Emergent Meaning and Narrative in the Digital Space
Addressing Tensions in Games and Game-like Media
“Most people want to be told a story. Leaving it up to a random number generator is dicey.” – Ed Del Castillo, Producer, Command & Conquer
Storytelling is one of the oldest forms of entertainment in human culture. Throughout the centuries we as a species have invented many ways to tell ourselves stories, and continue to do so. In recent years, games have become one of the most popular formats for delivering narrative, but as a new form of storytelling, narrative games face several critical issues. Although the problems facing this burgeoning medium are not insurmountable, they are uniquely twisted by the fact that games are, by their nature, a participatory endeavor – a transaction between the designers and players where the contact is far more direct than it is in other forms of popular storytelling. Although as an industry games are growing and are doing much better than older, more established mediums in the marketplace, critical thinking about games, and especially how narratives in games are constructed, is still in its infancy. There are at least two fundamental unresolved questions in games criticism –, “how do games mean?” and “how do games tell compelling stories?” Answering both of these questions is a much larger task than I am capable of achieving in this essay, but by the end I hope to provide a potential framework for how games mean and from there briefly propose a solution to the narrative question.
In many ways, the medium closest to video games is film; the games industry has been moving toward making games more ‘cinematic’ for many years at this point. Similarities between the two mediums abound: both are primarily visual, the productions are often large-scale, requiring the work of hundreds or thousands of individuals, both games and movies are entered in festivals, and both are created by small independent artists as well as large studios. The road splits when it comes to the way that games and films intersect with those consumers for whom most are ostensibly produced. While film-watching is a primarily passive experience where the viewer sits and watches the material presented on the screen without comment, games seek to engage their players on a fundamental level where the player has control over the game’s environment. As David Bordwell writes,
…the classical Hollywood film presents psychologically defined individuals who struggle to solve a clear-cut problem or to attain specific goals. In the course of this struggle, the characters enter into conflict with others or with external circumstances. The story ends with a decisive victory or defeat, a resolution of the problem and a clear achievement or nonachievement of the goals. The principal causal agency is thus the character, a distinctive individual endowed with an evident, consistent batch of traits, qualities, and behaviors.
This is not so in games: what occurs in a game is the result of a player’s actions as much (or sometimes more than) the work of game’s programmers, directors, artists, or anybody else involved in the project. The character, or “avatar,” often serves only as a vehicle for the player, not as an entity independent from the player’s influence. This influence can be as narrow as propelling the protagonist of the story forward from plot beat to plot beat in a generally on-rails experience or as wide as choosing which characters live or die, or what narrative threads are pursued from a massive tapestry of possible choices. This is quite a scaling-up of choices from what audiences encountered in 1934 when they viewed Ayn Rand’s Night of January 16th, a play with a game-like twist at the end which allowed a jury selected from audience members to decide whether a woman on trial was guilty or innocent, with two different ending sequences depending on how the jury voted. What was once seen as a gimmick is now the primary distinguishing characteristic of a major form of media. The open-ended nature of many game narratives allows for a new sort of narrative, emergent narrative, to manifest.
Regardless of whether their medium is physical or digital, games may be divided into two varieties. The first is emergence games, contain “the primordial game structure, where a game is specified as a small number of rules that combine and yield large numbers of game variations, which the players then design strategies for dealing with.” Chess, poker, backgammon – all of these games fit into this category. The second, progression games, are games in which “the player has to perform a predefined set of actions in order to complete the game.” This category is more recent, and many video or otherwise electronic games can be categorized as such. These categories are not particularly ground-breaking ways to think about game structure – it is only when narrative considerations are introduced that questions begin to arise. Due to their strong linearity and aesthetic ties to older mediums, progression video games are the easiest method for delivering narrative – there is a straightforward series of events that the player must carry out in a specified order as determined during the game’s creation and development process. These games are very developer-oriented (or author-oriented one might claim). The player exists within a predefined narrative in a predefined world with predefined characters, and the player’s ability to interact and influence these elements is strictly limited by the narrative that the developer wishes to present. Traditional methods of analysis can be easily tweaked and function fairly well when it comes to progression games. This is not the case, however, with emergence games.
