The Pyramid and the Web

Posted by on Jan 30, 2008 in Board Games, Musing, RPGs, Story, Will's Thesis | 9 Comments

Part of the point of my still-evolving thesis (first mentioned in my early post, “On Playwriting, On Gameplay”) is to develop a new vocabulary for branching or unfolding narratives — something loose enough to survive spit-balling and rough use, but rigid enough to be clear and evocative in serious discussion. A lot of the ideas wrapped up in this vocabulary are not new. They should be familiar to anyone who’s studied dramatic structure a bit, understands the Three Act or Five Act structure (see Jeff’s series on the Three Act form), or has dabbled in narrative studies.

Once the page has finished loading, click any of the images in this post to activate the gallery. Navigate using your keyboard: N=Next, P=Previous, Esc=Close the gallery.

In subsequent entries, we’ll look at the unique ways that games — board games, card games, and especially roleplaying and story-making games — can use these terms to describe and make new sense of real-time storytelling play and dramatic manipulation at the design level.

Sound high-falutin’? It is a bit. Pretentious, too. But I take this stuff seriously and if you’ve played RPGs, you’ll see right away how these terms can be useful in the dirty minute-to-minute activities of running a game.

Toward New Definitions

I’m using the term “GM” in here instead of “DM” or “Storyteller” or “Narrator,” though these terms are traditionally synonymous. GM is simply the generic term for the player who designs and “runs” an RPG adventure. Don’t confuse it with the newer MMOG definition of GM, which is different. And thus confusing.

We’re defining two bits of jargon here, neither of which is especially groundbreaking but both of which facilitate further discussion of narrative (not just plot) and dramatic tension (not just Act structure) in games. These terms are good for analyzing games you play and for creating gameplay moments, whether you’re a video-game level designer, a board-game creator, or a GM.

First up is the gameplay web. After that, we’ll look at a dynamic modernization of Freytag’s pyramid, modified for use in branching fiction and interactive storytelling. This new schematic structure measures narrative and emotional tension as dramatic elevation.

Gameplay Web

Basic Gameplay Web
Elemental Gameplay Web

You’re already familiar with this, I’ll bet. A gameplay web is a streamlined flowchart, which is old hat for plotting in RPGs, fiction, and all sorts of storytelling. A gameplay web is a category “above” a flowchart, though, insofar as a flowchart is one possible way to interpret a gameplay web — another manifestation of a gameplay web is the archetypal dungeon of D&D and countless other RPGs.

(The next chapter in this series, “Dungeons & Diagrams,” looks at gameplay webs and dramatic elevation in classic dungeon-crawl play.)

The basic components of a gameplay web are broad and easy to mess around with. In the example gameplay web (Image 1 of 6), the three main parts are singled out: the starting point (a), paths or transitions (b), and decision points (c), which are the hubs leading through other transitions into other decision points.

The paths or transitions (b) are the linear elements between decision points, which is an intentionally big and vague definition; they can be passageways or exposition, as long as inches or miles, corridors or continents, seconds or decades. The space between one decision point and another might be filled by a quick cut skipping over miles and days of travel (“When you reach the castle,” says the GM, “you have a choice to make.”) or it might be a winding catacomb marked out in five-foot squares and traversed with miniatures on a map.

We have a starting point but no ending point because we don’t know, when the web is created, what the end of the game story will be — the players’ choices at each decision point could lead to an ending at any point on the web.

Predetermining or scripting an end to an RPG adventure is a beginner’s method, perpetuated by the necessary nature of pre-written and published game adventures from back in the day. A published adventure needs an ending because it’s read as much (or more often) than it is played through, and its built-in ending makes it easy for the GM to identify the climactic moment of the story, and thus the rise of dramatic tension leading to it. By separating the established play space from the unfolding story’s dramatic elevation (see below), we’re better able to react to player choices in the moment and create an atmosphere of rising action without being tethered to some preconceived notion of an established plot.

Let’s look at a beginner’s gameplay web (Image 2 of 6), reduced down to an elementally small scenario. Players at the beginning progress to a single decision point, resulting in one of two possible outcomes: (a) and (b). Though different obstacles and choices may be put before the players in scenes (a) and (b), these are false decision points because there’s only one possible outcome, regardless of what decision was made back at the first decision point — all roads lead to the GM’s predetermined ending. This creates the illusion of freedom (which is sometimes enough) but is only a little better than being an audience member. It’s like watching a movie and then, on the DVD, learning that there was an alternate action sequence in the middle. The suspense of an inevitable ending is one possible way to ratchet up dramatic tension, but it doesn’t take advantage of the unique options of story-games and interactive narrative.

