Creating a Sense of Coherence in Open-World Adventure and Role-Playing Game Stories

The following is my most recent explanation of my thesis project.

We are interested in the application of interactive storytelling to videogames.  We want to improve story experiences in open-world adventure and role-playing games.  A game that features an open world allows its players to move freely in a large space with few or no artificial barriers, choosing what to do and when.  The flexibility of an open world and the fact that adventure and role-playing games tend to have strong story components make these genres an interesting place to explore interactive storytelling techniques.

Our central goal is to support the creation of open-world videogame stories that give players a sense of coherence.  To achieve this, we take a structuralist approach and partition stories into two types of scenes inspired by the concept of kernels and satellites.  First, a minimal set of fixed scenes form a core story with strong authorial control.  A game’s most central plot points become fixed scenes, thus acting like kernels.  The rest of the story emerges from a much larger collection of flexible scenes that can appear just about anywhere in story save a small set of preconditions.  Most flexible scenes act like satellites: minor plot points, or opportunities to develop story elements like theme.

We want to give players the freedom to explore flexible scenes however they wish as they move through the fixed scenes as designed.  A certain level of coherence is guaranteed when the content of the fixed scenes is itself coherent, but a story with few satellite scenes will have minimal aesthetic appeal.  The challenge, then, is to maintain coherence no matter how a very large set of flexible scenes is experienced.

Instead of arranging flexible scenes according to a strict definition of causal coherence, we want to create a “sense of” coherence.  By this we mean that not all events have to be causally related in explicitly obvious ways, but that players should have the sense that they could figure out the meaning of and relationships between events if they thought hard enough about it.

One of the major ways we achieve a sense of coherence is by managing the story’s progression.  We keep track of when certain story elements, such as theme and character, are reflected.  We then prioritize which scenes should be made available to players next according to a desired distribution of the story elements.  For example, if a particular theme was developed very recently, we want to prioritize scenes that reflect some of the other themes.  On the other hand, if it has been a long time since a theme was developed, scenes that reflect that theme strongly should have high priority.  A good distribution of elements ensures that story elements don’t feel out of place when developed, and that reminders of previous scenes are made throughout the story.

Another facet of creating a sense of coherence is the emergence of structure at run-time through the use of conditions.  Instead of defining causal relationships in a scene graph a priori, we allow authors to define prerequisites for their scenes.  Using prerequisites is a common technique, but in our design we push for prerequisites based on story state values in addition to game state.  For example, scenes might have prerequisites that only allow them to be seen once a particular theme has been developed sufficiently.  Alternatively, a scene might be best suited for the early development of the theme, and should not appear later on.  We want authors to think about flexible scenes in terms of how they function in a story’s development without having to worry about how they will fit within a series of causally related events.

In addition to controlling the path players take through a set of fixed and flexible scenes, we can improve the sense of coherence by adjusting the content of scenes.  In so doing, we want to give players interpretative agency: they should feel like there are deeper layers in the story not being explicitly told, and they should feel like they can interpret those layers in a reasonable way.

We are exploring three ways of dynamically affecting the content of scenes.  In the first, run-time criteria is used to choose a set of scenes that a recurring motif (say, an apple) can be featured in.  Observant players will begin to notice the motif over time and assign meaning to why it appears in certain scenes.  Eventually, they will expect something in particular to happen when a new scene with the motif begins.

Second, mix-ins give us pre-scripted opportunities to make connections to scenes the player happens to have already seen.  As Keith Johnstone points out in the context of improvisation, “feeding something back in from earlier in the story adds ‘point’ and creates structure.”  Characters, story elements, and dialog are all examples of source material that could be referred to in future mix-ins.

Finally, we can adjust the presentation of a scene to alter the player’s interpretation of otherwise unchanging events.  Choice of lighting, background music, camera angles, and even the weather can all depend on the story’s state at the time a particular scene is reached.  Perhaps the heroine of the story returns to the castle with the head of a dragon.  The mood evoked during the scene might be bright and cheerful if the player saw the dragon as an evil menace.  However, the mood might be more sombre if the player found out that the dragon was simply a loving mother trying to protect her hatchlings.  The final event stays the same, but the interpretation of it changes.

In summary, our goal is to give players a sense of coherence when exploring stories in open-world adventure and role-playing games.  We structure our stories as a set of fixed and flexible scenes.  Players can traverse the set of flexible scenes freely, barring any prerequisites that deem certain scenes inaccessible.  Flexible scenes are prioritized so that story elements are well distributed throughout the story.  We encourage interpretative agency by dynamically introducing recurring motifs, using mix-ins to make connections to earlier points in the story, and modifying the presentation of a scene to affect interpretation.  Through all of this, higher quality open-world stories will emerge while still maintaining a satisfactory level of interactivity.

Keynotes and Inspirations at Foundations of Digital Games 2015

I had a great time at this year's Foundations of Digital Games, but it was the talks and the hallway track with narrative folks that really left me inspired.  In this post I'll summarize the first three keynotes and some inspiration I got from one of them; you can check out my raw talk notes and the proceedings for more details.  The sketch notes I'm including are all by Chris Martens, a fellow narrative PhD candidate (though she'll be done soon!).

Tom Forsyth's Keynote

The first keynote of the conference was given by Tom Forsyth, who worked at Valve and then Oculus VR.  Tom highlighted some of the challenges we face when designing all-new experiences for virtual reality.  For example, motion sickness comes from there being an imbalance between what your eyes see and what your ears feel, so having players move up stairs causes issues (elevators are apparently much better).  It's also important to avoid most cinematography techniques, and to use an eye blink transition whenever possible.

