# Curating Simulated Storyworlds — Chapter 1: Introduction

> NOTE: THIS IS A MARKDOWN CONVERSION INTENDED FOR LLM CONSUMPTION.
> IF YOU ARE A HUMAN BEING, READ THE PDF INSTEAD:
> https://jamesryan.computer/papers/2018_ryan_curating_simulated_storyworlds.pdf
>
> The PDF is the canonical document, typeset with every figure and table. This file
> is a plain-text conversion of the LaTeX source, published HERE so that software agents
> can read the thesis without parsing a PDF. It is lossy and is not meant for humans.
>
> This file is chapter 1 of 14 only: Introduction (~4,600 words).
> See below for links to markdown conversions of the individual chapters.
>
> **Structure.** Sections are `##`/`###`/`####` headings. Cross-references are resolved to
> their numbers ("Section 3.1.1", "Chapter 5"), so a heading can be searched for directly.
> Citations are inline bracketed keys such as `[ryan2017grimes]`; every key used in the text
> has an entry in the bibliography, formatted as `- **[key]** Author. Title. Venue, Year.`
> Footnotes appear inline as `[Note: ...]`.
>
> **Conversion limits.** Figures and most complex tables appear as `[Figure omitted]` or
> `[Table omitted]`. Code fences carry no language tag. A few diacritics in author names
> were simplified from the Latin-1 source.
>
> **All files.**
>
> Index (this list, plus the abstract; small): https://jamesryan.computer/thesis/index.md
> Whole thesis in one file (large, ~1.6 MB): https://jamesryan.computer/thesis/full.md
>
> By chapter:
> - 01. Introduction (~4,600 words, ~30 KB) — https://jamesryan.computer/thesis/01-introduction.md
> - 02. Defining Emergent Narrative (~6,900 words, ~47 KB) — https://jamesryan.computer/thesis/02-defining-emergent-narrative.md
> - 03. The Pleasure of Emergent Narrative (~21,000 words, ~135 KB) — https://jamesryan.computer/thesis/03-the-pleasure-of-emergent-narrative.md
> - 04. The Pain of Emergent Narrative (~41,000 words, ~268 KB) — https://jamesryan.computer/thesis/04-the-pain-of-emergent-narrative.md
> - 05. Refining Emergent Narrative (~11,000 words, ~73 KB) — https://jamesryan.computer/thesis/05-refining-emergent-narrative.md
> - 06. James Ryan Generator (~2,800 words, ~16 KB) — https://jamesryan.computer/thesis/06-james-ryan-generator.md
> - 07. World (~20,000 words, ~129 KB) — https://jamesryan.computer/thesis/07-world.md
> - 08. Case Study: Diol/Diel/Dial (~9,300 words, ~59 KB) — https://jamesryan.computer/thesis/08-case-study-diol-diel-dial.md
> - 09. Talk of the Town (~25,000 words, ~162 KB) — https://jamesryan.computer/thesis/09-talk-of-the-town.md
> - 10. Case Study: Bad News (~22,000 words, ~132 KB) — https://jamesryan.computer/thesis/10-case-study-bad-news.md
> - 11. Hennepin (~32,000 words, ~209 KB) — https://jamesryan.computer/thesis/11-hennepin.md
> - 12. Looking Ahead: Sheldon County (~11,000 words, ~75 KB) — https://jamesryan.computer/thesis/12-looking-ahead-sheldon-county.md
> - 13. Conclusion (~2,700 words, ~17 KB) — https://jamesryan.computer/thesis/13-conclusion.md
> - 14. Bibliography (~29,000 words, ~233 KB) — https://jamesryan.computer/thesis/14-bibliography.md

