What are the oldest written rules of Go that are unambiguous?

Go is an ancient game. For most of its part, rules were transmitted orally and ambiguous, depending on judgement to interpret corner cases. What were the first rules to be at once (1) in writing (2) unambiguous[*]?

[*] By “unambiguous” I mean not depending on judgement, instead always able to be applied by mechanically following the written text. Examples: The AGA rules, Tromp-Taylor rules, any of the Ikeda rules. Not examples: Japanese rules.

Ambiguous seems to be a subjective judgement, and there is bound to be disagreement about which rules sets that label describes

1 Like

It is not a subjective judgment. You may have that mistaken impression from a superficial reading of the rules text. When implementing them in a computer program it is evident that some rules are ambiguous. This happened for example, to Ikeda when trying to program Japanese rules. That is why he wrote this treatise on rules as well as formulating his own. That was before AGA rules and Tromp-Taylor rules existed.

The Kata Go documentation hints at this ambiguity too:

A hope also is that this document can serve as a reference for anyone to implement any subset of such rules themselves - including a rigorous nearly-Japanese rules that bots can use for self-play or for competitive matches completely without need for outside adjudication or dispute resolution or any other protocol besides just the bots making ordinary plays.

I believe the nearly-Japanese rules should correctly handle a wide variety of details, so long as both players play to rationally maximize their score.

The qualification of “nearly-Japanese” is because it is impossible to implement exactly something that is itself inexact.

2 Likes

Well, I agree that the Japanese rules are ambiguous. I’ve even created an entire thread diving into various technical cases. See Odd Cases 🤔 in the Japanese Rules

Frankly, it’s a bit insulting for you to assume that I have a mistaken impression or have taken only a superficial glance at these rules. However, your mistaken impression of me seems reasonable given that you’ve only taken a superficial glance at these forums.

However, what I’m trying to point out is that you won’t find universal agreement about which rules are “ambiguous”. There will certainly be many that assert otherwise for the Japanese rules.

Among the various rules sets that you are familiar with, which do you consider to be unambiguous? I think that this topic could generate a significant amount of debate.

I think this is also an historical question, which is tricky for me to answer. I’m not so familiar with all of the many rules sets that have been written down over the many years of history. It’s hard to say what even counts, if some written forms did not gain much prominence or were largely forgotten by time.

I would venture to guess that perhaps some form of stone scoring rules must have been written down at some point over the past few millennia, perhaps earlier than many other modern examples.

7 Likes

I found this article that gives a candidate for the earliest written unambiguous rules:

Karl Robinson and John Olmstead became entwined in trying to establish a rational set of Go Rules. They published A Rationalization of Go in 1941 and The Structure of Go in 1945.

I could not find a digitized version of either of these books.

Ironically, this thread is a bit ambiguous :yum:

It seems like you’d define ambiguity as “inability to implement a ruleset on a (real) computer”. However, I’d argue that one could design a ruleset that is both rigorously defined, and impossible to implement.

For example, we could use this scoring rule:

After two passes, the score is given by the number of stones each player has on the board, given perfect play

This, I think, is both rigorous (it’s straightforward to count stones, and theoretically possible for a timeless god to iterate the game tree), yet also impossible to implement - the game tree is simply too large for any real computer to compute.

Another definition of ambiguity might be “having a situation where two rational judges may disagree on status” (e.g. L&D or legality). I’m not sure how the traditional rulesets perform by this criteria - it seems to mostly depend on how rigorously “living group” has been defined.

I’d be interested to see a scenario in Japanese where the status is up to interpretation, given that you both agree it’s ambiguous :slight_smile:

3 Likes

I’d place my bet that the first rule that fits those requirements is “the stones are placed in the intersections of the grid”.

4 Likes

I could really write a lot about this, but I’ll have to answer briefly due to lack of time and being on my phone.

