Edge In is a game inspired by Kennexions which uses metaphors as obscure hints regarding what other players are thinking. The hints are some obscure, in fact, that you generally don't find out what other players meant until the game is over; and in the mean time, every metaphor forms a constraint on your own play.

Some notes on terminology. The word "edge" is the graph theory term for a connection (that is, a line) in a network; the word "graph" is the graph theory term for a network; and the word "node" refers to the things connected by the edges.


The game is played with two or more people, with the help of paper, blackboard or whiteboard (though, see spoken variant below). At any given time, there will be a number of empty circles (nodes) with some connections. All players strive to have a "story" in mind at all times; a "story" is an idea of what each node refers to, such that all the connections make sense. Nodes can be people, objects, processes, or properties; one word, or a complex idea; anything works.

Before beginning, players should agree on what sort of story is considered valid. The main choice is fact vs. fiction, but players can agree to come up with a certain sort of fiction, such as a fictional culture; or a certain sort of fact, such as biology, TV shows, mathematics, or poetry.

The first player draws two circles and a connecting line. On the line, a metaphor is written, in the format "water : air". This signifies that the player has in mind a story such that the first node is to the second node as water is to air.

Play proceeds in turns, with each player adding a connection. This connection can be between two existing nodes, or between an existing node and a new node. Players are also allowed to connect nodes which are already connected, drawing a 'double edge' (or just writing the new metaphor next to the old edge).

It's worth noting a few things about how metaphors can work in the game. In one game, I played the metaphor "donut : coffee". I had in mind stories where one item gets immersed (dunked) in another, or where one thing is good when paired with another. Initially, the story was "engineer : hard problem :: donut : coffee". However, another player's winning story assigned the value "cop" to the donut side, since cops stereotypically eat donuts; and "stressed office worker" to the coffee side, since they stereotypically drink too much coffee. We can write this out as "cop : donut :: stressed office worker : coffee". This is considered valid play.

Another subtlety that comes up has to do with the way nodes are constrained. Suppose that a node is the "crops" side of "locusts : crops", while simultaneously being the "brick" side of "brick : tower", and the "tower" and "locusts" nodes are connected by "general : army". We've got a situation where a (metaphorical) general is (metaphorically) commanding the destruction of the (metaphorical) crops he's made of! What I want to point out here is that it's okay for a story not to have the exact same thing be destroyed by the locusts as is used for the tower. One valid story could have "computers" as the crop/brick, "experience as a programmer" as the tower/general, and "computer viruses" as the army/locusts. The computers getting destroyed by viruses are not the same ones which gave the programmers their experience, and this is fine. (The general:army link is a bit weak here though; apologies for not remembering the real example.)

In good play, players each have a story in mind and on their turn they fill in a metaphor which fits with their story. That metaphor will often cause other players to need to change their stories. In very competitive play, players will attempt to have two or more stories in mind, and play metaphors which are almost sure to drastically alter other players' stories. A final option is bluffing. It's allowable for a player to make a move without having any story in mind. However, a bluff may be called, ending the game. When a player "calls", the previous player must produce a full story, explaining how each metaphor fits. If they can't, the player who called the bluff wins. All players with complete stories produce them, and thereby win as well. The remaining players lose. It is usually interesting to hear incomplete stories from these players anyway. Sometimes they can be fixed, and in any case they help explain the moves the player made.

As players give their stories at the end of the game, they can write them out in the empty circles that represent the nodes. With more than two players, this gets messy, so it's not required.

Players may also freely join the game (if they can think of a story that fits) or bow out.

The game ends up very contemplative. Players must be willing to wait ten minutes or more as their opponents make sense of a move and adjust their story. On the other hand, that could be all consequence of my current skill level; people seem to be getting better fairly quickly.

Play Advice

Coming up with Metaphors

One thing I learned after a couple games was that I was very limited in the ways I was coming up with metaphors. I would look at a node, recall what I imagined it to represent, then ask myself for a good metaphor for that concept. For example, looking at "nose : face" I might have in mind the story "nose : face :: sun : sky". Then I ask myself what the sun is like, and come up with things like "day" or "flower". Then I might add all metaphor such as "day : night" or "flower : bee". Those sorts of moves are fine, but can be a bit too on the nose (if you'll forgive the pun). A more lateral strategy is to choose a less likely metaphor. You might ask yourself "what if the sun were like a stapler?" And staplers work with paper, so "stapler : paper" could be your move. What is the paper of the sun? I would say hydrogen atoms. The sun bundles them up via fusion into heavier elements.