*Dwarf Fortress* is beloved for the stories that emerge out of its simulated storyworlds. It has been called the most complex videogame ever [weiner2011dwarf,tyson2012getting,cox2012dwarf,elliot2014into,davis2014justin,sears2016how], and that complexity is always in the service of emergent narrative intrigue. The scholars Stepahie Boluk and Patrick LeMieux put it succinctly: “*Dwarf Fortress* begins and ends with stories” [boluk2013dwarven, p. 133]. As a fitting testament to its complexity, someone has written an entire book explaining how to play the game: the remarkable *Getting Started with Dwarf Fortress: Learn to Play the Most Complex Video Game Ever Made* [tyson2012getting]. In its afterward, author Peter Tyson identifies the game's narrative heart:

> The joy of Dwarf Fortress lies in the game's remarkable depth and the stories that unfold from the interactions between its many elements. From world and history generation through to the individual personalities of your dwarves, each component of Dwarf Fortress involves us in amazing stories making each fortress a unique one. By now, I hope you too have a few stories to share [tyson2012getting, p. 197]

 *Dwarf Fortress* players famously do share their stories [diaz2009dwarf], on web forums [immortal2017hall], dedicated websites [dfstories], in magazines [denee2010oilfurnace], and elsewhere. *The Sims* players share stories too [burkinshaw2009alice,isbister2016games,simsstoriesandlegacies,simstories], as do players of other games [murnane2018emergent].

This centrality of narrative in *Dwarf Fortress* has been articulated by cocreator Tarn Adams himself, who has called the system a *story generator*, as the writer Jonah Weiner recounts in a feature on the game that appeared in *The New York Times Magazine*:

> Many video games mimic the look and structure of films: there's a story line, more or less fixed, that progresses only when you complete required tasks. This can make for gripping fun, but also the constrictive sense that you are a mouse in a tricked-out maze, chasing chunks of cheese. Tarn envisions Dwarf Fortress, by contrast, as an open-ended “story generator.” [weiner2011dwarf, n.p]

 So what kind of story generator is this? In the field of artificial intelligence, computational story generation has a history dating to at least 1960 [ryan2017grimes], but many systems in the area do not work like this. More often, they tell stories by modeling a top-down process of plot invention [lebowitz1985story,pemberton1989modular,gervas2005story,skorupski2007wide,mcintyre2009learning,riedl2010narrative] or creative writing [turner1993minstrel,perez2001mexica,perez2004three]. But *Dwarf Fortress* takes a bottom-up approach: a simulated storyworld teems with simulated characters living simulated lives, and when stories are created by the system, it is through the happenstance of this character simulation. [Note: To be clear, by ‘storyworld', I mean a world, real or otherwise, whose events are narratable. This could be a fictional world that is recounted in a work of literature, or a simulated world out of which narrative may emerge. Of course, the latter sense of the term is most pertinent to the concerns of this thesis. If this term feels redundant or unnecessary, given ‘world', then hold on for now—later on, I will discuss examples of computer simulations that do not seem capable of producing narrative (more precisely: generating events that may be narrated).] This is called *emergent narrative*, which brings me to the first research question that is tackled in this dissertation:

> **RQ1. What is emergent narrative?**

 In emergent narrative, stories arise bottom-up through the mechanism of computer simulation. Often this means a simulation that undergirds an interactive experience, but emergent narrative can also obtain in noninteractive contexts, such as with simulationist story generators. In Chapter 2, I handle this research question by providing an extensively deliberated definition of the form, a first published account on the origins of the term ‘emergent narrative', and a brief overview of academic work in the area to date.

But why is this even important? Emergent narrative is important because it represents arguably the most successful approach to procedural narrative in the public consciousness, through its utilization in popular videogames such as *Dwarf Fortress* and *The Sims*. This success leads me to my second research question:

> **RQ2. What is the pleasure of emergent narrative?**

 Clearly there is a distinct pleasure constituted in works of emergent narrative, but what exactly is it and what are its contributing factors? While researchers have written extensively about the merits of emergent narrative as an approach to interactive storytelling [aylett1999narrative,louchart2004emergent,louchart2004theoretical,louchart2004narrative,aylett2005fearnot,swartjes2006fabula,kriegel2008emergent,swartjes2010whose], I am not aware of any substantial treatise on the distinct experiential pleasures of the form. In Chapter 3, I attempt to unpack the success of emergent narrative by discussing the approach in terms of a series of analogies to other forms, namely nonfiction, which work to yield a distinctive set of aesthetics that I call the *aesthetics of emergent narrative*. I believe that the pleasure of the form is rooted in these analogies and characterized by these aesthetics, and with this chapter I aim to provide the most extensive apologetics for emergent narrative to date.

