A hypothetical precursor game rules of ancient Go

I’ve read in the book The Evolution of rules in Go, hypothesis that Go originated as a type of AtariGo, and where a group only required one eye to be alive.

But I still feel those are still more rules than necessary. Hence, I’d like to propose a simpler ruleset that required very few assumptions and priors, even without rules regarding territory, but still looks like it would naturally evolve to Go as we know it today.

#1. Each side places a stone on the intersection of the board. One after another, and no pass.
#2. When a stone is placed it cannot be moved, a stone has 4 liberties and shares liberties with neighboring stones.
#3. When a group runs out of liberties, they are flipped upside-down and are considered captives of the opponents. but are not removed from the board, and captive stones don’t reduce the liberty of a group.
#4. Play until no spot is left open on the board, and the game ends.
#5. The player with more captives wins the game.

These rules have some very interesting effects.
1st, a game will always consist of exactly the same number of moves as the number of intersections. No more, no less, since no stone is taking out of the board, even the captives.

2nd, suicide is allowed by default (although players wouldn’t have the incentive to do so early on), but sometimes might be forced to, like if it is the last legal move on the board, and have to be played since there is no pass.

3rd, at the end of the game, if every spot is filled, except each side has just one eye space left, then the second to last move for that player is also suicide inside the opponent’s eye, and give one captive, in return the opponent is also forced to suicide give one captive as well. And the game will end in a draw.

4th, there is no ko of any kind existed, and a group will be completely safe once it has captured at least one stone inside. We can call it the “one captured eye area is alive” rule. It is not manually defined but derived from basic rules.

5th, creating separate eyes is crucial even if they haven’t captured any captives yet. It’s not because they make the group alive, but because they are captive traps, where your opponent at the end of the game has less “open area” than you, every move afterward is either filling dame points (space not in safe space, and not surrounded), and then backfilling their own areas that still have liberties left(and yields no points), and finally forced to give captives (to prevent suicide, and giving huge amount captives to the opponents).

6th, A safe “eye space” don’t have to be one intersection, as long as it is not sufficiently large enough for your opponent to create at least one “captive’s trap”, it is like a “safe space”, and can be left open till the end of the game. (and due to this, a seki with both sides has one eye in modern go would simply be alive, in no eye seki the one to play first loses, one side has an eye always win)

7th, when each side is left with only “safe space”, the best course of action will be backfilling your own safe space, instead of put stones in your opponent’s safe space and making it alive by default.

8th, experienced players don’t need to fill every space to be able to tell who wins the game, since it is essentially the same as a settled game in modern go. Fill the dame first, then similar to the area scoring procedure to fill in your own territory. And it can be done with area scoring as well, and there is no difference between these two “methods”.

9th, Players who have good global position sense still have the upper hand, while handicap stones cause a severe penalty. And you would want your groups to connect together as much as possible, and cut off your opponent’s groups, since in the end-game backfilling, your opponent needs more stones to form boundaries, and have to sacrifice more captives in your safe zone to prevent suicide.

10th, Due to trying to connect groups and cut off opponents, fighting would be even crazier since it would be much easier to live in a very small group. And 2nd line stones will be quite favorable since one eye on the side first essentially make it alive till the end game.

I think it is simple enough to be learned quickly, almost no ambiguity, a fixed amount of total moves, and yet interesting enough to be fun. And it shares the basic principle where the one with the larger “area” wins the game like Go, which could easily be converted to other variations of Go, by introducing the recapture area rules.

I wonder what would be the game of this precusor Go would look like if we just follow these 5 rules only?


Wait, doesn’t Atari Go have less rules than what you proposed?

#1. Each side places a stone onn the intersection of the board. One after another, and no pass.
#2. When a stone is placed it cannot be moved (arguably this isn’t really a rule so much as the lack thereof)
#3. When a group runs out of liberties it is captured (whether it remains on the board or not is irrelevant as the game has just ended)
#4. The first to capture wins.

The issue is ofc this is already vague (e.g.: it doesn’t describe which stone gets captured in the situation where a stone is played such that 2 groups don’t have liberties, and doesn’t describe how groups of stones work), but under normal atari go rulesets one eye (that is at least 2 spaces big) is already enough to be “alive” (or uncapturable – with the exception of certain seki situations like the 3-in-a-row with throw-in nakade)

1 Like

I just realize that rule #3, flipping the stone to mark as dead, still existed as a practice in modern scoring, makes dead stones easier to spot, and Chinese stones have traditionally been uneven (one convex).