In a future column, we’ll look at how gameplay webs function on the fly, as part of improvisational play (short answer: very well), but first, let’s put the length and width of a gameplay web together with the height of dramatic elevation so we can get some altitude.

Dramatic Elevation

Dramatic Elevation
Complex Dramatic Elevation (Side Angle)
Complex Dramatic Elevation from Above
Dramatic Elevation

We intuitively equate drama with altitude: Rising action. Heightened drama. “Ratcheting up the suspense.” Raising the stakes. The higher the story goes, the greater the peril. It’s much more thrilling to watch someone walk a tightrope that a sidewalk, isn’t it?

Freytag’s pyramid presents a five-Act model for understanding dramatic structure in which a story’s climax is diagrammed as something higher than the rising action leading up to it, or the falling action coming after. This notion of “rising toward climax” is common and intuitive now. Any reader of a movie review gets this stuff.

The basic parts of a dramatic elevation diagram (Image 3 of 6) are derived and modified from Freytag’s five parts. Instead of exposition, we begin immediately with the inciting action (1), which triggers the rising action that follows. Rising action is always followed by falling action (2), though the peak that marks the turn between up and down isn’t necessarily the climax (3). Smaller moments of falling action (2) might indicate a breather following a victory or time to lick one’s wounds following a defeat. The final stretch of falling action after the climax (4) includes the denouement.

Seldom, though, does a story stick strictly to Freytag’s symmetrical pyramid. Most stories’ dramatic structures are like mountains, not pyramids. Little sub-climactic peaks rise and perilous valleys fall between our heroes and their final victory or defeat. Indy gets the Ark, then he loses the Ark, then he gets the Ark again. Stories aren’t as tidy as Freytag’s pyramid.

In his defense, Freytag knew this — he was setting up an orderly means of appreciating old or ancient plays when he was drew up his pyramid. Whether you buy into the systematic appeal of the five-Act structure or not, whether you agree that Exposition is an Act unto itself or not (I don’t), the notion of elevation as a measure of drama is useful.

All we’re doing with dramatic elevation is exploding into three dimensions the idea of height as a measure of drama. Once you’ve broken it out from simple height, you can affix it to a gameplay web to simultaneously chart progress through a matrix of choices (the gameplay web) and the relative dramatic tension along the way, so we can see and appreciate the narrative emerging through play. (Images 4–6 of 6)

What’s next? Future columns look at gameplay webs and dramatic elevation in dungeon environments, how to use gameplay webs on the fly to improvise meaningful decision points and consequences, plus how to eyeball dramatic elevation during play to help minimize the influence of out-of-game tomfoolery on the game-story. Stay tuned, dear reader.

Don’t mistake appreciation with validation. The point here isn’t to grant some approval to play-stories that sync up just right with the model. The point is to get a better understanding of the stories emerging through your play sessions so that you can better control how drama rises and falls in later sessions. If you notice that your game adventures tend to lose a lot of dramatic altitude in the middle, you can take steps to raise the stakes and sustain suspense through to the end.

But what the hell does elevated drama actually mean? What is elevated drama?

The higher the dramatic elevation, the greater the peril, the more substantial the costs of failure, the more pressing the deadline, the farther the fall. Higher levels of drama come from battles against fiercer monsters, from attempts to disarm bigger bombs, from blistering gunfights atop taller skyscrapers.

This is a work in progress. Got an opinion or some paprika? Toss it in and we’ll stir it up. With just a bit of room to work with here, the best way to make sure that I’m tackling your questions or skepticism is to hear about it. The next column in this series is about webs and elevation in dungeons. What should come after that?

In the unpredictable and resource-driven environment of many games, however, the random factor means that climactic encounters might suddenly emerge. A few bad dice rolls and an early encounter with too many pathetic kobolds might put a character’s life in jeopardy — now the stakes of the battle don’t hinge on how much ammunition is used but on how many characters die. Some bad luck in a board game like Kevin Wilson’s ruthless frag-festival, Doom, can cause a dramatic climax (and a gruesome end) well before the gameplay web is fully explored.