But while it arguably represents the most successful approach to procedural narrative in the public consciousness, emergent narrative has characterized academic story generation's most famous failures, namely Sheldon Klein's murder-mystery generator [klein1973automatic] and Jim Meehan's *Tale-Spin* [meehan1976metanovel]. In the decades following those early systems, scholars and practitioners in the area have generally come to view the approach as at best a dead end, namely for the reason that emergent narrative tends to lack story structure. Here, an excerpt from a paper by *Facade* [mateas2003faccade] creators Michael Mateas and Andrew Stern expresses this critique:

> But for believable agents used as characters in a story world, strong autonomy becomes problematic. Knowing which action to take at any given time depends not just on the private internal state of the agent plus current world state, but also on the current story state, including the entire past history of interactions building on each other towards some end. The global nature of story state is inconsistent with the notion of an autonomous character that makes decisions based only on private goal and emotion state and local sensing of the environment. [mateas2002towards, p. 223]

 Mateas, who is my coadvisor, has proffered a more compact critique of emergent narrative in conversation: *it's just one damn thing after another*. While Ruth Aylett, Sandy Louchart, and collaborators—who carry the term ‘emergent narrative' as a banner—have worked to cultivate the approach in the specific domain of interactive storytelling [aylett1999narrative,louchart2004emergent,aylett2005fearnot], its critics have generally remained skeptical [mateas2003integrating,riedl2004intent,fairclough2004story,riedl2006linear,riedl2010narrative,young2013plans,robertson2018perceptual]. This raises my third research question:

> **RQ3. What is the pain of emergent narrative?**

 In Chapter 4, I unpack a series of harsh critiques of emergent narrative—namely ones proffered by Espen Aarseth and Noah Wardrip-Fruin—to identify a series of pitfalls that have worked to hamper the form. These hazards operate at both the level of underlying simulation and the level of surface expression, working to inhibit either the emergence of story structure or its identification by human observers (or both). This is the pain of emergent narrative, and with this chapter I aim to provide the most brutal critique of the form to date.

With this double-edged foundation, we arrive at the fundamental research question that is tackled in this dissertation:

> **$$ RQ4. How can the pain of emergent narrative be alleviated while simultaneously maintaining its pleasure?**

 One potential treatment for the pain is to *intervene* in the storyworld to remediate its affairs such that story structure reliably obtains in each simulation instance. Such intervention characterizes technical schemes like *drama management* [mateas1997oz,roberts2008survey] and *plot-level narrative planning* [young1999notes,riedl2010narrative], which employ omniscient agents that dramatically remediate a storyworld according to policies about story structure. But while such remedies may alleviate the pain of emergent narrative, they in turn kill the pleasure: when a simulated storyworld is polluted through the intervention of an external system, events no longer emerge, but instead spawn according to the policies of the modulating system. When this happens, emergent narrative ceases to manifest, and its distinctive pleasures disappear in turn. [Note: At its heart, this dissertation is an art manifesto that promulgates a fiercely *emergentist* approach to procedural narrative. It is written on behalf of a contingent that appreciates simulated worlds as real places where tiny abstract characters live out tiny abstract lives—this does not necessarily reflect an ontological position, but more essentially an *aesthetic orientation*. As part of a rhetorical strategy for promoting this orientation, in this document I use value-laden terms like ‘pure', ‘polluted', ‘tampered', and more to describe a variety of architectural patterns that are probably equally valid from an intellectual and technical standpoint. As such, my charged language is meant to instantiate this artistic position, rather than to express technical or intellectual appraisal. That is, if I for instance decry an architectural pattern that ‘pollutes' a storyworld, I mean to do so purely in terms of the aesthetic orientation for which this dissertation serves as a loose manifesto—even though such a pattern may be incompatible with this aesthetic orientation, as cued by my usage of value-laden terms, it could perfectly valid from any other standpoint. I believe that variation in an underlying architecture yields variation in terms of what is aesthetically possible at the surface level of a work of computational media—this is what Michael Mateas has termed the *interpretive affordances* of an architecture [mateas2001expressive]—and in this dissertation I introduce an architecture that was formulated expressly to target a specific set of interpretive affordances, which in turn anchor an intellectual framework that is the primary subject of this work. I will periodically return to these concerns below, but for now you have been warned: this is an emergentist manifesto.]