Emergence games need not have a narrative at all; there is no narrative inherent in a game of go fish, save perhaps the narrative that is created and later related when a player describes his or her play experience – for example, “Sarah and I were playing Go Fish, and she guessed my cards correctly six times in a row! I suspect that she may have been cheating.” These sorts of narratives are interesting, but they are not inherent to the game itself, arising only as a result of the play experience. Looking specifically at video games, there are many emergence games that include pre-authored narratives. Emergence games differ from progression games in that in progression games, the player may choose to interact with the presented narrative to a large degree, only tangentially, or not at all. The depth of presented narrative can range wildly, from rudimentary storytelling in games like Minecraft or Spelunky to deeply involved plot lines, in titles like The Elder Scrolls V: Skyrim or Left 4 Dead. These four games will serve, following a brief overview of the development of procedural generation in games, as my primary examples through the rest of this essay.
Any history of procedural generation and emergence in modern video games must begin with the 1980 game Rogue. In this game, the player navigates a dungeon that is procedurally generated as the player progresses, thus ensuring that no two plays of the game are the same. Rogue’s influence cannot be overstated; whereas many games are considered formative in their respective genres, only one modern genre is named specifically for its progenitor: the “roguelike” genre. While there are some issues inherent in the use of the word roguelike, and the exact nature of the genre is currently still being debated, in this essay I will be using the word to describe a game that meets the following criteria:
- Permanent player death: The player starts each new game session with a single life, and death results in all progress being lost.
- Procedural generation: The area through which the player navigates is a randomly generated space that is not constructed by the developer (“bespoke”) ahead of time. The effects of items of the same name may vary across plays, or in other words, a red potion may heal the player in one play and damage the player in another.
- Step-based gameplay: Movement takes place on a grid in the four cardinal directions. While you are not moving, nothing else is moving or acting. Each movement or action taken by the player allows all enemies in the level to take a movement or action.
This is by no means an authoritative definition, but it is illustrative enough to serve my purposes.
Roguelikes have traditionally not been a popular genre. The unfriendly nature of the gameplay put off widespread adoption of roguelike games by most players, who found games like Moria (1983), Nethack (1987), and Dungeon Crawl (1997) too punishing to be enjoyable. The genre found some success with Japanese gamers. The popular Mystery Dungeon series spawned a franchise that continues today with spinoffs featuring characters from popular series such as Dragon Quest (1993), Pokémon (2005) and Final Fantasy (2008), but games in this series have always been considered niche titles – they sell, but not often in blockbuster numbers, and when success comes it is attributable not to the genre but from the licensing of popular characters. It was not until 2011, when independent developer Edmund McMillen released The Binding of Isaac that the roguelike genre exploded into the gaming world’s collective mindspace, selling over two million copies. While not a true roguelike under my above terms (it did not feature step-based gameplay), Isaac’s unexpected popularity and massive sales led to a heightened awareness of the power of procedural generation and emergent gameplay systems in both the game-playing public and the community of game designers. While games in other genres had used emergent systems in the past to enhance their gameplay, Isaac served as a turning point. Never before had there been such a strong emphasis on the emergent nature of the gameplay; that which had once been utilized either sparingly or under layers of narrative intended to shield the game systems from players was now placed front and center. Emergence had become king.
The success story of Minecraft, a game by Sweedish company Mojang, is the story of the triumph of emergence in perhaps its most distilled essence. More so than nearly any other game, Minecraft allows its players the complete freedom to build, quite literally, their own story. Presenting players with an infinitely large world map comprised of randomly generated lakes, mountains, deserts, oceans and other biomes, players collect hundreds of types of blocks with which they can build, block by block, cities, robots, computers, or any other number of things that their minds can imagine. It is not dissimilar from a never-ending bucket of LEGO bricks – if you can visualize it, there is a good chance that it can be built. The game cycles between night and day – while it’s light out, players are free to build, but at night enemies will appear to attack the player and his or her construction projects. Players are encouraged to build bigger, discover new realms, accumulate weapons and armor and eventually slay a mighty dragon. None of these tasks, however, is required for play. A player may even choose to turn off enemies, and play in a purely “Creative” mode, eschewing the “Adventure” mode entirely. The preexisting narrative in Minecraft is almost non-existent; there is no text in the game outside of menus, save for a short poem that displays before the credits after the player kills the dragon in the climax of the game’s Adventure mode. There are non-player characters that live in villages, but they do not speak. The world, being entirely procedurally generated each time a new game is begun, has no preexisting history. The story is left entirely to the players. This lack of direct narrative has caused many to attempt to fill in their own stories. Notably, Jason Rohrer’s Chain World imposes additional constraints on the player in order to create a narrative: the player is allowed only a single life, and when the player dies, he or she must save the game and pass the save file on to another person. The new player inherits the world in the state that the prior player left it, and through a cycle of players passing the game from one to another, the world itself obtains a history, with prior citizens, “ancient” ruins and hidden treasures that players have left behind.