It might be as a relic for this kind of “no removal” rule, not just it is more stable on one side. And it would make a lot of sense to use flat wood pieces as “stones” and can easily be painted with not just black and white surface colors, but also other colors for their flipped sides.

Yes, it will be easier, but not in the same spirit as the one I proposed, where the “size of safe zone” analog to modern Go is crucial in winning, regardless of who gets the first atari.

AtariGo or other kinds of capture games might be related to some kind of shaped arrangement games, where stones are placed in certain arrangements to “capture” enemy stones (like Fangqi - Wikipedia), atari is just one type of shape. And most of these arrange games don’t even have once the pieces are placed can not be moved rule.

I suspect they are the cousins of Go, instead of derived from Go, and stem from some common precursor games. The rules I imagined is something that started the branch to become the “surrounding game”, even though it has the winning conditions of who “kill the most”. Who owns a larger area to win is derived, not defined by these rules.

I’ve thought about this too: how to make the ‘point’ of Go more intuitive.

My suggestion would be the following:

  1. The goal is to place as many stones on the board as possible -> the winner is the one with the most stones on the board.
  2. Each player places one stone on the board, one after the other, or may pass.
  3. If a stone, or group of stones, have no liberties, they are captured at the moment of placing a stone, and removed.
  4. The game must go on, or stop (no repeating sequences of boardstates)

From these three rules, and objective of the game, we get a lot of Go.

  • Territory is ‘safe space’ to put your stones later on. It’s obvious that more safe territory is better.
  • Eyes makes the territory safe, but as one quickly learns, you need two eyes.
  • In principle it is better with one large territory than many small, as each group requires two eyes to not be taken in the end.
  • KO, and end of game through passing, is resolved by the third rule.
  • Even modern scoring is almost a derivative of these rules. Counting territory is just a shortcut for how many stones you can put in.
  • Being efficient in taking territory (fewer stones for more area) translates to larger territory for the same number of moves.
  • Captures may need additional rules. If prisoners are returned after the war (nice treatment of POWs), one could also count how many stones you have left in your bucket. The fewer the better. Alternatively, prisoners could be kept, and a scenario exists where one player runs out of stones completely, at which point they are dead.

From an abstract point of view, it’s all about having more options than your opponent. Each available spot on the board is an option. The first to run out of options loses. Life and death.


This is actually what was done as a historical precursor to area and territory scoring. It is known as stone scoring. It makes for very simple to define rules, and behaves similarly to area scoring, except that it has a group tax.


I’ll take this opportunity to mention this page, where you can play against CrazyStone using stone scoring:
(scroll to the bottom)
It’s a good way to introduce beginners.

I agree that this more intuitive way is better for beginners to learn to play without complicated rules for counting and scoring after the game since it is part of the game itself. And as @yebellz and @le_4TC pointed out, from the historical records, we know stone scoring was used and their derivative methods of scoring were still in use all the way till the 19th century in China.

It leads to a style of play more aggressive in early games because cutting off your enemy groups make them “pay” extra eyes for that extra group to stay alive on the board in the end.1 group required 2, 2 groups required 4, eye space under this rule is a necessary evil to be avoided at all cost, except at the cost of losing groups, which also lead to different dame stones and diagonal connection could be considered ko threat if groups can be cut off into disconnected groups.

I want to point out that under the exchange captive rules where a “pass move” required extra captive to be given back to the opponents (or if you don’t have any, given one of your unused stone to be captive, since it is an “option” you forgo, under the rule of both sides has to take a turn to play), would result in a game where counting the captured prisoners yield the same results as counting who has more stones on the board. Since each player has to play a move, and if each has the same amount of stones, the one who has more stones on the board equal to who has less captive on the enemy side. And a game can be stopped early if both side only has stones half the amount of intersections. A game will never go beyond the number of intersections, where the one who runs out of stones first would lose by default. Both sides don’t even have to count to tell who wins.

This intuitive way of having more stones on the board has a drawback of games last much longer than necessary. And when the strength difference is very high, the much lower-ranked player would constantly throw in their stones into the enemy “safe zone” since they don’t know if they can live or not, and want to have more stones on the board (historically there was even a name for this behavior - 炮棋, throwing stones as projectiles beyond the border, a very vivid description). And if they have some ko exchange earlier, the stronger side might run out of stones to kill those throw-in stones when the weaker players just refuse to “captive exchange” to keep “dead stones” on the board. The captive rules have to be agreed upon as part of the rule, where both sides have regular exchanging captives of equal amount, till one side only has zero, or when one side runs out of stones, the exchange is mandatory. And I think it has a very deep root historically since it has always been called a “civilized fight” (君子之爭 fight between civilized/educated people).