This dissertation identifies an alternative mechanism that alleviates the pain of emergent narrative while simultaneously maintaining its pleasure: *curation*. Instead of intervening in a storyworld to *impose* story structure, the world is curated to *identify* stories that authentically emerged. The idea here is that well-crafted storyworlds do not actually lack narrative intrigue, but rather narratively potent material tends to be embedded in larger accumulations of unremarkable simulated material. This means that emergent narrative may actually have story structure, but the simulated material composing such structure may be embedded in a larger accumulation that itself lacks story structure. Thus, when human players tell stories about what has happened in *Dwarf Fortress* or *The Sims* gameplay, they are in fact carrying out a process of curation—narratively potent material is extracted and then assembled into narrative artifacts with discernible story structure.

What if a system did this automatically? This is the impetus for the subject of this dissertation: *curationist emergent narrative* (or *curationism*, for short). This framework represents a refined approach to emergent narrative that entails the curation of raw simulated material into actual narrative artifacts that are mounted into full-fledged media experiences. This procedure is instantiated in a generalized curationist system architecture, which is itself inspired by Hayden White's model of historiography as a constructive storytelling procedure [white2014metahistory]. In the architecture, a module called the *chronicler* inscribes the transpiring phenomena of a simulated storyworld in a *chronicle*, which a *story sifter* sifts through to extract material that a *narrativizer* uses to construct a *story*, which is finally mounted into an actual *media experience*, thereby enabling human encounter. This makes story generation a process not of *inventing* stories, but of *recounting* stories about what has actually happened in a world. In Chapter 5, I discuss this framework in terms of its intellectual position, technical approach, generalized architecture, and variants of that architecture.

In the process of outlining the curationist approach in that chapter, I incidentally handle another research question that is implied by the previous ones:

> **RQ5. How has emergent narrative been arguably the most successful approach to procedural narrative and at the same time perhaps its least successful approach?**

 I contend that curationism actually accounts for how certain works of emergent narrative have been tremendously successful, while others have been outright failures. Again, the answer is simply this: curation. Projects like *Dwarf Fortress* and *The Sims* have flourished because human players are willing to curate their simulated storyworlds so that manifest emergent narrative may be encountered (thereby unlocking the aesthetics of the form). Such curation may result in genuine narrative artifacts [dfstories,simsstoriesandlegacies], which may be mounted in full-fledged media experiences [denee2010oilfurnace,burkinshaw2009alice], or it may take the form of a kind of unconscious narrativization that drives the basic interpretative procedures that humans employ to make sense of an ongoing simulation (or any experience). I call the latter phenomenon *mental curation*, and in Section 5.3.2 I show that it instantiates a degenerate variant of the curationist architecture.

On the other hand, when human observers are *not* willing to do curation, a work of emergent narrative will fail miserably. For instance, Sheldon Klein's much derided story generator produces monolithic reports on all the various happenings of its underlying simulation, and the result is a jumbled mess that humans have typically been unwilling to curate into actual emergent narrative (thereby precluding the aesthetics of the form). Thus, success in emergent narrative has typically been contingent on whether or not humans are willing (or enabled) to curate a simulated storyworld. In the case of videogames like *Dwarf Fortress* and *The Sims*, humans have largely been willing because, through gameplay, they develop attachments to the storyworld and its characters. This is not typically the case in simulationist story generation, and so we find a generally negative appraisal of that approach. But what if systems automatically curated their own storyworlds?

