, while Hidekazu is pushed by Gen-an, he shows his double waist and picks up one station that is often overlooked. Half a year before the retreat of Takewa, it was a prelude to the time when the signs of the decisive battle between Gen-an and Hidekazu were felt. It was Inoue 8th Dan and Hidekazu 6th Dan. In addition, since the trace of Hidekazu was heard in May of the following Tenpo 11th year, he has been calling himself Tsuchiya's surname until then. Inaba Mamoru Kugai worked as a hatamoto of Gohyakkoku and a large guard (head of the guard of Edo Castle), and at that time he was a Go player who often invited high-ranking players to his home to hold a Go party.

[DeepL]In 1835, Genan and Hidekazu played three times, winning one game, losing one game, and beating each other one time. By the way, Hidekazu, while being pushed by Genan, showed double-heartedness and picked up one game that was often overlooked. It was a preliminary game when the signs of a decisive battle between Genan and Hidekazu were felt six months before the defeat of Takezawa. It was between Inoue 8dan and Shuwa 6dan. Also, in May of the following year, Tempo 11, traces of Hidekazu were heard, and until then he took the surname Tsuchiya. Inaba Morikugai was a Go player who often invited high-ranking Go players to his home for Go meetings, even though he had been a 500-koku Hatamoto or a Daibancho (the head of Edo Castle).

By the way, didn't DeepL mess up the date? Isn't Tenpo 10 1839? This game is in GoGoD as file 1839-05-16a.



--------------------

Going back to MyNavi, the PDFs seem great because it takes away most of the headache of translation. No fiddling with a camera. And the text doesn't need to be fixed or reviewed. But I prefer physical books if given the choice.

I'd overlooked you were talking about translated commentaries only. Sorry. But as regards the AI samples you showed.... Not being an American I very, very rarely say Wow (and I never say awesome or cool), but in this case of extreme wowity - WOW!!!!

I had extreme trouble even matching the English to the Japanese - two different universes.

Just to comment on a handful of the examples in the Shuwa text:

- "winning one game, losing one game, and beating each other one time " should be "one win, one loss and one game left unfunished.
- Yes, Tenpo 1 is 1830.
- Genka should be Genan
- double waist/double heartedness should be stubbornness/tenaicty (nimaigoshi - it's from sumo, but in go Rin Kaiho is famous for it)
- Shuwa didn't "take" the name Tsuchiya - he had it; he was born with it.
- Inaba Morikugai should be Lord Kugai of Inaba and he wasn;t a daibancho but an oobangashira and that wasn't the head of Edo Castle (which was the shogun, of course). The oobangashira was the head of the oobangumi which was the protection force of any of the five shogunal castles. The Japanese here says more or less that, too ("head of the Edo Castle protection force).
- Half a year before the defeat of Takewa makes it sound like a famous battle. The reference is actually to Honinbo Jowa.
- Picking up a station that is often overlooked??? Picking up games by emerging one point ahead (through nimaigoshi) stubbornness) somehow seems to have just a little different nuance.

As I understand it, the way modern machine translation works is to put a very heavy reliance on finding existed translated texts (different ones 99.999999% of the time) on the internet that marry up in part to the desired text. There are usually differences and so a selection is made based on a combination of percentage match and frequency of translation equivalents (e.g. Hidekazu for Shuwa bevause Hide and Kazu are very often used in boys names.) This works a treat for standard or common phrases (it is raining today) and goes haywire with anyth8ing not so bland. To compound that basic problem, the computer can only work with what has been input by usually untrained users. So, if a user thinks 大番頭 should be read daibancho instead of oobangashira, that what deepL and google get to work with. Garbage In Garbage Out.

I'm sure they'll get better with time, but learning Japanese for yourself will give you so much more. It may seem like a huge task, but I covered the first London Marathon as a journalist. This was a time when ordinary people believed only superhumans could run 26 miles. When I interviewed many of these ordinary people after they had completed the race, the sheer joy and pride they felt (and continued to feel long after) made even me say Wow. (And I have to admit that seeing a squad of Gurkhas run the race instep in perfect military formation almost made me say "awesome." I didn't say it but I thought it.)

John Fairbairn wrote:I had extreme trouble even matching the English to the Japanese - two different universes.

Looking at this again, if I had never read about Shuwa in English then I would have no idea what the translation was trying to say. But having already read something about these events, I was about to piece a few things together. Not a good translation but since I didn't need to do any spell checking (like with game move commentary) then it was worth the effort for me.

The translated names are hardly ever right so I get good use out of the GoGoD NamesOct2019XML.xml.

John Fairbairn wrote:I'm sure they'll get better with time, but learning Japanese for yourself will give you so much more.

I think so too. And it has already. Luckily Go has limited vocabulary and grammar. And it comes with diagrams. Otherwise I probably wouldn't have bothered. Reading normal Japanese (without a dictionary) is something I don't even aim for, but my ability to read some Tsumego prompts/hints (without a dictionary) came easier than I first thought. So there is hope for me to read Go content. Plus, the Japanese Go scene is fun to follow.