[raw talk notes]

Sketch note by Chris Martens, via Twitter

Robin Hunicke's Keynote

The next day we heard from Robin Hunicke, who produced Journey and is seriously inspirational.  She began by talking about Wattam and its designer Keita's inspiration for the game.  The zaniness levels of that game are right up my alley, making me want to get a PS4 even more...

Most of Robin's talk was about her inspiration for Luna, which as she puts it, was kind of a beast.  She talked in depth about the inspiration noted on the game's website: "origami, shadow boxes, abstract sculpture and minimalist illustration."  Near the beginning of the design process, she apparently spent six months alone in her apartment, folding paper.  It turned out that folding paper digitally wasn't very fun, but that didn't mean some of the lessons learned from origami couldn't apply to a game's design.  Luna looks wonderfully whimsical, and I am hoping I was mistaken about it being a VR game (or at least, not VR-exclusive) because I would really like to try it.

[raw talk notes]

Sketch note by Chris Martens, via Twitter

R. Michael Young's Keynote

The first academic keynote of the conference, R. Michael Young is in charge of the Liquid Narrative Group at NCSU.  He described some of the main areas his group looks at in the world of narrative after reminding us that "narrative is big. Really big."

The systems his group builds are evaluated based on whether they produce narratives that can be understood by humans in particular ways.  They break narrative down into story (everything that happens inside the story world), discourse (the choices the author makes in how to tell the stories; what goes into the telling), and interactivity (what happens when a player goes into the role of a character? How do we design the story and discourse?).

The group's most used tool seems to be artificial-intelligence-based planners.  Planning looks at how to automatically sequence actions in the face of a novel set of environmental goals.  In narrative, this might mean anything from having characters scheme to achieve their own goals, to authors planning to mislead then reveal.  Stories are broken down into the smallest possible units (such as individual actions), then built back up.  Many problems arise from the use of standard planners, which often tell uninteresting stories.  One of the ways to improve the outcomes is the group's current work in progress that attempts to express character traits through the action sequences created.

[raw talk notes]

Sketch note by Chris Martens, via Twitter

Inspiration

Michael's distinction between story and discourse got me thinking about my own thesis project (read a somewhat out-of-date description here).  I realized that a large part of what I'm doing feels more like discourse than story.

It felt even more clear to me when I thought about Mieke Bal's definitions of story and fabula from Narratology: Introduction to the Theory of Narrative.  A fabula is "a series of logically and chronologically related events that are caused or experienced by actors" while a story is "a fabula that is presented in a certain manner."  Most of the techniques I am working on actually don't affect the fabula so much as the way the fabula is experienced.

One of the areas I struggled with in my thesis proposal was justifying why I didn't want to use planners.  There are some definite reasons that were easy to articulate, such as difficult authoring.  However, I also had this unarticulated understanding that planners weren't quite right for the design approach I took, but wasn't sure exactly why.

By thinking about story and fabula, I was able to realize that I'm not trying to arrange actions into a story so much as allow navigation through a set of coarser story pieces featuring fixed actions.  I want players to be able to explore a mostly fixed fabula in different ways, leading to different interpretations of it.  In the process, the resulting story should still have a sense of (but not necessarily actual) coherence.

As a result of this insight and another cool idea that came up during the conference, my thesis project's focus is tightening up very nicely.  I'll share more about that sometime in the future.

Spring Research Update

It's been a while since I last did a research update. With spring (sort of) arriving, there's no better time to reflect on a winter's worth of hard work.

First / Dean Gugler 

Coherent Emergent Stories

Most of my effort in the last 8 months has been dedicated to teaching, but despite this, I managed to make progress on my PhD and thesis project.  In the fall, I spent time putting together my thesis proposal, trying to make the content as close to final-thesis quality as I could.  Then I proposed in December.

Since then, I have been dabbling with a next-iteration prototype to test my story ideas.  Instead of trying to craft an entire game, I am focusing on what I call a "story explorer."  I am designing the prototype to be as data-driven as possible so I can quickly and easily test many different stories and approaches to arranging those stories with my story engine.

Gram's House

The Gram's House project is a labour of love, and I am so excited to see how far it has come since I came up with the idea years ago.

Lately we've been hard at work on the NSF AISL Pathways grant we were awarded to study the effect of story on teaching computer science concepts with games to middle school girls.  We have been working on prototypes for three analog games to be used in informal settings.  The game cover the concepts of data representation (specifically images), data organization (searching and sorting), and algorithms (writing and reading precise instructions).

It has been a lot of fun coming up with the game designs, but also very challenging.  I really want to make sure we have something more than a lightly gamified activity.  I want the games to have inherently interesting and motivating goals that happen to require understanding of our CS concepts to achieve.  I want the games to present interesting and meaningful choices to players, and have at least some degree of replayability.  I'm not sure that our current games have all these features, and I am convinced that we can come up with even better designs.  Hopefully our resident story and game design expert Lorraine Hopping will stay patient with my constant pushing, because she has been an amazing asset to this project and has a lot more experience than I do!

Something else exciting is that two of my first year students may be joining the procedural content generation grant team at Northeastern University in Boston this summer.  I am beyond thrilled to be able to enable this kind of opportunity, and I can't wait to see what they are able to accomplish.

In addition to the summary of Gram's House on my own webpage, we have started an official project site hosted by Northeastern.  We are still working on adding content, but that should be a good place to find information about the project in the future.