So far I have outlined the first half of this dissertation, *Part I: Why and How*, which is about why curationism is a viable framework and how it might be instantiated in actual projects. In total, this half represents what is to my knowledge the most extensive (and most comprehensive) treatise on emergent narrative to date. The second half of this dissertation, *Part II: How and Why*, approaches curationism from the opposite direction—this half is about how I have instantiated the curationist method in my own work and why it appears to have been a good approach. In this way, it wrangles another essential research question:

> **RQ6. What would a curationist project actually look like?**

 Given my articulation of the curationist approach, such a project would be sure to incorporate a simulated storyworld, curation, and a media experience. As I explain while outlining variants of the curationist architecture in Section 5.3.2, the media experience and storyworld may be coextensive or decoupled, and curation may be carried out by one or both in either an online or offline manner.

I have spent the last six years developing a series of simulation engines—*World*, *Talk of the Town*, and *Hennepin*—that are intended to drive *second-order media experiences* set in their simulated storyworlds. These engines are each the subject of a dedicated technical chapter in the second half of this dissertation. In tandem with the development of these engines, I have worked independently and in collaboration to develop a series of experiences set in the worlds that are created by those systems. Following each chapter reporting a simulation engine, I provide another, in tandem, that outlines an experience that was driven by the engine. These experiences are: *Diol/Diel/Dial*, a hypertext world encyclopedia that compiles the massive total history of a *World* world; *Bad News*, a collaboration with Ben Samuel and Adam Summerville that takes the form of an installation-based immersive experience (with a live actor) set in a *Talk of the Town* town; and *Sheldon County*, an ongoing project concerning the procedural generation of narrative podcasts each recounting a unique *Hennepin* storyworld. While the simulation chapters report technical contributions, the experience chapters are meant primarily as case studies that concern the instantiation of curationism in the experience at hand (or the fatal lack thereof). In turn, they answer my sixth research question by providing a troubling example of missing curation, a powerful example of human-powered curation, and a first exploration of fully automatic curation. Moreover, through the articulation of a set of actionable strategies that I have developed both in my simulation engines and the media frameworks that curate their generated material, this series of chapters answers the following additional research questions:

> **RQ7. What are some design patterns for crafting simulated storyworlds in ways that facilitate curation?**

**RQ8. What are some design patterns for curating simulated storyworlds?**

I should include a brief note on the research methods employed in this dissertation. As for Part I, my method is scholarly argument in the humanist style, which means I have succeeded to the degree that you are convinced. Like I have mentioned, technical contributions are reported in Part II, but I do not personally see much value in them. As others have articulated before me, computational techniques that are intended to support media experiences cannot be truly appraised except through actual implemented experiences of the kind that they are meant to support [mateas2001expressive,stern2008embracing,horswill2013science,koenitz2014five,wardrip2014envisioning]. From this orientation, technical contributions in computational media are meaningless, and in fact not contributions at all, when they are divorced from the media contexts for which they are intended. In this dissertation, the real appraisals of my simulation engines appear in the case studies of the media experiences that they enable, since it is only in those contexts that the technical approaches taken by the engines can be evaluated. As such, my primary evaluation scheme in Part II is the *case* [berlant2007case], and in fact this answers the call of Ian Horswill, who is on the reading committee for this dissertation:

> computer science is traditionally, and understandably, resistant to work on cases. One cannot generally get a Ph.D. for writing a program; one gets a Ph.D. for developing a general theory that will help others write some class of programs. But I see no alternative but to relax this constraint for AI work that has specifically aesthetic goals. We have to be open to the case because the case is all we have. [horswill2013science, p. 84] [Note: This quotation originates in a paper titled “Science Considered Harmful”, to which R. Michael Young has recently written a rebuttal titled “Science Considered Helpful” [young2018science]. The dialogue affords an illuminating look into the nascent field of computational media, as it attempts to define itself.]