Dereck Yu’s Spelunky is a game with slightly more narrative than Minecraft, but still one deeply rooted in the emergent tradition. In an homage to Indiana Jones, the player steps into the shoes of the Spelunker, a bullwhip-wielding fedora-clad archeologist and explorer who has set out to plunder an ancient tomb for artifacts and search for the legendary golden city. The Spelunker will loot idols from boulder-trapped altars, confront the spirits of the dead and come face-to-face with the god Anubis before his quest is complete. The game is divided into five distinct worlds comprised of four levels apiece. The worlds and theming are predetermined by the game ahead of time, but the individual levels are procedurally generated. The computer builds the cave, checks to make sure that an unobstructed path to the exit exists so that the level can be completed, and then populates the level with enemies, treasures and traps. The levels in Spelunky are designed such that every object has the potential for interaction with every other object, allowing for unexpected sequences of events to occur. For example, the Spelunker might trigger an arrow trap, which fires an arrow. The Spelunker jumps over the arrow, so the arrow flies into a shop, damaging the shopkeeper. The shopkeeper then becomes enraged and attacks the Spelunker, who is hit by a bullet and thrown backwards into a spider, which he bounces off of before landing in a spike pit and perishing. If the Spelunker had simply taken the arrow and the damage that the arrow would have done to him, he would not have triggered this emergent chain of events that led to his untimely demise. Indeed, telling stories of how one perished in Spelunky is a popular pastime within the community of Spelunky players:
Christ. Killed by a freaking rolling stone trap because I fell off a ladder without noticing. Of course, I only released the stone in the hopes that it would kill all the giant spiders, so they probably wouldn’t[sic] gotten me in the end. – Jick
I had a silly moment in today’s daily challenge. After getting cursed with the ball & chain I decided to experiment and broke up the Moai head using it like I saw in the eggplant run video (without any particular reason to do so, I didn’t have an eggplant or anything). It worked out pretty well and I went ahead and suicided using a jetpack to ensure that I lost the ball in chain.
I respawned in the Moai head, but I had destroyed the floor so I couldn’t use the exit. I panicked for a few minutes (and wasted a few ropes and bombs on something I pretty much knew would not work) and then remembered that both exits on that level went to the same place. Duh. – rchicken
My least predictable death probably occured in the temple and I think I might have even seen it on here in one form or another: I enter a level, nearby a crocman triggers an arrow trap, the crocman teleports away from the arrow and telefrags me. All that happened in less than 1 second so there was no way for me to avoid it or anything. – JustSmall
Although Spelunky is not the only game with a thriving community that exists outside of the game as a social group (massively multiplayer online games are perhaps the best example of these environments), the existence of a prominent fan community presents an excellent example of what Bruno Latour was getting at when he described his theory of actors, agents and networks. What Latour illustrated with the use of a gun exists within Spelunky. To paraphrase Latour, when the player originally sits down to Spelunky, he only wished to play, but now with the game experience at hand, he wishes to share his play with others. The network cannot be escaped; the game itself makes sure to remind the player of how the current play session measures up to that of the player’s friends via leaderboards and rankings. The presented narrative of Spelunky is ultimately irrelevant. The paper-thin nature of the story and complete reliance upon external preexisting cultural artifacts for context is excusable, however, as that is not where the truly important story of Spelunky lies. The meaningful narrative of Spelunky is constantly being formed and re-formed over and over again by the fan community that has grown around it. The stories that are told about Spelunky become the story of Spelunky, and the game has the capacity to produce an inexhaustible amount of story so long as players continue to engage with the game and record their play experiences.