How Gameplay Affects Stories in Games

I am very interested in the role of storytelling in videogames.  Although I do like games without stories as well, the story is often the part of a game that engages me the most.  For my thesis proposal, I spent some time thinking about how stories integrate into games, and realized that some of the existing ways to analyze them didn't quite suit my needs.  Below is an excerpt from my proposal discussing a spectrum I came up with instead.

We are interested in the application of interactive storytelling in videogames. In particular, we want to create a more satisfying story experience in open-world adventure and role-playing games. A game that features an open world allows its players to move freely in a large space with few or no artificial barriers, choosing what to do and when. The flexibility of an open world and the fact that adventure and role-playing games tend to have strong story components [1] make these genres an interesting place to explore interactive storytelling techniques.

Combining IS and games introduces a new challenge: a balance must be achieved between creating a satisfying story and maintaining a reasonable level of player control in both the story and gameplay. This is no easy task, as Costikyan points out: “Divergence from a story's path is likely to make for a less satisfying story; restricting a player's freedom of action is likely to make for a less satisfying game” [2]. It also brings an important difference between interactive storytelling and storytelling in games to the forefront. In traditional interactive storytelling, the IS system is primarily responsible for arranging the story being told, making adjustments according to user interactions. When telling interactive stories with games, however, it is the gameplay that drives the story forward and determines how it should unfold. Gameplay is what we call the interaction between players and a game system's rules [3]. Gameplay includes physically moving through an open world's space, fighting enemies, conversing with non-player characters, and collecting items.

Figure 1: Our spectrum of how storytelling integrates with gameplay

Some gameplay actions might allow players to interact with the game's story elements, and others can indirectly affect the direction of the story. Viewed as a spectrum (Figure 1), the least interactive types of game stories come as a result of gameplay not having any effect on the story at all. The most interactive stories are directly affected by gameplay. In the middle, stories are affected by gameplay, but only indirectly. Many authors who wish to create compelling story experiences in games aim to create experiences as far right on the spectrum as possible, but several challenges make this more difficult than it may seem.

 Figure 2: Lebowitz and Klug's storytelling in games spectrum

Lebowitz and Klug [1] also proposed a spectrum to analyze videogame stories (Figure 2). As presented, non-interactive stories are at the far left. The level of control that players have over the story increases to the right of the spectrum. Implicitly, the spectrum also captures how much control an author has over a story, with the largest level of control to the far left. Furthest right, the author has no control at all; games that Lebowitz and Klug call fully player-driven are highly emergent with no set plot.

We are more interested in how directly gameplay affects a story rather than the magnitude of control the player has, partly because gameplay and story often feel like separate activities. Furthermore, most open-world games we are interested in fit into the open-ended story category of Lebowitz and Klug’s spectrum, which they characterize as games that allow players to progress how they wish and that often (but not always) feature open worlds. Some open-world games allow players to explore and complete various tasks without any relation to the story. Others have similar tasks to complete, but doing so will cause the story to change in some way. We want to be able to distinguish games within the open-ended category, and we are able to do so with our spectrum.

In the following, we discuss the types of strategies used to implement game stories across our spectrum and discuss problems that arise for each. We begin at the far left of our spectrum, where gameplay has no effect on story. This does not necessarily mean that the game is not an example of interactive storytelling, as the player may be able to interact with the story in ways that do not change it. The player might choose which parts of the story to consume and when, explore the world the story is set in, or interact with the characters within. Games in the Zelda series, such as The Legend of Zelda: Skyward Sword, feature fixed stories but allow the player to converse with non-player characters and take on missions that have no story consequences. Not allowing gameplay to affect a story results in stronger authorial control over how the story unfolds. However, it also introduces a risk of creating gameplay that does not align well with the narrative being told, especially in games that offer much freedom in terms of how to approach gameplay. A story told about a hero that in gameplay can rob innocent citizens at will leads to ludonarrative dissonance, a term originally coined by Clint Hocking. Red Dead Redemption openly tracks the player's heroism, but it does not matter how heroic the player is deemed to be: non-player characters treat the player character the same way, and the same tasks are required of him. The worst case of a story not affected by gameplay is a “constipated story,” where strictly alternating presentations of gameplay and story fail to interact with each other at all [4].

Stories that are indirectly affected by gameplay lie in the middle of our spectrum. With such games, players do not directly make choices that alter the course of a story. Instead, players interact with the game system, resulting in changes to the game state. How the story is presented or arranged depends on the current game state. For example, a portion of a story might only be available if the player character has enough experience points. Many areas of Fallout 3 were technically available all the time, but the player could only survive long enough to explore them after gaining enough experience points through gameplay elsewhere. Alternatively, the outcome of a plot point might depend on how evil the player acted in previous encounters. In Fable II, the player has opportunities to be good or evil through various tasks. Which approach is chosen unlocks some new tasks, and changes the player character's appearance on screen. Some games with branching stories, including those with multiple endings, are also found near the middle of the spectrum. In BioShock, for example, gameplay actions indirectly cause the story to change. The game ends differently depending on what the player decides to do with certain characters that can either be spared or harvested for player benefit.

The structuralist approach of defining an implicit graph at run-time could be applied here; gameplay statistics can be used in scene preconditions, for example. Although players can only affect the story indirectly, most players will be satisfied. In two surveys of game players, Mallon and Webb [5] discovered that players actually prefer episodic and directed story experiences over unrestricted freedom, and Lebowitz and Klug [1] found that while the majority of players place great importance on a game’s story, they do not require full control over it.