After digging a bit deeper into games might historically related to the precursor of Go Some musings on a "reverse" ko rule - #14 by claire_yang, I came to a realization that what we considered as intuitive and as basic like the moment of removing “dead stones” might not be that clear if the original goal of the precursor games are about capturing in the first place.

I feel it is very important to think the acts of “make a move”, and “remove stone when it run out of liberties” are in effect two “actions”. If remove stones that run out of liberties is an action has to be done by the opponent, suicide as we know it is an act to waste the action of the enemy, and would be part of the precursor rule.

It also worth pointing out that the reason I believe the “dead stones” didn’t get removed from the board is older, is it meant there is no “secondary game” under the table, or outside the board in this form. The “captive jar” is in plain sight on the board and part of the game with no additional “captive jar” required, and players could not have cheated if they want to (like hiding some stones to begin with), since the board has to be filled by default, and it would be very clear to know if someone hid stones and not play it.

I also think that when enough of these games being played where stronger players start to realized that they don’t always have to play to the end, and by counting the safe zone to know who has more hypothetical captives to end the game. And this “quick counting method”, might be why we have two-stage of the game, playing and scoring where the latter half originally was just convenience and more advanced gameplay reserved for the “elites”, who are good at math and have extra tools to “record” and count the hypothetical captives. We need to remember digit numbers themselves are inventions appear much later in history, and for a long time in history, education like basic math is also only reserved for elites, and require some extra tools like counting rods or abacus to get it verified.

The comparison of who has more captives with rules required a full board is much easier even if you don’t have those tools, since players just need to check who has a longer captive lineup with intuitive counting.

So I believe the precursor of Go without filled captives quickly diverge from the original full board when they were played by the elite, and then been taught like a type of different games only to nobles at the time, who has the means and knowledge to carry on the “scoring phase”, which is in itself a game of sort, requiring to imagine the game to be played to the last, with extra tools to get it right, and like most modern “game expansion pack”, it doesn’t change the game at all, but only some superficial add-ons to attract “wealthy players”.

1 Like

could not have cheated if they want to

It’s a bit speculative to assume that the rules were such that cheating would be minimized.

Anyhow, I don’t know so much about the precursors to Go, but a few general notes about pretty much any feudal society: the nobles didn’t want to do what the peasants did. So, a noble’s version of Go could and perhaps would include counting, flipping stones, capture jars, and so on. A peasants version of Go, if such a thing existed, would be centered around ease of play and understanding I think. Flipping stones require fancier equipment, so I think that’s out. AtariGo is nice in this way because it’s immediately clear who has won, and it’s more exciting as a spectator sport perhaps. Perhaps even involve rudimentary gambling. Problem with modern Go is that the skill difference is so big between even a few stones, that I think peasants would not play this form because if you have time to study the game a little bit, it will be very clear very early that you’re not playing a fair game. An AtariGo variant makes the playingfield a bit more even, at least on the surface level.

One could however imagine a variant played for money or some resource, say grains of rice or coins or some such. Whatever you surround you get. And you cannot pass. Getting territory would be crucial. Eventually one wins the board because they were more effective in their use of stones. Then you only need a rule for KO, no prisoner jar. Maybe even handicap was introduced as an idea to convince stupid players to play the hustlers. Though this requires people to have enough of whatever they’re gambling with.

Still, I think this is too complex, if a precursor game was played by the commoners, though I see that it is speculated that one of the most ancient boards they’ve found was found in a watch tower, indicating it being played by soldiers.

Very speculative idea: what if it did indeed come from astrology as some speculate, or more specifically, from divination? Throw some black and white stones on the ground, remove those surrounded, and see which color is most dominant. Then, if they’re unsure if something is surrounded or not, just move the stones a little bit according to some rule.

1 Like

I find this questionable, as with both Chinese ruleset and Ancient Chinese ruleset, the captive jar is irrelevant for scoring (and I hear it is quite common to just dump it into opponent’s stone bowl in China)

I don’t think flipping the stone is actually required if it was played by peasant. Here is a picture of a modern variations of a shape rearranging game still being played in north-west region of China.