I'm sure there's much more advice about coming up with metaphors that I haven't figured out yet, and just as much about coming up with satisfying stories.

Coming up with Stories

At least one player I know thought that it was easier to think about how the metaphors in play might apply to people than to physical objects. For example, they might decide a node stands for "politician" and see where they get from there.

It's (arguably) a good strategy to try to come up with stories which are very different from those other players have in mind. One way to do this is to take a metaphor which might have an obvious value association, like "toxin : nutrient", and try to put something good on the bad side or something bad on the good side. For example, we might put "toothpaste" on the "toxin" side and "junk food" on the "nutrient" side, since toothpaste kills plaque and junk food nurtures it.

Similarly, metaphors like "fox : rabbit" will tend to cause two nodes to be "against" one another, and metaphors like "bird : flock" will tend to cause two nodes to be "on the same team". If you can come up with a story which makes the "fox" cooperate with the "rabbit" or the "bird" fight the "flock" (without stretching the metaphor), you'll usually be able to make life very difficult for the other players.

In play with other people, I don't usually use pen and paper to work out a story, but here's some advice that worked for me when analysing a game on my own. Draw the nodes as large circles with room for notes, leaving room for the connections but not their labels. Then, looking at one connection at a time, think about properties you want the node to have; for example, confronted with "fox : rabbit" I might write "something that hunts" (or just "hunts" for short) in the circle on the "fox" side. Having a list of properties doesn't really guarantee the connections work, but it aids thinking.


Why does Edge In work out the way it does?

One thing that surprises new players is that there can be other valid stories than the one they have in mind. At times, there is a strong illusion that everyone must be thinking the same thing. But the reality is that players rarely think along the same lines as one another. The game tends to actually end; players at the end tend to be surprised by one another's solutions. And the surprise isn't usually because of one player using an obscure idea other players weren't aware of; it's because one player thought of something another player didn't think of, but could have thought of. Why is this the case? The human mind is usually amazingly good at metaphor and analogy, at jumping to associations.

I think this says something about how broad and complex the space of human ideas really is. Players don't converge to a single story because there is not one big metaphor that lies behind everything.

A caveat: I've only seen six games of Edge In at the moment. The most moves played in one game has been ten. I've already seen one game where two players ended up with the same story. So actually I could be completely wrong about what tends to happen with longer play or more experience.


I've seen people think the game is impossible, stare at the board for four moves or so, then suddenly come up with a valid story and join in. People seem to be able to acquire the skill. Moreover, we seem to be getting better at the game via multiple plays. I think there's something being practiced by Edge In that goes beyond just analogical thinking. Something having to do with working under many constraints, and being able to start from scratch and look for a new solution when the situation changes.

In my view, human consciousness (more specifically WEIRD consciousness - Western, Educated, Industrialized, uRban, Democratic consciousness) has been shaped over the centuries by a series of technologies which change the way we think about ourselves and our world. These include maps, writing, mirrors, clocks, indoor lighting, and more recently, photographs, movies, eyeglasses, computers, cars; to name a few. These technologies pervade our lives enough to change the way our brains work; for example, clocks allow and encourage us to think of time as objective and shared; mirrors give us a concrete concept of what we look like to others, and force us to (at least slightly) think of ourselves as objects in the world. Technologies which change our consciousness also change our brain structure; for example, by learning to read, a person gives up a portion of their brain near the visual cortex which would otherwise be used for recognizing faces. Are illiterate people experiencing faces in a fundamentally different way than those of us who can read? (The difference is that illiterate people process faces with both hemispheres, rather than just one.)

Games are a consciousness-shaping technology too. Learning to strategize in games makes us think about goals and actions in a different way. Even someone who doesn't play games absorbs these ways of thinking through culture; words like "strategy", "rules", and "win", and expressions like "your move", "the ball is in your court", "a strong position", etc. give non-players an idea of how game players think. Of course, games have been with humans a long time, so this particular sort of consciousness took hold millennia ago. But the question is, does a game like Edge In contribute some new form of consciousness? How are we reshaping our brains by playing?