Games with quests commonly feature gameplay that indirectly affects story [1, 6]. A quest system in a game is used to organize what quests are available and when. It will also take care of offering quests to the player, possibly through conversations with non-player characters or markers in the world the player interacts with to trigger a quest. Many games with open-ended stories feature quest systems. Each quest contains a fragmentary story and provides short-term gameplay goals [7, 8]. Story consistency is enhanced by making the quests self-contained and largely independent [8]. As a result, completing a quest reveals a portion of the game's story that the player would not otherwise see, but rarely affects any other part of the story. The only indirect effect of choosing a quest through gameplay is the addition of the non-essential bit of story contained within. The stories revealed in quests would likely be more interesting if they connected better with what happens in the core plot; instead, they often feel like busy work, or little more than a way to improve your character's statistics.

If quests were less task-based and more goal-oriented, they might lie closer to the right of our spectrum where gameplay directly affects story. Quests in videogames tend to focus on tasks such as fetching items or engaging in combat instead of higher-level goals that can be achieved in various ways [7, 9, 10]. Instead of asking players to complete tasks, quests could be used to introduce meaningful conflict between characters that the player must resolve through gameplay, allowing for more direct interaction with the story. Games in the Mass Effect series use quests to introduce conflict between characters. Sullivan's GrailGM framework [10], designed to work with game mechanics based on social moves, also allows for conflict-oriented quests. GrailGM makes quests available based on the current social dynamics between player and non-player characters in the story. Players have the choice to complete the quest as put forth by the quest giver, or to create conflict by going against the quest giver's wishes. If more quests were explicitly connected to the game's main storyline, the gameplay act of choosing a quest could give a stronger sense of affecting the story.

There are games that more clearly lie to the right of our spectrum. Some games make their core mechanic a direct interaction with the story, as in Telltale Games’ The Walking Dead series. In The Walking Dead, players are primarily asked to decide how the main character should respond to various situations. Heavy Rain offers similar gameplay with one interesting difference. When players fail to complete an action move, the game does not ask them to try again. Instead, the following scene or scenes continue, but the exact content is affected by whether the player has failed or succeeded. Sometimes gameplay as simple as which quests the player chooses to embark on can directly impact the story. For example, Fallout: New Vegas tracks statistics about who the player is most loyal to. Which quests the player completes can affect loyalty and, by extension, how certain parts of the story will play out.

A challenge of games whose gameplay directly affects the story is to find ways to allow players to make meaningful dramatic decisions. The Walking Dead is a game that almost entirely consists of what appears to be dramatically significant choices. However, most choices the player makes impact only the next line of dialog, nothing more. While perceived agency can be important [11], it is easy to have the effect wear off when few or none of your choices are dramatically relevant. 

---

[1] J. Lebowitz and C. Klug. Interactive Storytelling for Video Games. Focal Press (2011).

[2] G. Costikyan. Second Person: Role-Playing and Story in Games and Playable Media, chapter Games, Storytelling, and Breaking the String, pages 5 13. MIT Press (2007).

[3] K. Salen and E. Zimmerman. Rules of Play: Game Design Fundamentals. The MIT Press (2003).

[4] C. Crawford. Chris Crawford on Interactive Storytelling. New Riders, 2nd edition (2012).

[5] B. Mallon and B.Webb. Stand up and take your place: identifying narrative elements in narrative adventure and role-play games. Computers in Entertainment (CIE) 3, 1 20 (2005).

[6] J. Howard. Quests: Design, Theory, and History in Games and Narratives. A K Peters, Ltd. (2008).

[7] E. Aarseth. From hunt the wumpus to everquest: Introduction to quest theory. In F. Kishino, Y. Kitamura, H. Kato, and N. Nagata, editors, Entertainment Computing - ICEC 2005, volume 3711 of Lecture Notes in Computer Science, pages 496 -506. Springer Berlin Heidelberg (2005).

[8] M. Trenton, D. Szafron, J. Friesen, and C. Onuczko. Quest patterns for story-based computer games. In Proceedings of the Sixth Arti cial Intelligence and Interactive Digital Entertainment Conference (AIIDE) (2010).

[9] C. Lindley. Developing Interactive Narrative Content: sagas/sagasnet reader, chapter Story and Narrative Structures in Computer Games. High Text Verlag (2005).

[10] A. Sullivan. The Grail Framework: Making Stories Playable on Three Levels in CRPGs. Ph.D. thesis, University of California Santa Cruz (2012).

[11] M. W. Fendt, B. Harrison, S. G. Ware, R. E. Cardona-Rivera, and D. L. Roberts. Achieving the illusion of agency. In D. Oyarzun, F. Peinado, R. Young, A. Elizalde, and G. Méndez, editors, Interactive Storytelling, volume 7648 of Lecture Notes in Computer Science, pages 114 -125. Springer Berlin Heidelberg. (2012).

On Completing a PhD Proposal

In Mid-December, my PhD thesis proposal was accepted, leaving me ABD ("all but dissertation"). It was quite the journey to get there, and I have some hopefully useful insights to share from the experience.


if you've ever wondered... / toby

If you're curious, much of the information about the doctoral proposal for our School is likely widely applicable. We are supposed to finish the proposal within 6 terms of registration, but in reality that's not very common. Many students do their proposal much later in the process, after completing most of the work needed for the thesis. We tried to do my proposal a bit sooner so we could get some solid feedback earlier in the process. I'm really glad we did, but more on that later.

