Playing "Sino-Tibetan formations" ("1-3-5s")

Lately I’ve been getting a bit bored of my opening systems.

I like to play the high Chinese, so I thought: what if we expand the idea?

To make a Sino-Tibetan formation, you need:

  1. A stone in the O square

  2. A stone in the X square

  3. A stone on

Beyond that, the choice is yours.

1 Like

I just played the Manchurian fuseki against an 8k, in a game that reminded me of some AI insights.

Here, I expected the joseki wB+ C S18 O+ A 口 etc.

The bot likes, as I remember it does, wC+ B, continuing into wC+ B C18 A+ R13, establishing a group with right-side access.

What I found really interesting, though, is that it even regards wA C R13 O Q14 as an even game.

I’m wondering whether the meta in the Manchurian is properly R15 口, with R15 Q16 falling victim to the aforementioned C+ line.

I played the Manchurian system again today.

There are several possible continuations:

  1. D–B, or the Ussuri Line

  2. C–D, or the Argun Line

  3. C–A, or the Amur Line

Recently I’ve been playing the Amur, but I’m thinking of trying the Argun soon.

In the game, White played the vulgar exchange A–B. One might think that this sequence relies on a ladder and that Black must now play X and not Y, but actually Black can still play Y.

White still loses about 15% WR, but that’s the same as what the bot claims I lost from playing the Manchurian to begin with, so I can’t criticise. It’s only a point or two anyway. In any case, White should probably play C directly.


This ladder necessary for White to capture tightly. The game result looks better for White, though, even if he does have the ladder.


There are some complex bot variations possible, but the game could easily proceed simply like this:

That looks about even, or, more accurately, a little better for Black.

The game continued this way, though, with some mistakes from both sides. I was happy with this, noting that A and especially B are inefficiently placed.


Fun fact: the Manchurian was in Hurt / Heal Fuseki and finished 18th of 24, between the nirensei–nirensei splt and “Hosai’s formation” (which could also be called the “rotated sub-orthodox”).

Its SL page was created in 2001.

1 Like

Just curious, what’s the reasoning behind these names?


The version with a 5-3 point and high center stone was named the “San Andreas Fault Fuseki” by KoBa a few years back.


Manchuria is another name for Northern China or regions in it, developing the name from the high Chinese, which the Manchurian system is based on. That name isn’t mine; as I said, it’s existed for at least twenty years. The pun is that the “high Chinese” is “high”, or north.

“Sino-Tibetan” is a language group containing the Chinese languages. Similarly, the “Sino-Tibetan formations” encompass the Chinese openings.

The Ussuri, Argun, and Amur are rivers in Manchuria.



Map of the Sino-Tibetan or “Trans-Himalayan” languages, with the Chinese languages in red.


I stumbled (again) on this eighteen-year-old SL discussion of a 5-4 noseki in the Manchurian.

What I found more interesting, though, is what the OP said he usually played, at IGS 6k in 2003:


As a fairly long-time on-and-off 5-4 player, I’ve never even considered descending with (5). I play (5) as a tiger’s mouth or full triangle connection.

The interesting thing about this (5) is that it seems to be a stronger encouragement for White to play (6) in the corner, although Black is shown continuing locally with (7).


I found these cases in Waltheri of professional appearance of the full shape with (7):

  • Ito–Shuwa 1845
  • Hashimoto–Kanbara, date unknown
  • Iwamoto–Kitani, data unknown
  • Go–Sekiyama '32, played on an amost empty board
  • Inoue–Sekiyama '41
  • Kubouchi–Hashimoto '46
  • Suzuki–Hayashi '49
  • Ko–Yamashiro 2013

Here is the Go–Sekiyama game.


Agreeing with our Professor van Driem, the Institute of Linguistics in Berne prefer to call them “Tibeto-Birman”. :face_with_monocle:

Wikipedia defines the Tibeto-Burman languages as the non-Chinese Sino-Tibetan languages.

ie. the non-red colours on the map

1 Like

We’re fine with Trans-Himalayan as well, because Sinitic, Böd and Burmese seemt to be closer to each other than to the rest of the family.

BenKyo is reviewing the Go–Sekiyama game now.

This makes me think: if only the Sino-Tibetan formations with a hoshi are Chineses, that means that the others – like the Manchurian – are Tibeto-Burman.

Or just “Tibetan formations”.

1 Like

I played the Manchurian against @shinuito in a komiless correspondence game in the regulars’ handicap tournament.

He hit me with the surprising (16) Q6. I failed to find strong continuations like (17) P6 and (19) N5, instead playing the compliant (although blue) (19) O2, allowing myself to be squeezed.

However, my position would’ve still been fine if I had played away on (21) and permitted White to cleanly squish my cat. I made the spirited blunder (21) O6?, leading into the position on (29).

At this point I thought White would play the light (30) N8 (blue), but he chose the confrontational P8 instead, forcing the game into an exclusively tactical phase after my necessary cut at N7.

Suffice to say I collapsed in the subsequent fighting.

Especially if I had the position again I’d try (19) N5 N3 N8.

Continuing my riverine naming practice, I’ll call this the Tumen Line.

1 Like

(6) enters the Weldon Variation of the Amur Line, named after Alex Weldon of the SL discussion.

1 Like

On the other hand, Manchu is not part of that language family at all. But I actually like Sino-Tibetan in that context if it just means “related to the area where China and Tibet are”.

1 Like

10s blitz game with @8fledermaus8

In thie game we discussed (5) R17 P17, the Ussuri Line. It’s worth noting that by this point Black is already receiving less than 30% WR.

I’m considering next trying this “Sungacha Variation” (the Sungacha is a tributary of the Ussuri River).

White continued with a slightly slack joseki, allowing me to climb back up to about an even game. However, the position of Q8 had become dubious.

We played an anti-avalanche in the top left; I fixed with the thick tiger’s mouth and White invaded the Manchurian at the counter-Chinese point. It was still a 40–50% game.

The bot would like White to play at A here and crawl, but that’s not how I (at least) play Go. The established and human move here is the atari B.

This usually continues wB+ 1 2 3+ Ax, but fleder played the slightly worse wB+ 1 3+? ext. 2. I was nudged above 50% WR, albeit only for a short time. The focal point left above 3 was larger than either of us realised, and wasn’t played for some time.

The bot suggests that I can maintain 55% if I immediately turn there and make Q9 happy.

If the proper joseki had been played, I’d be at 45% since Q9 would be a touch redundant. This would still be a totally even game, of course.


I had the idea that if I were to play something like this, I’d rather Q9 be on Q8, as a “South Manchurian”.

Otherwise it might be too close to my influence.