I think that there are various cases that involve the “enables” clause that are unclear, even for rules enthusiasts that relish analyzing complicated rules beasts. A lot more can be said about this broad class of rules beasts, but I will point out a specific form of these discussed in this thread: Is white alive under Japanese rules? - #11 by yebellz

The particular example in the specific post that I linked highlights how resolving life and death status can involve computationally intractable problems (such as figuring how big a vacated space needs to be to enable one to play a successful, living invasion).

Another example of difficult interpretation is the psuedo-seki position given in this post: Odd Cases 🤔 in the Japanese Rules - #22 by elsantodel90

This latter example also touches upon how there is an ambiguity of authority, that is, the written Japanese rules are not definitive and the ultimate authority is decisions (for disputed ambiguities that arise in professional games) made by the Nihon Ki-in.

You specifically mention “rational judges”, but who/what exactly qualifies as such? It seems reasonable to suggest that OGS moderators are “rational judges”, but there have been differences in interpretation (usually resolvable by realizing that someone made a mistake) in even simpler cases (such as misjudging a bent-four-in-the-corner situation).

6 Likes

Could we “fix” the territory scoring rules in the following way?

Play continues until both players wish to score the game and agree on the live / death status of stones, or until both players pass, in which case all stones are considered alive.

Edit: Scratch that, it only works for Area scoring rules …

Edit Edit: I guess it might work if the game is counted as in Tromp Taylor Rules in the event that players don’t agree on live / death of stones.

I have more research on ancient historical principles and customs, instead of modern rules definitions and rigorous computational implementation. But I do know trying to turn customs into rules happened during the late 19th century to early 20th century when there were several organizations and associations existed that had different customs, and when the Nihonkiin established they they needed to consolidate these differences hence the beginning of codify customs.

The introduction of Go to a larger crowd outside of East Asia also happened at the end of the 19th century to early 20th century, but a lot of them at the time, still looked up to the Japanese customs as the template (Nihonkiin hadn’t even officially codify them until after WW2). Hence, people are free to write any rules or customs as they see fit (hence the period in China where the current Chinese rules started to get codified in the 1950s to 1960s, and the Korean players brought back Japanese rules. And as far as I know AGA rules and others were matured after the wave of foreign tours by East Asian players since the 1980s, and very recent events. None of them were designed with the computational implementation in mind at the time.

If we really want to go far back in time, the pure stone scoring rules are probably the most unambiguous and easy to implement by computer. Literally, just count the stones, and everything else depends on players capturing and life or death on the board. They don’t even have superko or any other stipulation as far as we know (there were ancient texts, that mentioned not over 3 times for an action called 炮碁, which might be related to repetition or capturing, and might related to ko, the meaning of the custom is still debated, but simple ko rule might be the only common custom at the time).

Edit: I need to go and check on Ing’s rule papers from the 1970s and 1980s. I remember seeing paragraphs of Ing Chang-ki discussing rules with foreign players and professors and edge cases. And there might be some historical clues and references still left from the rules actually used outside of East Asian rules.

And here is an auction page of the 1941 book Rationalization of Go that had axioms and theorems in the preview

And I don’t see it as an implementation of a rule though, more like studying set theories.

1 Like

Not sure as you can play go in squares. Maybe capturing by surrounding is the first rule

This depends on the level of correctness you expect from those rules - how well should they correspond to the actual go game that people play?

Without any such requirement, i think FIDE had written rules dating back to the beginning of the last century. Oc they are a bit lacking in correctness - understandably since they were written as chess rules - but for an “anything goes” go ruleset they still fit the bill. :slight_smile:

If otoh you mean rules that describe real go as it is played, then the safest answer is probably “not yet”.

1 Like

Those are interesting and illustrative cases, thanks!

Perhaps “rational” is not the right word, but what I was trying to get at with the term “rational judge” is that external factors like time or skill would not play a factor. If a moderator is not able to recognize bent four-in-the-corner, then I would not consider it a rational judgement.

The idea being that two judges should be able to independently reach the same conclusion, given a complete understanding of the ruleset.