The “equipment” required can easily be much simpler, where the precursor don’t even need a game board at all, the only things players need are pieces for two sides than be distinguish from one another, and then draw lines on the dirt ground to start. And upon out of liberates the dead stones are immediately taken out and line up on the side (which is still be done today for intuitive counting, as seen in the picture). And players can just dig a hole on the ground, or fill it with other objects (like twigs, or anything) to represent a spot has previously been played, and no one could play there again.

The mention of “captives” (俘子) dates back much older since Tang dynasty counting method, which is actually the origin of the Japanese territory scoring method (數路 count the road), where the records clearly showed the scoring method required the players to keep captives and fill in the other side, and then compare the “road/empty spot” left behind, but with extra eye points for each “block” of group deduced from the “road count”

The comment on the record shows
共二百四十三着 黑杀白六子,白杀黑六子。 黑有四十路,白有三十九路。
Total 243 moves, Black killed/captured 6 white stones, white killed/captured 6 black stones. Black has 40 “roads”, and white has 39 roads.

Modern scoring will tell you there are 52 black territory, and 51 white territory, fill in the captives, there are 46 black territory, and 45 white territory, but black and white each has three separate groups (the dame point that can connect groups don’t need to be played), so each reduce 3 times 2 eyes 6 territory, and we get black 40 roads, and 39 white roads.

Clearly a part of this method is transformed into the later Japanese territory scoring, by omitting the “reduce eye space” rule. While Chinese scoring took a different turn and kept the reduce eye space rule as group tax, and switch to area scoring, since it would only require one side to count where every dame points have to be filled (some believe is the anti-cheating incentive to promote this change to area scoring, since captives are no longer required for scoring).

I don’t think flipping the stone is actually required if it was played by peasant. Here is a picture of a modern variations of a shape rearranging game still being played in north-west region of China.

Very interesting picture :slight_smile: (how do I attribute a quote to you? I only get the text itself)

So a precursor could simply be, as you argue (if I understood) - play until no moves can be played, mark captured stones, the one with the most captured stones win.

I still feel that marking captured stones, even if by marking the ground or using another type of stone, or flipping the stone, is a complicating factor. If this game was played by noble and peasants alike, then ease of play is paramount. It is always doable to find some workaround, but, makes it harder. It might be the case that it was as you say, but I’d personally prefer if this part could be taken out. It seems more likely to me at least.

So if the nobles who can devote more time to the game and perhaps also can do simple arithmetic, they stopped playing until the board was full. Perhaps as a sign of ‘skill’, i.e. they didn’t need to play it out. But the leap from not being able to play on conquered land, to being able to play there seems a bit big to me. It seems easier to just have a rule like, play until you can’t or won’t play anymore, most stones win, stones without liberties are taken out. Seems much more straightforward. And let’s say there’s no KO rule, which could lead to a stalemate. Well, perhaps it did? It’s hard to create a ko so fundamental that the game hangs on it, if you’re a noob like me. The experts however, might have devised such a rule for tournaments to avoid just this, or maybe not. Chess after all can end in a draw and has always been so as far as I know, and komi in Go is quite recent. But, thinking about it, I think the ko rule would pretty quickly be adopted as it would be frustrating to have a good game going and then the other refuses to give up the ko.

In other words, I don’t see how taking out dead stones and playing until you can’t play anymore (first to pass loses basically), is less likely precursor than leaving stones in? Seems like rules anyone could learn and understand. Then, on their first ko, they learn that it’s not allowed because they’d be stuck there and that’s no fun. From there, it’s easy to see how territory became a thing. And captured stones is in a way a reflection of territory. If the same number of stones are played, then I don’t know if there is a situation where the winner is not the one with the most captures.

If one has a rule, first to pass loses, stones must have at least one liberty or else be removed, you must play a stone that changes the game. This ruleset takes much time, just as stone counting and capture counting does. But, as the smarter players could see that playing in the opponents territory would mean capture, they would play in their own safe havens once there was no places to expand. This took shorter time (no need to think), but still too long. Then they figured they could count captured stones, territory, or stones on the board, if both players agreed to stop playing before one player could no longer play a legal move.

EDIT: Thanks for an interesting topic. Sorry if I seem to ‘attack’ your opinions, I’m merely thinking out loud.

You select a paragraph of the text, and a Quote option will appear

Let me describe the game they are playing in the picture, a type of Fang-qi. It’s a family of game variants not one specific game, a game about trying to form certain shapes. The simplest one being a “square”, 2x2 solid 4 stones in a square form (hence the name).