As I will make clear throughout the rest of this document, the curationist framework of Part I actually coevolved along with the simulation and media practice that I recount in Part II. In this way, and in a certain sense, Part I tells a story about Part II, which itself tells a story about Part I. As I was preparing to write this dissertation, Jonathan Lessard, another member of my reading committee, encouraged me to outline the evolution of my practice using the method of *autoethnography* [ellis2011autoethnography]. I took that to heart, and as a result each chapter in Part II begins with a personal recollection of the intellectual and technical contexts in which the subject of that chapter was developed. As such, this dissertation wrangles a final research question, and it is perhaps the most difficult one:

> **RQ9. How did I get here? [Note: While I include this somewhat lightheartedly, I do think it is a serious question, especially in the context of a PhD thesis. As a digger in the historical record of computational media, I have read through many dissertations that were published decades ago. When I do so, I am generally more interested in the intellectual contexts in which the student carried out the subject project than in the technical and scientific details that are reported in the document. As such, acknowledgments sections are frequently more potent (from a contemporary intellectual standpoint) than dissertation bodies, because in the latter there is a tendency to eschew any recounting that may be perceived as personal. When a scholarly work lacks contextual information concerning its own development, it will be at best incomplete and at worst intellectually dishonest, and so in this thesis I seek to provide such information freely and abundantly. While all dissertations are personal narratives, this one is more personal than most.]**

This is the first ever dissertation in computational media. The Computational Media PhD program at UC Santa Cruz started up last year [stephens2016uc] as an offering of the Department of Computational Media, which was itself established in 2014 [stephens2014uc,koenig2014new]. Prior to that, a group of leading scholars in the field convened on UC Santa Cruz in the summer of 2012 under the moniker of the *Media Systems* project. [Note: A project website is hosted here: https://mediasystems.soe.ucsc.edu.] In “Envisioning the Future of Computational Media: The Final Report of the Media Systems Project”, my coadvisors Noah Wardrip-Fruin and Michael Mateas articulate how work in this emerging area weaves technical, creative, interpretative, and collaborative threads to produce four kinds of outcomes:

> *Artifacts*: the outcome of making novel computationally-driven media.

*Capabilities*: the outcome of developing computational, representational, and design approaches that enable new forms of media.

*Insights*: the outcome of studying the technical, historical, and cultural creation and function of computationally-driven forms of media, both old and new.

*Educated Practitioners*: the outcome of interdisciplinary education and training in computational media. [wardrip2014envisioning, p. 21].

 In turn, the requirements for this nascent PhD program stipulate that a dissertation in this area should utilize these various modes of work in turn (or in tandem) to produce outcomes of the sorts just listed:

> The Ph.D. dissertation is a significant contribution to the CM field. It may emphasize technical, media-creation, or interpretive aspects, but must have a significant secondary contribution from at least one of the other areas—and may also emphasize areas equally.

 This dissertation is the result of technical, creative, interpretative, and collaborative work that has resulted in a series of artifacts, capabilities, and insights, as well as one educated practitioner. To be more specific, in this document I provide the following contributions:

-  *A definition of emergent narrative.* This definition is arrived at through an extensive deliberative process, and it is meant to join three distinct traditions that I view as instantiating the same essential form: the simulationist school of noninteractive story generation, as exemplified by systems like *Saga II*, Sheldon Klein's murder-mystery generator, and *Tale-Spin*; videogames driven by emergent character simulation, such as *Dwarf Fortress* and *The Sims*; and the bottom-up approach to academic interactive storytelling that is exemplified by the ‘emergent narrative' research programs of Ruth Aylett, Sandy Louchart, Mariet Theune, Ivo Swartjes, and their respective collaborators. This contribution appears in Chapter 2, though my inclusive treatment of both interactive and noninteractive works is reinforced throughout this document.