Unlike Minecraft or Spelunky, Bethesda Softworks’ The Elder Scrolls V: Skyrim has a heavily authored narrative experience waiting for players who decide to step into the fantasy world of Tamriel. Skyrim’s story about the return of dragons to the country of Skyrim after centuries of absence comprises some twenty hours of gameplay, and there are additional stories involving wizards, thieves, assassins, demons and more, each just as long as the central plot thread which is considered the “main” narrative experience. However, despite hundreds of hours of bespoke, fully authored content, Skyrim stands as one of the most successful emergence game experiences to date. Perhaps more than any other game on the market, Skyrim embraces the sort of storytelling methods that Salman Rushdie observed and spoke about after watching his son play the western-genre cowboy game, Red Dead Redemption:
One of the things that is interesting about it to me is the much looser structure of the game and the much greater agency that the player has to choose how he will explore and inhabit the world that is provided for you. He doesn’t… in fact, doesn’t really have to follow the main narrative line of the game at all for long periods of time. There is [sic] all kinds of excursions and digressions that you can choose to go on and find many stories to participate in instead of the big story, the macro story.
Skyrim takes to heart Sid Meier’s definition of game: “A game is a series of interesting choices.” The entirety of Skyrim’s play experience is choice – choosing one’s avatar race, gender, appearance and skills; choosing how to interact with the non-player characters and creatures in the world; choosing whether to pursue or ignore narrative hooks and threads; choosing which problems the world presents are worth interacting with. More than either game previously discussed, player choice functions as the driving element of every narrative aspect. While the land of Tamriel and the impending draconic invasion serve as the fabula of Skyrim, it is the player’s choices and the events that unfold from them that serve as the game’s syuzhet. Although both are required for a complete story, the question arises as to whether they equally important. Is the story of Tamriel created by the game’s designer before play begins, during the programming stage? Or is it instead that there are many individual Tamriels, authored either solely by the player as play progresses and from the process of the player’s choices? Could it be that Tamriel is authored jointly, as a melding of the designer’s preexisting structure and the player’s cascading decisions? Jonathan Culler writes, “one could argue that every narrative operates according to this double logic, presenting its plot as a sequence of events which is prior to and independent of the given perspective on these events and, at the same time, suggesting by its implicit claims to significance that these events are justified by their appropriateness to a thematic structure.” This analysis works well enough for a pre-authored work of literature or film, but not for the interactive world of Skyrim, where the sequence of events is not determined prior to or independently from the formation of perspective. As a result of the medium’s bringing-into-being narrative via the act of play, Culler’s argument begins to fray, breaking down under the strain of transposition from literature to game. Games need a new form of analysis, a mode of understanding authorship and narrative that exists apart from our pre-established rhetorical pedagogy. Until such is established, we as critics will lack the necessary toolkit required to meet games on their own terms, to “read” them as a new and independent form of expression, and to interpret in full the complex structures and problematic dissonances inherent to the form.
Valve’s four-player cooperative zombie-themed first person shooter, Left 4 Dead, utilizes emergence elements in yet another way, splitting the difference between the nearly irrelevant presented narrative of Spelunky and the intricately constructed but ultimately inconsequential preexisting narrative in Skyrim. Players step into the shoes of one of four humans who have survived a zombie-fueled apocalypse and are now trying to survive against the hostile flesh-hungry hordes. The story of the game is communicated primarily through brief, one to two sentence interactions between the four characters and various environmental elements placed throughout the levels. Graffiti on the walls within safehouses tell stories of families split apart by the zombie plague. Rows of chairs in a park before a gazebo and a zombie in a white dress speak to an interrupted wedding. A crashed helicopter and a HAM radio tell the story of an unsuccessful evacuation attempt. There is as little or as much engagement with the story as the player wishes; if so desired, the player can simply run forward, shooting zombies and ignoring everything else in the environment. It is these zombies, and the tools that are used to combat them, that allow for an emergent narrative. Left 4 Dead uses procedural population of enemies and equipment to create an experience that is different each play session. A room infested with enemies during one session may be filled with loot in the next, or completely empty. “Special Infected” zombies, who have stronger capabilities and present a greater challenge to the player, spawn in the world according to a complex series of calculations based upon the player’s current performance in the game, and may appear either not at all or in great numbers. An artificial intelligence known as the “Director” monitors the player at all times, building out the contents of a level mere seconds ahead of a player’s experience of that level. Via the use of a complex analysis system, the Director stays one step ahead of the player, setting the scene, monitoring what it believes the tension level of the player is and working to provide a unique experience each play session. It is valuable to note that the game’s presentation is directly meant to evoke a B-movie zombie film, from the name of the Director to the film grain that overlays the screen to characters cheekily referencing classic genre films from Soylent Green to Ghostbusters to Planet of the Apes. By appropriating the techniques of a medium whose conventions of meaning are understood, Left 4 Dead hopes to sidestep many of the sticky problems and questions surrounding how games fundamentally convey meaning.