It has even simpler rules where there is no “liberty” to “objectively” determine which stones are dead, but they are subjectively selected by opponents. The goal of the game is to form certain shapes. And the board size doesn’t matter (don’t even need to be an odd number of the same number, it can be 5x5, 5x6, 5x7, 6x6, 7x8, etc. depending on the regions, or as long as both sides agree). The following is the general rules of the most basic kind, and existed in some forms or another across all of them.

Each side has half the amount of pieces to the grid numbers (if it is an odd number, the first one to play gets one extra), and then each player takes turns to put a stone on the grid. During this first phase, players can place their pieces wherever they like (even just randomly), but normally it would be ill-advised to try to form the shape at this phase (the reason will be explained later).

The second phase starts by picking a certain amount of pieces of your opponents with equal amount and remove them from the board (if it is an odd number of grids, the one that has one less piece can remove one extra to balance the pieces left, the amount is also varied from different variations and place to place). And the strategy of this phase is obviously to break apart the opponent’s shape, and this is why you’d try “not” to form shapes at the initial placing phase, so your opponent will have no priority to break them apart at this stage, a cat and mouse game.

And the final stage is played where each side moves pieces still on the board, to form the shape. The rule of how a piece can move and how far, from jump across pieces, move in a line, move diagonal, all kinds existed in different local variants. Every time a shape is formed, one of the opponent’s pieces of choice is picked and “captured” to be removed. The pieces that haven’t formed a shape need to be picked first before shape formed pieces, otherwise, the game will go on like a random move without strategies. The game ended when one side can no longer formed a shape. Usually there is no need to count captives, since the last one to form a shape will usually have the advantage to kill off all the opponent’s remaining loose pieces that haven’t formed a shape.

As described above, these rules seem pretty raw and flexible. And they are so widespread and diverse in many different places, it is highly likely they are very old, and morphed to different local variants over a long time with their highly adapted rules, low requirement, and lots of flavors to be added and evolved.

However, my interest lies in that there is an extra method in some of the varieties, and is related to the topic we are discussing here - “marked pieces”, especially marked captives. This extra step method is applied during the initial phase, or more precisely it is a merger of the first phase and second phase, by marking some spot on the board to be “dead space” where neither side can place a piece there (even an entire line can be marked as forbidding). The purpose of this method is very simple - speed up the placement and removal phase, and go directly into the fun moving and capturing phase (instead of placing till the board is full, and then removing. Doing two things at once).

It is easy to see where I drew my inspiration for my hypothetical Go precursor. It is essentially a very special variant of this type of shape rearranging game that omits the last moving and capturing phase. Think about it, on a very small board, if some of the early variants in the ancient time applied the speed-up marking captive method and repeatedly played by many players, and then as players got stronger and better, they would soon realize that there are only so many different board positions after the removal worth playing, and most don’t even need to play out the moving and capturing phase to be able to tell the final outcomes. Hence, the logical next step, is to increase the board size to have more variations and positions, and maybe keep adding more and more different shapes, until someone thinks of a way to “objectively” picked those “marked spot”, in order for the game to be on-par in positions and competitive even for the strongest players to have fun.

Finally, at some point, it ended up evolving into a game where it is more and more about the beginning strategy phase to gain an advantage than the later part of moving and captures, since the latter part will be long decided in the initial phase, no matter how hard the side with “losing positions” tried to capture stones later on.

I would even postulate that ko fight is essentially another modification of the second phase and the speed-up captive marking process, defined as moving pieces to previously held captives’ spots in limited fashion, most likely some of a mixed of the placement phase and moving and captures phase. It would be like trying to merge all phases in to one, with minimum number of steps as possible. The initial rational would be as long as they have potential to form more shapes, some “loose pieces” could be “sacrificed” inside enemy strong positions to have a better chance in advantage positions elsewhere. It would be the same reasoning as modern ko, just under different rule sets with different definitions of “solid/alive shapes”.

Don’t worry, I like to take in different perspectives. And some of your thoughts are very close to many others speculate of how Go evolved in a relatively recent time period, before the 5th century AD. From historical records there is a word for what you described of playing until both sides couldn’t play anymore - 溢, literally translated as “filled”, and we know that this method of “scoring” still existed up till 5th to 7th century AD, before the Tang dynasty “counting road method” become mainstream (in modern view, a hybrid of territory scoring with group tax). And from records, we also know that ko rules before that period of time, are not “fixed”, since they talked about player need to negotiate rules they agree on, related to how they can draw in pieces into opponent’s “territory”, even how many times it can be done (strange right? limited how many ko fights can be played). And there was even some form of komi (a very odd komi system, with a group of 3 count as one unit, where one side need to pay for the other side in unit, 3 units needs to be pay to offset first move advantage, around 6 to 9 komi), and tie breaking rules to prevent stalemate. It is already pretty mature around that time, but rules seem to still varied in different places and haven’t been unified.