The Document

The first step of the proposal is to write the document. I started working on this during the summer while also developing a prototype game using my interactive storytelling framework, then part-time in the fall when I went back to teaching. However, several iterations were required, the first being on the general structure of the document. This is what we settled on in terms of chapter layout:

• Introduction
• Background
• Project Goals and Design
• Work Completed
• Future Work Plan
• Conclusion

I put most of my effort into the second and third chapters. The background is where you have to not only show that you have a good grasp on what's out there, but also clearly show the gaps you are trying to fill. I used the Project Goals chapter to really spell out what I was trying to accomplish, and describe the storytelling framework we had published at Foundations of Digital Games.

Both of these chapters needed multiple iterations even after I fixed the overall structure of the document. We tried to bring them as close to final thesis-level quality as we could in the time we had, hoping we could reuse a lot of it again later. This is what I spent most of the fall doing.

My process for the background section is interesting to look back on: although I knew from the beginning what the text in that chapter had to accomplish, it still took me three to four iterations to get there. I fixed one problem each iteration. First, I improved the organization of the chapter so each major section had a clear organizing principle. Then I analyzed the literature more critically, then more effectively highlighted the gaps. I reorganized the sections again, and finally added better conclusions to tie up each section. The final product actually ended up being decent!

In some ways I would have liked to spend more time on the document. I noticed a few too many problems when re-reading it a few weeks after giving it to the proposal committee. For example, I cut back my introduction just before submitting, and later I realized just how... bad it was. But, to be honest, the end result — passing the proposal — would not have changed. Sometimes you have to know when to stop, even if that reason is that you simply can't spend any more time on it.

The Oral Examination

Three weeks after submitting the document to the committee, we had our proposal examination. It started with me giving a 15-20 minute talk, followed by two rounds of questions from the committee.

When preparing for the talk, I re-used a technique that helped me with a past conference presentation: I wrote a script to figure out what I wanted to say. After some feedback from my supervisor on that, , I ended up iterating on that, too.  I ended up finishing my slides a little too close to the actual presentation time. I had originally hoped to run through the talk at home at least a few times (once with my husband), as well as take notes about my background section and key project information so I would be really well prepared for questions. I did none of that. Although I still passed, I think those with some more time might really benefit from trying these ideas.

These are the slides from my talk, which give a decent enough idea of how I structured it in the end. As you can see, I centred the organization around the proposed contributions of the thesis.

Because of my lack of planned practice, I ended up being pretty nervous for the talk, and it showed.  Which is a shame, because I'm generally pretty good at presentations and keeping nerves in check.  But I guess it turned out well anyway, since some of the committee understood what I was trying to do better after the presentation.

The question portion of the exam was... interesting.  I feel like I was barely asked any questions at all.  I mostly got feedback about the proposal document (lots of constructive criticism!) and suggestions for where to go from here.  The committee suggested some ways to narrow the scope of our work and suggested ways to avoid having to make an entire game to test our designs.  This was incredibly relieving! Nothing I've done so far has been wasted effort, and future plans now look more achievable somehow.  The plan is to meet back up with the committee in six months to go over an updated plan since things are likely to change enough that we don't yet know what the final thesis will look like.

Based on this outcome, I am very glad we did an "early" proposal.  I highly recommend biting the bullet and proposing your thesis as early as you can.  Your supervisor should know if you've got enough to pass.  Get valuable feedback early and perhaps even save yourself some work, as happened in my case.

How Small Changes in a Game's Story Can Make a Huge Difference

I recently wrote to the author of a book I love, Think Like a Programmer.  I had wanted to ask some questions related to using the book for one of our first year classes. But it turned out that V. Anton Spraul also happens to be interested in stories in games, the topic of my thesis.

Leave me alone, I'm reading. / Hey Christine 

We had some interesting discussion on the topic after he read our Foundations of Digital Games paper on coherent emergent stories. I thought that some of his suggestions of how this sort of system could work were so spot on, I wanted to share them here.  So, with Anton's permission, here is what he said:

So I'm intrigued by your idea, especially by how it could be employed without the player knowing -- using a "tension" value to control the type or volume of music in a scene, for example. What if a game gradually lowered tension over time outside of character interaction or combat, so that a scene would actually play differently just because the player took a long walk before a key confrontation? Cool stuff. Or a game with Bioware-style dialogue interaction where certain choices were not always present, and not because certain information had previously been discovered or not, but because of the emotional state of the player avatar at that moment, as influenced by prior events? Perhaps ultimately, a game could be made which always arrives at the same scene, but the outcome of the scene is largely controlled by prior actions in a way the player wouldn't predict--so maybe the player can only shoot the final boss if he or she is angry enough to do it. (I was let down by the endings of the otherwise brilliant Deus Ex and Deus Ex: Human Revolution because they didn't depend on anything that happened five minutes before the end of the game.)

I like the tension example especially.  We had pictured using different lighting, or camera angles, depending on the path the player took previously.  For example, maybe the ending of the game is the same either way, such as the princess returning to the kingdom after "taking care of" the threat of the nearby dragon.  Simply knowing that the player befriended the dragon and learned its behaviour was a result of it protecting its child would already cause a different interpretation of the culminating scene than if the player had slayed the dragon.  But adjustments in lighting, for example, might emphasize this.  The idea that additional small changes based on tension could reflect the urgency with which the player acted seems all the more interesting!

This kind of approach is meant to allow players to have an effect on a story's outcome without the need to create much (or anything) in the way of additional assets.  I think the potential is enormous.