It’s the job of the game designer — whether it’s the board game designer making decisions years before you play or the GM making decisions right then and there — to recognize the dramatic elevation of particular actions and moments in play and fuel those moments with the proper narrative energy.

It’s the difference between a loud whimper and a bang. Actual play, success or fail, might determine whether the players and their characters wail or cheer, but recognizing and controlling the sound volume should ensure that wailing or cheering is dramatically satisfying. Know when to dial it up and when to turn it down.


  1. John Arcadian
    January 30, 2008

    I like the interpolation of dramatic elevation on the Gameplay web. It takes the flowchart approach to scenario design and incorporates the rise, climax, and fall of action into the story as more than just a format for a generalized plot. I think a common tendency has been to work Freytag’s Pyramid into scenario design for the whole scenario, or as the pacing model for a night’s session. Understanding that climactic rise and fall occurs in each situation, or at each point in the flow of the game is hard to grasp, especially when we draw our inspiration and examples from linear forms of media.

    “Higher levels of drama come from battles against fiercer monsters, from attempts to disarm bigger bombs, from blistering gunfights atop taller skyscrapers.”

    I think this definitely holds true, but I think there are other ways that dramatic elevation can be achieved within individual scenarios. Thinking of NPC responses or in game actions in terms of “this elevates the action level” “this maintains the action level”, or “this lowers the action level” can help in controlling the flow of these situations. Entering combat definitely elevates the action level, continuing combat obviously maintains it. Performing special maneuvers elevates, while stepping out of combat to heal lowers the action level.

    The thing is, these raise, maintain, and lower the action level based on how the players and the Game Master play it out. Continuing combat in a blaise way can definitely lower the action level, since the action level is what is actually occurring at the table between the players, not on the map with the characters. Stepping out of combat to heal can raise the action level if the player “dodges between arrows while uncorking a bottle, gulping down the liquid inside and tossing the empty glass vial at one of the kobolds to distract him”. Even if this resolves as nothing more than the character moves and heals himself, it dramatically elevates the action level. It will all depend on the involvement of the players in the character’s actions.

  2. Will Hindmarch
    January 30, 2008

    Right on, John. One of the follow-up articles I’ve got talks about managing in-character and out-of-character influences based on how they affect dramatic elevation. Also, not all reversals or setbacks actually lower the dramatic elevation; maybe most don’t. Sometimes a failure just ratchets up the tension — failing to rescue the princess in Scene 7 means she’s another casualty if you fail to stop the evil wizard from melting the castle in Scene 8.

    Plus, what about the difference in resolution between dramatic elevation for a single play session versus dramatic elevation for a whole campaign? Or an episode versus a series?

    After that, I want to tackle a bit about gameplay webs for whole campaigns, and not just individual adventures. Then we’ll mess around with how different webs can interact.

    Hopefully there’s a lot of mileage to this jargon. My hope here is to inspire GMs to compare notes and lessons. I feel like too many RPG theories are built around “solving” some central dilemma and therefore shutting down the discussion, you know?

  3. Queex
    January 31, 2008

    If you’re going to have directed graphs, you should really include the arrows right on there- there’s a lot of literature about graph theory that’s interesting and pertinent.

    The characterisation of nodes as ‘decisions’ is a little flawed, I think; they could just as easily be a random element (the outcome of a fight or the result of a roll) or where two potential paths reunite. The path of nodes at the start of your example is really just one long edge, by the definition you give.

    That said, it’s an excellent way of thinking about stories and I look forward to the next part.

  4. Will Hindmarch
    January 31, 2008

    Thanks for the kind words, Queex. I’m not an educated grapher, so I’m going to make some mistakes on these for a while, I’m sure. (My time’s spent reading narratology and ludology right now, but I should really mix some graph theory in there.)

    The random factor inherent in many decisions is already a part of a subsequent article. The players may not be making an explicit decision to succeed or fail in battle against a monster, for example, but they will make decisions in play that affect the outcome. Also, generously, the dice do sometimes decide.

    I’ll continue to refine gameplay web examples as time goes by, but the example web I’m using now is intentionally flawed in some areas for use in future examples demonstrating acceptable linearity and pitfalls within the method. I hope. We’ll see how I do.

  5. Tommi
    January 31, 2008

    I think these are useful tools, but you can count on a random mathematician-in-training to ask for more details.