There are recorded about “palace game” (宮棋), essentially refer to games played by nobles, maybe even kings’/emperors’ wives, concubines, servants played them. And some are confirmed to be some types of shape rearranging games, some are chances games (but using Go board, or other board games board, by rolling dices or even throw the pieces on the board to generate random moves, etc.), but some are definitely variations of Go, since their description fit, and some kings/emperors themselves are known Go players. They definitely have the means and efforts to make them as complicated as they like, or even made up rules just as they feel like to. And from archeological discovery we known other nobles also played Go at 2nd century AD with earliest known stone pieces from tombs, and clay/ceramic board at 2nd century BC, and not from tombs, but broken slabs buried under walls. It’s very likely Go is already in the form we known at that point, and its root needs to go much further in history.

From games like Tibetan Go, we know there were special Ko rule, only allow ko after a “cool down” period, essentially make the captured area forbidding for a period of time. And they have strange rules like pieces have to be formed to certain shapes before they can move in, or those shapes can gain extra points. By captures stones on certain spot of the board can yield extra points (even if those stone trying to capture them ended up dying), that is different spot has different point value. Under these weird rules, it seems likely the logical concept of equal moves are not the most important things to be consider about early on, not even “ko” itself is a set rule. And the goal is more about the act of capturing, and creating shapes than encircling territories. And there are many taboo moves, where some spot or shaped cannot be played at all, and has nothing to do with ko or liberty shortage. It feels like some rules are missing, and leaving a relic behind.

All these left me wonder that exactly were the evolution phase of Go, and the curious case of ko rules, which even modern rules cannot fully resolve them. And I have a more or less very plausible path from shaped rearranging game to a version of my hypothetical precursor. But still unsure about how ko rules evolved from that proto-game to what we know in the historical text, and today.

If my assumption is right (see the details in previous reply), ko might have nothing to do with preventing the same position being repeated, but has a deeper root related to how the proto-game was played in its primitive form. And the act of removing captives, might have been a 2 step process. That is, the captive still stay on the board even if they run out of liberties, and the opponent has to use one extra move to “remove” them. Before captives are removed from the board, they would still be considered “existed”, but in limbo, where it could lead to some very interesting encircle from both inside and outside atari that current rule won’t allow.

Interesting. You’ve researched this a lot. I don’t think I’ve got more to offer, besides, one thought that came up while reading your latest reply. Given all these different rules, considering that in these times people generally didn’t travel as easily as now. One can easily imagine that many variations of the game was played different places, and one must agree on the exact rules beforehand. Just like in a lot of card games. We remember those scenes in old westerns: “5 card stud, jokers wild, 3 on the table, river doubles, and no swaps” and everyone are on board with that. It could be similar with Go.

So what is the minimal? A grid and two sets of stones seems to be the same. Liberties seem to be the same. Capture and removal seems to be the same (besides your hypothesis). Then they can differ on whether pass is allowed or not, if there’s group tax, if it’s territory or prisoners that matter, special win loss conditions like ‘first to pass’ loses, ko rules, etc… They would all create similar games in terms of playing, but the strategy would be different. But, leaving stones on the board seems to me be the one rule that makes the game most different. Perhaps some played this variation too because it would be faster.

I think in exercises like this it’s worth it to look at how much a rule change changes the game. It’s not a given that a precursor game was ‘simpler’, but any given change to the game should be ‘small’. Take the history of chess rules. While not as ancient, it’s gone through many iterations, in many cases making the game simpler. For example, the queen at some point could move 2 spaces in any direction the first two times it moved, then only one space in any direction. That’s more complicated than today. Winning could also be done by capturing all your opponents pieces (minus the king) which is also an additional rule, and pawn promotion was at times a more complicated matter than put in whatever piece you feel like.

So I would think that, if a game change messes up the applicability of a persons skill at the game, then it’s less likely that that change was a common variation. Poker is poker is poker, but there are many different ways the cards are dealt, swapped, etc. This changes the probability of things, but not the fundamentals. I still know what’s valuable or not, who wins, and so on.