My FDG 2014 Cruise Ship Conference Experience

You may recall that I was going to a conference on a cruise ship in April.  Well, I'm back from Foundations of Digital Games 2014 and am happy to report that I have found another new favourite conference and community.  The conference went well and I made some wonderful friends. Win win!

It was a strange experience, being on a cruise ship for (mostly) academic purposes.  This was my first time on one, and to be honest, I actually prefer the resort experience more when it comes to vacations.  An overwhelming sense of "fake" was prevalent on the ship, and while resorts aren't necessarily better, on a cruise all you have is the boat.  No beach, no grass, etc.  I also didn't love the dark, cavernous feeling on most decks of the boat or the lengthy process to embark and debark.  Even the mall was kept dark and lit with neon lights most of the time.

But there is a big advantage to hosting a conference on a cruise ship: nobody can leave! This was really great for building community.  It was easy to find other attendees and spend some social time with them.  For example, on one of the early nights, there was a disco party happening in the mall.  At that point I was alone, wandering around, wondering what to do.

When I ran into some friends (old and new), I finally had someone to dance with, even if we were stuck with disco for quite some time.  I would not have danced disco alone, but with them, I had a blast.

I have to admit that the upper deck with the pools was a nice place to prepare for my paper presentations (lab mates, if you are reading this: pretend I prepared weeks in advance and practised at our meetings).  Sitting on a swinging chair looking out on the ocean is a good way to relieve last-minute stress.

And boy, was I stressed.  I wasn't worried about the actual presentation being good, but rather whether the audience of heavy-hitters in the stories-in-games field would think the work itself was any good.  It was a rare moment of feeling the imposter syndrome.  To make matters worse, I had two talks almost in a row! Good to get them over with, but no chance for feedback in between.

Fortunately, everything went very well.  The talk was good, and the questions afterwards were even better.  A lot of the people I was intimidated of in the first place made a point to tell me that my talk was interesting.  Later in the conference I even got to have an extended conversation with one of them, giving me both confidence and ideas.  (Learn more about what I presented if you're interested.)

After my talks and a couple of other interesting paper sessions, I escaped on my own for a bit to decompress.  The sun was starting to set, which was the perfect time to take a stroll around the boat.

The next day, the ship docked in Cozumel, Mexico, where two of my new friends and I went on a tour of Maya ruins (apparently you aren't supposed to include an "n") and visited a gorgeous beach.  I was really glad to have my talk behind me at that point as I could completely relax and enjoy it!

 

The last day of the cruise included more interesting talks and a lovely reception and dinner to cap it all off.  I left the following morning on a high, and already trying to figure out how to ensure I attended next year's conference.  I left feeling like I had finally found "my people," from my awesome roommate to the researchers with the same interests.  Thanks FDG, and hope to see you again soon!

Interactive Storytelling in Games: Next Steps

Last night I gave a lecture for our undergrad society on some of the more interesting recent developments in interactive storytelling, along with a preview of my own thesis work.  Below are my slides.

Interactive Storytelling in Games: Next Steps from Gail Carmichael 

Cruise Ship Conference

I've never been on a cruise before. Who would have thought that my first opportunity to sail would be for an academic conference on videogames? Come on, admit it. You're jealous.

via Wikimedia

The conference is Foundations of Digital Games, and the photo above shows where we'll be living for about 5 days in April.  This past fall, my supervisor and I worked really hard to get a paper we'd been sitting on into good enough shape to submit, and wrote up a whole new paper on my thesis work.  I was nervous about whether either would get in, but lo and behold, both did!

The paper on my thesis work, A Framework for Coherent Emergent Stories, got in with generally positive reviews, despite the very embarrassing fact that two important diagrams ended up as black boxes.  The one more negative review was actually extremely helpful - we will definitely be improving our write-up with those comments in mind.

The other paper was about non-linear stories in traditional media and games.  It was hard to know how this one would fare since the topic is more closely related to games studies, making me a bit of an outsider.  It was accepted in the work-in-progress track, which I am definitely satisfied with.  Lots of really useful comments in those reviews, too, so while this isn't the more important of the two papers, we should be able to improve it.

I have to admit that these successes are really great news after a recent string of rejections.  My publication luck is finally beginning to pick up!

Slides from 'Coherent Emergent Stories in Video Games' / GHC13

I gave a talk at this year's Grace Hopper on what I've been working on for my thesis project:

Coherent Emergent Stories in Video Games
Crafting satisfying narratives while preserving player freedom is a longstanding challenge for computer games.  Many games use a quest structure, allowing players to experience content nonlinearly.  However, this risks creating disjointed stories when side quests only minimally integrate with the main story.  This talk introduces the problem of nonlinear storytelling in games and discusses our flexible, scene-based story system that reacts dynamically to the player’s actions.

My slides are embedded below and you can learn more on my website.

Working With Google Spreadsheets Data Directly

On Wednesday I talked about assigning numbers to my princess-and-dragon story.  I have all my data in an easily parsable format in a Google Spreadsheet.  I decided to use Python to do my story-related computations because of the fast start-up time and the existence of the Google Data Python Library.

Things started well enough thanks to a decent quick start guide, even though I've only used Python a few times before.  I was able to connect to my spreadsheet without issue and start working with the spreadsheet's feed.