A tentative guess: the game is an extreme version of a normal conversation. In conversation, we usually have just one guess as to what the other person is saying, and we will come up with a new guess the moment they say something which doesn't fit. By being extremely ambiguous and constantly proving players wrong, Edge In encourages seeing properties of the space of possibilities, rather than just holding one guess.

Dual of Hipbone

Imagine playing Edge In with blank lines instead of blank circles. Players would spend their turn adding an item to the board and/or a line. Since having some sort of connection between things is easy, the only way this would be interesting to play is if players have fairly high standards for the metaphors. Since each metaphor is more or less disconnected from the others, the players might as well reveal right away whether or not they can come up with a connection.

The resulting game is more or less like HipBone. The main difference is that either player can score points on any given connection. Also, if we stay strictly to the rules of Edge In, it seems that the first time one player can score a point which the other cannot, they win. The game depends on players’ honesty in an awkward way: one player draws a line, then both players attempt to think of metaphors which fit on the line, then they both reveal what they’ve got. Do we require them to write the metaphor down privately, so that the second person to speak can’t steal the first person’s idea?

Now imagine playing HipBone but with metaphors on lines rather than phrases in circles. Players put down a metaphor on their turn, then score points by presenting a story (‘story’ in the Edge In sense; a list of items which go in the circles). A point is scored for each line which makes sense in the story, and the sense of "makes sense" used here is the one from HipBone; ie, both players must OK each point, and there is negotiation allowed ("I'll give you this one, but you owe me").

If we take IntergameProtocol seriously, the analogies and disanalogies between the two games seemingly say something about what’s elegant and inelegant. What more general game would more truly encompass both Hipbone and Edge In?


Alternate rules, listed roughly from most to least concrete.

Play without paper

Edge In plays surprisingly well without writing things down. The metaphors end up more poetic and verbose, since they're spoken instead of written. I recommend aiming for fictional stories, and speaking to the other players as if you are discussing some secret by using metaphors and code words. Nodes can be referred to by their existing metaphors; for example someone might say "my sources tell me that the eagle is a doorframe, and the music box is the old paint which flakes off the frame" when connecting existing nodes, or "the eagle is a doorframe, and it has flaking paint" if adding a new node. Players can review everything and make sure they're on the same page by listing all the titles attached to each node.

It's also possible to play by assigning letters or numbers to each node, but this leads to less verbal review, making it more likely people will flip some edges.

Starting with two edges

The GBG as described by Hess often starts with two seemingly contradictory concepts and attempts to bring them together. A good way of implementing this in Edge In is to begin with four nodes, A, B, C and D; and to have one player connect A and B, and another connect C and D, using very different metaphors from one another. Play proceeds normally from that point.

No Metaphors

The game also works fine if straightforward language is used instead of metaphors. For example, a line might be labelled "is part of" or "destroys". Use arrows to indicate which direction the relationship goes; the node at the base of the arrow comes before the text, and the node at the tip comes after.

Actually, I haven't directly tried the plain language version; instead, I played a more limited version where the edges from ConceptNet? were the only allowed moves. Choosing a pool of well-defined connections like this sounds very limiting, but in my experience it definitely works.

The advantage of not using metaphors is that there's no chance of players forcing things to fit. Consider the metaphor "horse : road". A horse is taller than a road, smarter than a road, faster than a road, more organic than a road, and less expensive than a road; pick any two English words, and they can be argued into a kenning with horse and road. Edge In becomes trivial if players are willing to make such arguments. If this troubles you, I recommend the no-metaphor version; it's about equally rich and engaging.

Fixed Board

This variant hasn't been tried, but seems straightforward enough.

After seven turns or so, a game of Edge In tends to become highly interconnected. Play slows down as it starts to take longer and longer to check everything. This is mostly fine; the game doesn't slow down in a boring way. But if you want a more uniform experience, it might make sense to play on a fixed board; that is, a set of lines and circles arranged ahead of time (or right at the beginning of the game). Each circle in such a board should have just two or three lines connected to it.