With the examples now in place, let us return to the discussion of games and film. Film is an understood, mature medium, while games are not. As Clint Hocking elucidated while describing the Kuleshov Effect in 2011, “[At t]he lowest and most fundamental level, meaning in film is generated not by the moving picture itself, really, but by the juxtaposition of the images, backed by the editing.” While Hocking utilized the Mechanics-Dynamics-Aesthetics (MDA) framework advanced by Robin Hunicke, Marc LeBlanc and Robert Zubec to make a claim that games derive their meaning, i.e. “mean,” via their dynamics, he admits that his analysis is not gospel, but is rather an attempt to solve the same issue that this essay addresses. While the MDA framework is useful in “bridg[ing] the gap between game design and development, game criticism, and technical game research,” it is not suited to provide an explanation for this fundamental question – how do games mean? This lack of fundamental understanding regarding conveyance of meaning informs the narrative question: if the medium itself is so fundamentally immature and/or not fully understood, how can a meaningful narrative emerge? A more elegant answer to the question than Hocking’s utilization of a discipline-straddling framework is this: games mean via the interactive process by which a player actually plays a game. In much the same way that the Kuleshov Effect would be meaningless within the medium of film without an audience to interpret the images being cut between, within the medium of games meaning is derived via a player’s interaction with the game itself. One might choose to view games and film as two sides of the same coin: both are visual mediums, one passive and one active. Within film, the observation of a manipulated image forms meaning, while for games interaction and manipulation of an image create meaning. This hypothesis is borne out by observing the way that the four games’ narratives manifest. In both Minecraft and Spelunky, the narrative is literally created by the interaction between player and game. By contrast, Skyrim and Left 4 Dead, with their preconstructed stories still require constant interaction in order for their stories to progress. Unlike films, which can be turned on and will play without further input even with nobody to observe them (though obviously without observation the meaning of the film cannot exist), the narrative of a game will advance without a player to advance it. Games can even take what would be a single moment of film and stretch it to an entire experience, thus “redefining the standard by which we judge film.”
The resolution of the meaning question allows us to look forward to a second question: how do games create compelling narrative? In other words, how can a game tell a story that stands on the same level as those works of literature and film which are today revered as masterpieces? An answer is found in an examination of how games utilize, or in some cases dispense with, traditional narrative structures: plot, character, theme. As most modern narrative games are digital in nature and thus bounded by the computer systems in which they exist, they are defined by their computational nature. Computation and plot have tended to sit on opposite sides of a very large room in the history of literary criticism, but more recently effort by critic Franco Moretti have worked to quantify plot, recognizing that “literary studies have experienced what we could call the rise of quantitative evidence.” His methodology consisted of creating character maps, network visualizations based upon actions and interactions. Utilizing Hamlet, a spiderweb of characters was formed, linked by vertices representing points of interaction. A digital game can have a trickier time creating such a web in one regard, as it is possible that characters who never interact in one player’s experience of a game become intimate friends in another player’s, but by the same token, the digital nature of the medium allows a computer to instantly calculate exactly who is interacting, to what degree this interaction takes place, and even how often this interaction occurred across a large population of players. Many game companies are already collecting such information; tracking not just character interactions but every possible statistic, from how often players choose to play as a certain race or gender of character to how often a specific weapon is fired. No other narrative medium is as equipped to handle large-scale quantitative analysis of its subject, and no other narrative medium is as willing. By utilizing the quantitative techniques Moretti designed to deal with questions of plot in literature, we will be able to map the data provided onto data from other mediums. Books, films, games – the structural barriers between the mediums will begin to break, allowing us to see points of intersection and divergence in narrative style and presentation more clearly.