But from there things got... complicated.  The quick start guide didn't cover what I needed to do, and I kept getting referred to the spreadsheets API that made mention of other languages, but not Python.  This isn't the first time I've found documentation by Google to be lacking in some way.  I could probably figure things out eventually (perhaps by looking at the source code), but I didn't really want to spend that much time on getting the data - I would rather spend time processing it!

Luckily, @ArcTanSusan and @sharonw gave me a great suggestion on Twitter: gspread.  It was made for Python specifically, has good documentation, and offers much simpler code for doing what I needed to do.  Sign me up!

Making the switch saved me a bunch of time in the end.  The moral of the story: use gspread to work with Google Spreadsheets in Python, and search for alternatives before just using Google's stuff.

Princesses, Dragons, and Assigning Numbers to Stories

I've been working on a prototype story for my thesis project ("Coherent Emergent Stories").  Although I have a couple of stories on the go (such as the one I made for my GRAND poster), the one I am working with now was loosely inspired by The Paper Bag Princess.  This is an informal representation of some of the nodes:

In addition to these nodes are many more satellite nodes that, at the very least, further the story's themes or develop its characters, but that are optional and can be seen in any order.

My task now is to turn the ideas behind each node's availability (where should the player be? What knowledge should they have? etc) into numbers.  We are currently using a modifier system to calculate suitability of a node at any given time.  This means that I need to represent things like mood and knowledge as numbers.  This is much easier said than done!

I will also be writing some code that will represent formula(s) for computing a scene's score so that we can test what scenes will be available for various game states.  I decided to use Python, since its quick and easy scripting nature is just the thing needed here.  It also seems easy to connect directly to a Google Spreadsheet, where my story data currently resides.

Tomorrow, we plan on spending the morning playing with the prototype and adjusting the data and calculations as needed.  I'm really curious to see how far off the mark my numbers end up being!

Stories and Games at GRAND 2013

Last week, I attended the annual meeting of the GRAND (Graphics, Animation, and New Media) research network, held in Toronto.  Although the research and discussion presented and held at the conference spanned much more, the focus for me was on games and stories in games.

The presenter I was most excited about seeing was Jane McGonigal of Reality is Broken and Superbetter fame.  She believes that gamers are actually practicing some rather useful skills when they play.  For example, they learn to be hopeful and creative, two of several things that we should want people solving the world's greatest problems to be.  I reviewed her book a couple of years ago and still find that it influences my thinking on games.  Although I already knew most of what she talked about at GRAND (having been a fan for a while), I loved seeing her in person, and loved even more that my friends and colleagues now buy into her ideas as well.

A surprise for me was how much I loved Terry O'Reilly's talk.  I admit I'm not much of a CBC follower (unless they're airing an Ottawa Senators hockey game), so I didn't know who Terry was ahead of time.  He spoke about the power of stories, mostly with respect to marketing and advertising.  One of my favorite quotes:

Make people feel your message, not just understand it.  -Terry O'Reilly

Besides being an extremely good talk, it was fascinating how much I connected with his message with respect to games.  In particular, I found myself being convinced by him (and less directly by Jane McGonigal earlier) that stories can truly make a difference in learning with educational games.

On Tuesday night I presented my nicely designed research poster.  I was quite pleased to see a few other really great posters.  My favorite poster (possibly of all time) described Tiffany Inglis's research on pixel art in the form of a comic strip.  Check out the poster on her project page.

Finally, on Wednesday, the last day I was at the conference, I attended the Women in Games panel.  The panel featured Grace from Fat, Ugly, or Slutty, Cecily from Dames Making Games, Anita of Feminist Frequency (which is most recently focusing on tropes vs. women in videogames), and Brenda of Silicon Sisters, a women-lead game studio in Vancouver.  The discussion was fascinating, and I saw a lot of what I do with women in CS shine through, even though involvement in games can be much broader than programming/CS.  It was also really neat to see what Brenda and her company have been working on, since I had been chatting with Brenda about stories in games the previous evening at my poster.

Even though I had less than two days between trips (poor baby Molly!), and even though I could only stay for two days, I'm really glad I ended up coming to GRAND.  I feel energized as I move into my attempt to get a lot done research-wise this summer...

My Beautiful GRAND Conference Poster

I'm a strong believer in creating conference posters that look good.  If they have a striking resemblance to printed papers, in my opinion something has gone really wrong.  With that said, I have to say I had a lot of fun designing my most recent poster.

Although this image is slightly out of date from the final print version, it gives a good idea of what I was going for.  You can get the gist of the research by looking at it, but it does not contain all the information a paper would.  That's what the poster presentation itself it for: I will have the opportunity to discuss the work more deeply.

There is no reason whatsoever that a poster can't be both beautiful and functional, so I encourage you to see what you can come up with the next time you create a poster! If you have an example to something you're particularly proud of, I'd love to see a link in the comments.

Edit: You can now check out the high resolution PDF of the poster if you'd like.

Story Graph Example Inspired by The Paper Bag Princess

Without giving too much away, I wanted to share an early example of the types of story graphs I'm working on for my thesis.  This was my first example, and it's inspired by Robert Munsch's Paper Bag Princess.

Without even telling you what all the colours and notations mean, you can probably gather that there are some nodes that must be seen in a particular order, and others that can be seen any time.

Eventually my examples will have much more of the latter type.  The end goal is to be able to decide which nodes should be available in the first place according to the player's current state in the game, and then to dynamically modify the scenes in small, simple ways to ensure they make sense and connect back to previously seen nodes.

Hopefully, this will result in more open and coherent emergent stories in games.

Coherent Emergent Stories

My thesis project is moving forward, which I'm thrilled about. Here's a high-level abstract that summarizes the system I am working on.