Crossword-Puzzle Style

Completed games can be presented to people without any of the accompanying stories, as sort of a crossword-puzzle-like challenge. Solving such a challenge feels more difficult than playing an actual game, since you have to take metaphors in all at once rather than as they’re played.

Another option is to use a computer to generate metaphor networks, and then try to solve them. Depending on how you do it, this can have the problem of being too easy, or of coming up with terrible metaphors, or of creating a metaphor network which is unsolvable.

I’m not sure what approach I’d take if I were trying to design a daily or weekly Edge In riddle (like for a newspaper). It seems like that would involve totally different skills than normal play. Certainly I’d try for interesting symmetry, though.

Restricted Stories

As part of the normal rules, players may agree to restrict valid stories to some specific domain. But perhaps it's worth saying a few words about more speculative restrictions.

Players could agree that all stories must contain a tree. Or, more realistically: if working with purely fictional stories, players could agree to have at least one node represent a murder (in order to get a murder mystery vibe).

Players might aim for stories consisting only of verbs, no nouns.

Players might decide to require two working stories in order to win, rather than just one.

Shared Story

In this version of the game, players write out a value in each node they create, along with the metaphors placed between nodes. The players still have private stories, but this way there is a public story throughout the game too. Play proceeds until all players feel like ending the game, and the goal at the end is to have a story that is as different from the public story as possible. There is no formal scoring; instead, the player with a story most subjectively different from the public one wins.

The intuition behind this game is that we all are chained to a public, shared world, but would like to operate in private worlds which are different from the public one, while continuing to be able to function in the public world.

Keep Playing after a Challenge

When a player "calls", the previous player reveals a story which satisfies the constraints. However, if the challenger can add a new link which breaks that story, play may continue. (After all, there aren't any valid stories out in the open after that point.) Of course, if the challenged player can't produce a story, or the challenger can't ruin it with a link, the game ends as usual.

Public Stories

Each player reveals their current story after each move. Each move must break the previous player's story. Style points are earned for making minimal changes to your story over time; for example changing a node from "sky" to "airplane" instead of coming up with something completely new.

In order to keep the game growing, rather than just becoming a denser and denser nest of lines, this variant has an extra rule. Players are allowed to make two moves on their turn if one move adds a new node and the other does not.

Less Mystery

This variant makes the most sense when working with fiction.

New nodes get text labels just like in the "shared story" variant, but in this variant the labels limit everyone. For example, a player might add "tea" as a new node, connected to a node "hobby" via the metaphor "eye : storm". Different players would interpret this differently in their private stories; for example, the tea might be kombucha, and the hobby is microbrewing; and the meaning of the metaphor is that kombucha tea is the only fermentation the person has learned to do well. Another player might decide the hobby is something like swordfighting, but the person takes peaceful breaks to drink tea.

This variant could be combined with Restricted Stories, above, to give a full Whodunit type game: have the nodes be people and clues, and require that in any valid story, one of the people listed actually did the deed.


An appealing generalization of Edge In would be to allow metaphors between metaphors. Unfortunately, this was hopelessly vague in practice. We tried allowing players to draw blank lines, then draw a metaphor-bearing line between some non-blank line and the blank one. The idea was that a working story should fill in the blank line, as well as filling in all the empty circles; and of course the blank line should obey the metaphor pointing to it.

Unfortunately this seemed to turn out hopelessly vague. All three players had a different interpretation of it. So, a slightly more specific operation is described here as a rules variant.

Instead of the normal move, players may draw a blank line; they can add a blank circle which it connects to as well. Then, the player circles one side of a metaphor from elsewhere on the board. For example, if a metaphor "ineffective : futile" is in play, the player might circle "futile". Then, the player makes a small circle on their blank line, with a colon to one side of it; that is, the line basically gets labelled " O : ". Then, the player draws a line from the newly circled term (for example, "futile") to the small circle. Then, the player labels this new line with a new metaphor; for example, they might write "inedible : vile".

What this move accomplishes is to constrain what the "small circle" might mean. "inedible : vile :: futile : O". (This might be completed by something like "pointless" or "dangerous".) Any valid story must assign a metaphor to the edge which contains the small circle. A future play could put a second small circle on the other side of the colon on the same edge.