Thinking about it, leaving captured stones in wouldn’t change the game all that much. Only ko’s and snapbacks, also some cases of false eyes. But, if you’re someone who learned to play with captures staying on the board and meet someone who doesn’t, the whole idea of ko’s and snapbacks is foreign to you. These players would be in for a surprise or perhaps would not understand this variation. The other way around is much more general. I could easily play a game with captures staying on the board and still know what’s going on, though some strategy might not be as useful anymore.

However, a benefit of marked prisoners is that you can play with only dirt and a stick, making marks in the sand as you play. One thing worth considering too is, did they always play on the intersections and not inside the squares? It’s much easier to play a variation of Go on paper (or dirt) if you use the boxes. Easier to erase or alter marks without messing up the board.

There is a clue in lexicography from an ancient dictionary written around 1st century BC, where it is the oldest text we know started to use the word 圍棋 (but written as 圍棊) to refer to Go. And it seems to be a new word or regional word since the description of the word is only one line - 圍棊謂之弈。自關而東齊魯之間皆謂之弈, translated as Go is also called yi(弈), in the eastern part of China, they all call it yi 奕 in two of the ancient powerful states Qi 齊 and Lu 魯. These states existed since the very early day of the Zhou dynasty before the 1st millennia BC and were high nobles. Confucius himself came from one of them, and Confucius was also one of the earlier scholars to mention yi 奕 in his work. So we are fairly certain at the 1st millennia BC, Go or the precursor that continued to use the same word for nearly a thousand years is some kind of game that high nobles were playing (but perhaps only popular in this part of China). And in old Chinese literature, if something only uses one character with a simple form, it usually means it has a very ancient origin rooted in old Chinese language, where double characters word like 圍棋 are used when people found something similar and added an additional word to distinguish them (like chess, and Chinese chess).

So, we know there was some kind of games called qi 棊, that existed at the time, which is slightly different than yi 弈. And people added the word “encirclement” 圍, to emphasize the difference. Hence in 1st century BC, it was at least a game about who gets a larger part of the board. And it begs a question, is it possible qi 棊 in itself actually wasn’t a game about encirclement?

There is a second clue right before the yi 奕 entry - "簙謂之蔽,或謂之箘,秦晉之間謂之簙,吳楚之間或謂之蔽,或謂之箭裏,或謂之簙毒,或謂之夗專,或謂之𠥙璇,或謂之棊。所以投簙謂之枰,或謂之廣平。所以行棊謂之局,或謂之曲道。 "
I won’t be translating them all, but the long descriptions basically boiled down to a list of words that all refer to the game 簙 in different regional dialects (蔽,箘,箭裏,簙毒,夗專,𠥙璇), and the most interestingly - qi 棊 as well. This means some people (probably not many since it is listed in last) also used the word 棊 for a different type of game 簙. It gets more interesting that we know the Imperial families of the Han dynasty (the time period where this dictionary is composed) came from a place called Song 宋, where it used to be a minor state. But Song got conquered around the 3rd century BC by no other than its powerful neighbor - the state Qi 齊 (one of the two powerful states I mentioned above where people call the game of Go - yi 奕).

So the most likely scenario is that Go in its precursor form (no matter what it was) was popular in the eastern part of China in these two powerful states for ages. At the same time, a family of games (possibly way older) was also popular throughout China but already split into various forms in different local states and take roots in regional dialects. Where one of the state Song 宋 has a variation called qi 棊. When the Song state was incorporated into the neighboring Qi state, people started to borrow words and adapted Qi culture, and start to play the imported yi 奕 game. But in oder to distinguish them, they used an additional word to describe yi 奕 as 圍棊. Luckily, the people from this region gave rise to the family later forming the powerful Han dynasty who speak the local dialect, and brought the game to their new center of power in the western part of China, and spread it across Han territory.

However, we are leaving with one question - what exactly was this widespread family of games (簙,蔽,箘,箭裏,簙毒,夗專,𠥙璇,棊) about, and makes the people from Song 宋 used it as an analog for yi 奕? We have some clue about what it is - It was some kind of bo 博, later a variation called Liubo 六博 became very popular throughout the Han dynasty as well. But we are not sure the reconstruction of its rules is complete, and how much was it differed from the older bo 博, and how much regional difference between each variation. Although, they are definitely popular enough to be used together as 博奕, representing boardgames in general.