Crafting satisfying narratives while preserving player freedom of action is a longstanding challenge for computer games. Many games use a quest structure, allowing players to experience content nonlinearly. However, this risks creating disjointed stories when side quests only minimally integrate with the main story. We propose a flexible, scene-based story system that reacts dynamically to the player’s actions.

In the proposed system, stories are defined within a graph where nodes represent scenes and edges represent causality. Nodes are tagged with information including possible locations for the scene, the plans or goals connected to the scene, and the agents and objects involved in the scene. At any time, the distance from the player’s current game state to nodes in the story graph is measured according to five dimensions of nonlinearity: time, space, causality, agents involved, and the player’s goal. The system will use the distance to determine what nodes should be available at any given time. Scenes will be modified dynamically according to when and where they ultimately take place, ensuring that each node has a narrative connection to its predecessors. This system allows for potentially connected stories driven by player action, leading to a more cohesive emergent story.

Interactive Storytelling: Slides from CUSEC 2013

Here are my slides from yesterday's talk at CUSEC 2013.  There are normally embedded videos in the slides; this version links to the videos online instead.

Interactive Storytelling (CUSEC 2013) from Gail Carmichael

Procedural Rhetoric and Theme in The Walking Dead

I've been thinking about procedural rhetoric and its relationship to a particular theme in a story. For example, a major theme in The Walking Dead is humanity: how do you maintain your humanity in such trying times, and how far are you willing to allow it to slip in order to survive?

How do the game’s mechanics support this theme? Most of the mechanics are superficial in their relationship to the story (i.e., classic point-and-click adventure style of interaction). But then there are the story choices. Many are dialog choices which don’t end up having much real impact in terms of story results, but some are black and white choices like whom to save.

These latter types of choices seem to fit more with the idea of procedural rhetoric. You are faced with a decision you have to make quickly and on the spot. Neither choice is good, and sometimes the outcome is essentially predetermined anyway in the sense that the zombies will kill no matter what. This reflects how it might feel in real life as well, when the adrenaline is pumping and you have only moments to react. Sometimes your actions feel futile, and often somebody has to die.

But I think these are really the only mechanics you could argue represent procedural rhetoric relevant to the humanity theme.

I want to figure out if there's a way to represent the theme more deeply in the game’s rules/mechanics.

SPOILER ALERT...

For example, one episode involves a farming family who we eventually learn are cannibals. They had decided to survive by eating flesh of people who were injured and going to die anyway, though they eventually rationalize eating even hardly injured people. How could the game reverse the roles so the player ends up doing something like that and actually feeling like it was the right choice despite the moral dilemma?

...END SPOILER

I want it to be more than just the story, even if the story can branch for these kinds of choices; I want the rules of the game to represent the theme better. I am not convinced the current setup of Walking Dead would be the best for this, given the style of mechanics, so what might?

Research Article About Me: Creating Compelling Computer Games

There's a nice story about my research on the Carleton website.  I'm particularly happy to see it because it hasn't been easy to figure out my exact thesis project, and it feels good to have finally found my path.

A team of Carleton researchers is trying to find out why so many computer games shy away from using nonlinear storytelling techniques – that is, techniques that help present stories out of chronological order. Traditional media like films and novels use all kinds of interesting nonlinear techniques, like those found in Run Lola Run, Groundhog Day and Memento. Many games tend to stick to fairly simple techniques like flashbacks, but more sophisticated approaches could result in more games with critically acclaimed stories.

Read the rest here!

Resources for Testing Game Story Ideas

I will soon be trying out simple nonlinear narrative ideas for my thesis research, and wanted to find (freely available) tools that will make doing so much easier.  Here are a few, some of which are great for beginners and non-programmers.  (Know of any other great tools? Let me know in the comments!)

inklewriter

 
If you want to create branching stories quickly and easily, this free graphical tool might just be for you.  It's not likely to be useful for me without access to its source code, but non-programmers should have some fun with it.

Ren'Py

This is how Ren'Py describes itself:

Ren'Py is a visual novel engine that helps you use words, images, and sounds to tell stories with the computer. These can be both visual novels and life simulation games. The easy to learn script language allows you to efficiently write large visual novels, while its Python scripting is enough for complex simulation games.

I like that beginners can use its simple scripting language to create interesting stories, and that I will be able to use Python to make more sophisticated games.  This is probably the first tool I'm going to download and try since it should be really fast to get going on the simplest of ideas.

Fabula

As per the Fabula website:

Fabula is an Open Source Python Game Engine suitable for adventure, role-playing and strategy games and digital interactive storytelling.  Fabula can be used as a library to develop your own games. As an alternative, you can use the Pygame-based graphical editor and the default game engine that come with fabula.

With this tool, you get into programmer's territory, but still get a lot for free.  I like that the engine was designed with storytelling in mind, and that like Ren'Py it uses Python, which should make it fairly easy to work with. I will likely give this one a try early on as well.

Orx

According to the Orx project site:

Orx is an open source, portable, lightweight, plugin-based, data-driven and extremely easy to use 2D-oriented game engine. It has been created to allow fast creation of games and prototypes.

The reason I include this project is because I think 2D is the way to go to test story ideas, and such an engine would make creating a 2D game much easier.  If the other tools aren't flexible enough to customize things as much as desired, then this might be the way to go (assuming you understand C/C++ programming).  I probably won't use this one right away since I don't need the power, but I have a feeling it could come in handy later on.