-  *An apologetics for emergent narrative.* While others have argued extensively about the merits of interactive emergent narrative, in this document I work to unpack the essence of character simulation itself, which results in the identification of analogies to nonfiction, lived experience, worldbuilding, and art brut. Here, in particular, I spend considerable effort arguing how emergent narrative actually works more like nonfiction than fiction, which I do by reviewing an extensive array of definitions of nonfiction that I then apply to a work of curationist emergent narrative—Tim Dinee's *Oilfurnace* [denee2010oilfurnace]—to show that the definitions are satisfied. Finally, I identify a set of aesthetics that are yielded by these analogies, which I call the *aesthetics of emergent narrative*. This contribution appears in Chapter 3.

-  *A polemic against emergent narrative.* In addition to providing what is perhaps the most extensive defense of emergent narrative, I also contribute what may be the harshest attack on the form to date. In this document, I unpack and extend the criticisms of emergent narrative proffered by various writers including Espen Aarseth and Noah Wardrip-Fruin to identify a series of pitfalls at the levels of both simulation and curation. As to the former, I primarily articulate how the crafting of character simulations often fails to target emergent structures that have the discernible features that humans associate with narrative. In a section on curation issues, I unpack Aarseth's denouncement of a troubling (and ubiquitous) conflation of simulation and narration—the raw transpiring of simulation (or a trace of that unfolding) is called a story—to assert that actual emergent narrative can only manifest through a procedure of curation. This contribution appears in Chapter 4.

-  *A refined approach to emergent narrative.* This comes in the form of my framework for curationist emergent narrative, which is the subject of this dissertation. In producing this contribution, I proceed from the material contained in both my apologetics and my polemic for the form. Here, my challenge to myself was to devise a refined approach to emergent narrative that takes its harshest appraisals as an intellectual starting point. This is the primary contribution of this dissertation, but Chapter 5 is specifically devoted to the curationist framework.

-  *A set of artifacts, in the form of three simulation engines and three media experiences.* These are the completed simulation engines *World* and *Talk of the Town*, the completed media experiences *Diol/Diel/Dial* and *Bad News*, and the in-process engine *Hennepin* and in-process experience *Sheldon County*. These systems and experiences are each the subject of dedicated chapters that appear throughout Part II of this dissertation.

-  *A set of case studies in curationist emergent narrative.* For each of the three media experiences that I report, I provide a case study that contemplates the experience in terms of the curationist framework that is the subject of this dissertation. In total, these case studies provide an actionable account both of how curationism may be carried out and why that is worth doing. As to the former, I identify a series of design patterns at the levels of both simulation and curation.

-  *An autoethnography of my simulation and media practice.* Across the openings of the chapters in Part II, I tell the story of how my simulation and media practice evolved over time. Incidentally, these case studies also explain how the curationist framework at the heart of this dissertation was developed, since all of this coevolved in the Petri dish of my life. This is a contribution because it elaborates on the core contributions of this thesis by explaining why I made the decisions that I made, why the decisions were good ones or bad ones, and critically how I came to appraise them later on.

-  *A history of emergent narrative.* For the last year, I have been carrying out a massive project involving the excavation of materials pertaining to the forgotten history of computational media. [Note: My first outputs from this project are a conference paper on a forgotten 1960s story generator [ryan2017grimes] and a workshop paper that provides an extensive account of a 1578 work of interactive branching narrative [ryan2018garden].] While this dissertation is not an historical study, this has been my mindset recently, and so inevitably I provide historical context along the long path of this document. This includes a tracing of the origins of the term ‘emergent narrative', which I do not believe anyone has provided before; a first substantial account of the earliest known work of emergent narrative, *Saga II* (1960); an excavation of Sheldon Klein's murder-mystery generator, working in the style of Noah Wardrip-Fruin's study of *Tale-Spin* to show how the system has for decades been misunderstood; and numerous details here and there on the history of computational narrative and computational media, more broadly.

With this thesis being the first of its kind, in a sense the most essential question that I am attempting to answer here is this:

> **What is a computational media dissertation?**

 I hope that I have done well to answer this question. As you journey through this tome, godspeed!