Games are very old, but are still new as a vehicle for narrative delivery. Whether the game is physical or digital, an emergence game or a progression game, played alone or in a group or online or unwired, new and as-of-yet untapped pathways still exist for creating new stories, finding new modes of expression and building entire worlds. Without a method of understanding meaning, we can only take the smallest of baby steps on any of these journeys. It is my hope that my interpretation and analysis has sufficiently clarified the problem such that we can move forward and utilize the medium’s unique proclivity for quantitative analysis to realize the full narrative potential of the medium. Storytelling, whether it be via speech, painting or the written word, is perhaps humanity’s oldest endeavor. It is only natural that the methodology progresses alongside the rest of our species’ advances.
 For the sake of this essay, I will be writing, unless otherwise denoted, of games that are primarily concerned with telling a story and having a narrative structure. Games with other concerns certainly exist, but are not of primary interest in this discussion.
 David Bordwell, “Classical Hollywood Cinema: Narrational Principles and Procedures.” in Narrative, Apparatus, Ideology. A Film Theory Reader. (Columbia University Press, 1986).
 For the purposes of this essay I choose to reject the definition of avatar as put forth by Rune Klevjer, instead embracing what he describes as “a vehicle of communication and self-expression.”
Electronic Arts claimed that there were over one thousand variables that influenced the narrative of their Mass Effect 3, released in 2012. Justin McElroy, “Mass Effect 3 pulls in over 1,000 variables from Mass Effect 2” (Joystiq, 2010). http://www.joystiq.com/2010/06/14/mass-effect-3-pulls-in-over-1-000-variables-from-mass-effect-2/
 Jesper Juul, “The Open and the Closed: Game of emergence and games of progression,” in Frans Mäyrä, ed., Computer Games and Digital Cultures Conference Proceedings (Tampere: Tampere University Press, 2002). 323-329. http://www.jesperjuul.net/text/openandtheclosed.html
 Tanya X. Short, “Never Say Roguelike.” (Gamasutra, 2013). http://www.gamasutra.com/blogs/TanyaXShort/20131119/204988/Never_Say_Roguelike.php
 Michael Suszek, “The Binding of Isaac reaches two million in sales.” (Joystiq: 2013). http://www.joystiq.com/2013/04/21/the-binding-of-isaac-reaches-two-million-in-sales/
 Jason Rohrer, Eric Zimmerman and Jenova Chen, “The Game Design Challenge 2011: Bigger than Jesus.” Presented at the Game Developer’s Conference (2011).
 Forum Discussion thread: “VGHD #49 Assignment: Spelunky.” (The Forums of Loathing, 4 May 2012-10 Oct. 2014) http://forums.kingdomofloathing.com/vb/showthread.php?t=192836
 Reddit Discussion thread: “Your silliest, most unfortunate deaths?” (Reddit (Spelunky Subreddit), 2014)
 Latour, Bruno. Pandora’s Hope: Essays on the Reality of Science Studies. Cambridge, Harvard University Press. 1999.
 Salman Rushdie, “Video Games and the Future of Storytelling,” (BigThink, 29 Nov. 2010) http://bigthink.com/videos/video-games-and-the-future-of-storytelling
 Andrew Rollings, & David Morris. Game Architecture and Design: A New Edition, (San Francisco: New Riders, 2003). 61.
 Jonathan Culler, The Pursuit of Signs: Semiotics, Literature, Deconstruction. (Ithaca: Cornell University Press, 1981). 178.
 Mike Booth, “The AI Systems of Left 4 Dead,” presentation, (Valve Corporation, 2009). http://www.valvesoftware.com/publications/2009/ai_systems_of_l4d_mike_booth.pdf
 Clint Hocking, “Dynamics: The State of the Art,” presentation, Game Developer’s Conference. 2011. http://www.gdcvault.com/play/1014597/Dynamics-The-State-of-the
 Robin Hunicke, Marc Leblanc, Robert Zubek. “MDA: A Formal Approach to Game Design and Game Research”. (Chicago: Northwestern University, 2008)
 Hocking appears to have been approaching this in his 2007 response to Roger Ebert, “On Authorship in Games,” but stopped short of drawing the conclusion himself, as he was interested in questions of artistic merit. http://clicknothing.typepad.com/click_nothing/2007/08/on-authorship-i.html
 Jay David Bolter & Richard Grusin, Remediation. (Boston: The MIT Press, 2000). 97.
 Franco Moretti, “Stanford Literary Lab Pamphlet 2: Network Theory, Plot Analysis”. (Stanford: Stanford University, 2011). http://litlab.stanford.edu/LiteraryLabPamphlet2.pdf