The strange thing is that we have archeological artifacts for Liubo 六博, which have gameboards (it has two boards, one for moving the pieces, one for the rods related to keeping scores and using several types of dices/rods) that is very different from Go, and it is a game involving chances, and generally considered as a type of mixed racing and capturing game. Although its gameboard will immediately remind people of the board used in Nine men's morris - Wikipedia, and also astrology maps (I think this is the reason why Go seems to be associated with astrology often and it might still be true). It is possible that the precursor developed in Qi and Lu states were modified from a common ancestor game that diverged into various kinds, and kept as peasants game in the form of bo 博 due to its gambling component, while yi 奕, popular for the upper-class evolved in a very different direction, and became more sophisticated.

One final thing very interesting and worth noting is that there was a third type of game called Sai (賽 I swear it’s just coincident), that was often associated with bo 博, and the difference seems to be that Sai 賽 doesn’t involve the gambling/chance component, but was purely a game of skill. There are some debates as to what it was, and there is just too little evidence to know for sure. I suspect that it was probably a kind of shaped rearranging game similar to Fangqi which has some common roots with bo 博, and split off earlier from some primitive games (maybe a type of merels/morris) where their gameboards can be all kinds of sizes and designs (astrology cultures could be mixed in before this early stage or even possibly be the inspiration of it). The variations involved more with moving and capturing became the skill part of bo 博, while the capturing and shaped rearranging components were focused in Sai 賽 (probably with different types of boards, and some used grid boards evolved into Fangqi today).

Hence if we trace back the origin of Go, one likely path might be this.

  1. Thousands of years ago beyond recorded history, some astrology concepts inspired or got mixed in with a primitive type of merels/morris games.

  2. It split off into the precursor of Bo 博 with components of gambling, moving and capaturing, where the precursor of both Sai賽 and proto Go focused more on shape arranging, strategy placements, moving and capturing.

  3. The precursor split into Sai 賽 which focused more on shape arranging, moving and capturing (and later on continue to evolve into Fangqi), where the precursor of Go - yi 奕 focuses more on strategic placement due to being introduced to wealthy upper class, but still maintain the capturing and possibly some moving components, where people hundreds of years later, still were able to recognize its similarity to a variation of 博 - qi 棊.

  4. During the Han Dynasty, the alternative name of yi 奕 - 圍棊, maybe a variation of yi 奕, were popularised by the imperial families and maybe kept refined its rule to be more generalized, and possibly focused more on encirclement, and loses the component of moving, and has some kind of ko rules that were still incomplete.

  5. During the first half of the 1st millennia AD, 圍棊 started to become 圍碁 Weiqi, where the stones no longer used wood as common materials, and as it spread to different regions, different ko rules, scoring rules, even suicide, forbidden spots, atari sequences were evolved, and due to China split several times into several regional powers, and re-integration was short-lived, these differences were kept. Some variations spread to neighboring regions like Tibet, Korea, central Asia, until the next reunification in the Tang Dynasty in the 7th century AD, where the mainstream scoring rule - counting road method replaced others (like the primitive 溢 scoring requiring whole board filling), and ko rules started to mature and simplified to the form we can recognize today.


This is a very interesting question, I’ve previous suspected that it has some deep culture root way back in their precursor form, due to even Go itself wouldn’t be affected playing inside a grid, instead of on the grid. And there seems to be no reason why any of them have to be played on the grid.

Although, there might be an advantage for playing on the grid, like digging a hole on the intersection, to place pieces more firmly, it will be harder to line up inside a grid, and human eyes tend to like more regular shapes than zig-zag line up, and can trace an imaginary line even if we only saw extended tips at both ends (you can try it at home, draw a line, and cover the center with a paper, and your eyes would still try to see a line behind it). I personally often try to “fix” the uneven stones while playing on a real board, simply because it helps me recognize the shape and reading better somehow.

1 Like

You’ve really done your research. Fascinating :slight_smile:

I was thinking about the possible connection to astrology/astronomy/divination, and found this: Machinamenta: Excerpt 8: Divination and Games

Don’t know how much worth it is, but it is interesting to consider that if the goban reflects the universe/life/will of the gods, and the stones initially were thrown randomly on the board for divination purposes, then as the post notes, a ‘stone thrower’ (sticks in the case of Egypt) could be thought of as not playing a game, but throwing stones according to the will of the gods. It’s not too long a leap to think that nobles played it as a game, while certain commoners used it as a divination board, as the post mentions. Just food for thought, but the rules could be merely a reflection of how to interpret the distribution of stones. But, I assume then that this inspiration would perhaps be reflected in the chinese characters describing the game.

Sidenote, chinese is a very interesting language in how the symbols themselves mean so much beyond the mere word they describe.