    1. The gameplay web is, or is at least to be by the pictures, finite. This is useful in that analysing finite graphs is pretty well-researched subject. How is the finiteness justified?
    In computer and board games, if ignoring psychological factors, the options are usually finite, even if numerous. (In computer games they are by definition finite, since computer can only store a finited amount of knowledge, I would intuitively think.)
    How is finiteness justified in case of roleplay, which usually includes very large amounts of human decision-making, and where all the important factors are not always codified in the rules (which may offer an uncountable amount of options)?

    2. Are all the relations between the events symmetric? Or, the opposite, does a pair of vertices (a, b) exist such that there is an edge from a to b but not from b to a?

    3. The pictures again imply that the angle (derivative) of dramatic elevation can only change at decision points and that it changes at every decision point. Are these correct, and if yes, on what basis?

    4. All the graphs are of the kind that can be drawn on paper without the edges crossing. I presume this is for convenience and clarity, as opposed to being an actual part of the models.

  6. Jeff Tidball
    February 1, 2008

    Three things:

    (1) The spatial model here — dramatic height mapped on a web of options — is fantastic. Fantastic. I think it will meet the test of being useful in practice.

    (2) This part makes me nervous: “The higher the dramatic elevation, the greater the peril, the more substantial the costs of failure, the more pressing the deadline, the farther the fall. Higher levels of drama come from battles against fiercer monsters, from attempts to disarm bigger bombs, from blistering gunfights atop taller skyscrapers.”

    It makes me nervous because there’s real peril for a hypothetical game-story creator in thinking that elevation comes from a quantity of hit dice. I think that it almost never does. And you can see that assumption go horribly wrong in your average RPGA adventure or summer blockbuster.

    (3) I recently picked up the widely lauded The Visual Display of Quantitative Information from the local library. Worth a skim, at least, but it was less useful for my purposes than I had hoped it’d be.

  7. Darrin Bright
    February 1, 2008

    Extremely interesting. The gameplay web in particular echoes a lot of my thoughts on scenario design.

    Some reservations, however:

    1) In theoretical terms, the diagrams are dead-on. In practical terms… are these descriptive of gameplay that has already happened, or prescriptive of what *might* happen when gameplay starts? While this looks like an amazing diagnostic tool, as others have pointed out, once you throw in human decision making and differing opinions on how to quantify “tension”, the gameplay might be too dynamic to graph until after gameplay is over (at which point the gameplay is no longer a web, it’s a narrative spine or storyline).

    2) Assuming the gameplay web can be at least partially prescriptive, who is responsible for creating the initial gameplay web? Is the GM entirely responsible for all possible decisions, nodes, and endpoints? If the players are involved, how much control should they get?

    3) I’m not sure I like tension being graphed vertically. Is this measurement from the GM or player standpoint, how do you define tension, and is this measurement made at the moment the game is being played or afterwards as part of post-game analysis?

    4) I think we’ve seen some designers comment already on the “fun thing” that drives a game engine, be it leveling up in D&D, collecting bricks in Catan, or landing on Boardwalk. In order for there to be any tension, however, there’s got to be “unfun” things, and I don’t think we’ve seen much discussion about that. The players obviously don’t like unfun things, but the game would be boring and lack any tension without them.

    Call of Cthulhu is one of the most successful games at creating tension (gobloads of unfun things), but I often find myself wondering and marveling at how it does so. Robin Law’s recent blog on expectation frames may have hit the nail on the head, though. Does the expectation frame define most of the unfun things and essentially outline the basic structure for failure outcomes?

  8. Jeff Tidball
    February 1, 2008

    Here’s something that struck me last night as I was trying to fall asleep: I want a game mechanics to change at different heights of tension. I don’t want to roll the same d20 to hit, with all the same bonuses and penalties, whether I’m fighting Asshat the Lackey or Grimthor the Campaign Boss.

    Dogs in the Vineyard takes a giant step in the right direction, but I’m not aware of a game whose mechanics respond to dramatic height in a particularly robust way. I mean, any game with some kind of hero point mechanic does, to some extent. But surely there must be a way to improve that.

  9. // story, games, together
    March 25, 2008

    […] take a closer look at how gameplay webs and dramatic elevation work in practice—and how these things relate to that classic icon of RPG adventure architecture: […]


Leave a Reply