Meta-meta-edges are possible, too, and so it's fine to draw several blank lines on one turn so long as they're all being constrained by one top-level metaphor.

Toward a GBG

The "start with two edges" variant above makes Edge In slightly closer to the Glass Bead Game. What else can be done in that direction?

Library of Metaphors

The GBG is played in reference to an Archive of patterns, which can grow over time. If Edge In had a standardized Archive, it would be like a rulebook stating exactly what a metaphor means. For example, players could consult the rulebook to discover whether "fox : rabbit" must match with things which are actually antagonistic toward one another, or instead could match any hunter and hunted. (Dart:dartboard::fox:rabbit for example.)


Edge In is "contemplative" in the sense that you learn to think hard about getting a working story, and you deliberately hold the story in your mind once you've got it. (I also find that I "smooth out" the story over time, finding tweaks which make it a stronger match for the metaphors in play.) However, this sort of contemplation is very active, not exactly meditative or mindful.

An option after the game is over is to contemplate the results Kennexion-style, expanding the links produced by the winners to get compound kennings. I've found that when I actually do this, I run into weak links in the metaphors that keep the kennings from working well. Such a problem wouldn't exist if the game were played by drawing from a curated library of metaphors.

A more speculative option for contemplation after play is to flip every single metaphor, and see if you can come up with a working story for the resulting graph. Of course, it’s possible that this will just be unsolvable; but the idea is that flipping all the edges should work, since it means “taking the dual” of the story. (In category theory, the dual is defined by flipping all connections; however, category theory has arguably more specific connections, so maybe it works better there than here.) If we take the idea of a “dual of a story” seriously, then each valid story we produce on the original ought to have its own dual.

It’s also worth noting that some mathematical objects are self-dual. A self-dual story would be one which still works when every metaphor is simultaneously flipped.

Scholarly Requirements

The GBG requires years of scholarly training to play, since individual moves require moderate- to expert-level understanding of a wide variety of intellectual fields. I can imagine Edge In being played that way; a few days ago I heard a discussion which could be summarized as "topology : pointless topology :: pointless topology : ??". This is like the game-state "A : B :: topology : pointless topology, and B : C :: topology : pointless topology". (The idea, or part of the idea, was to be able to have a topology with three open sets, any two of which could be combined to obtain the entire topological space, but all three of which could not.)

Maybe the right concept here is "Edge In, but restricted to metaphors which were important cultural or scholarly realizations at some point". This would include edges like "Turing machine : von Neumann architecture", "melody : chord sequence", "chord sequence : comma pump", "Euclidean geometry : Riemann manifold", "speech : writing", etc.

Cooperative or Solitary Play

Edge In is a competitive game, whereas the Glass Bead Game is either played by one person, or by several people cooperatively adding to an unfolding pattern.

Perhaps there isn't much difference, though, except in degree of skill.

Imagine a community of Edge In players who got so good at thinking of stories which work for a given set of metaphors that they could no longer beat one another. They would look at a game and know exactly what sorts of stories could exist; but that doesn't mean they would actually list the stories. There's a skill that's appearing already in players of Edge In, of being able to know what sorts of stories work for which parts of a web, and knowing that we'll be capable of making up a story there if challenged. (In other words there will sometimes be one or two nodes we don't even bother thinking about during play, because it's so easy to create the story there. But of course further play can make easy nodes hard again.) Players so skilled that they can't beat one another would also know if some particular move created an unsolvable board. The rules governing valid patterns — patterns which admit at least one solution — would be the "true" rules of Edge In, and would be much more similar to the Glass Bead Game as conceived by Hermann Hesse.

Amongst such a community, or with such knowledge, it would make sense to play until only one valid story remained, or until any additional specificity would be mundane, or only reduce the symmetry of the set of valid stories.

Of course, the question arises of whether this is really possible. TheNovel is set in a world where there is no more intellectual progress; but it's possible that human ingenuity has no such bound. Any new ideas in a community of expert players would enable individuals to beat one another by knowing something that makes a new story available to them (even if the set of metaphors is standardized).