Phrontisterion 7

June 23-24, 2012

After a seven-year hiatus, Phrontisterion returned for its seventh life. The weekend was marred by atrocious weather: it rained and was cold, so we spent most of the time inside. Fortunately, the group was small enough to fit comfortably in the main room in my house. As far as I could tell, this was the only June 23rd in local history on which it rained, and it was certainly one of the coldest. Next year Phrontisterion will be in July.

Rather than write most of the report myself, I asked attendees to write up their own reports. I present them here in the order that they reached my mailbox.

Rod Humble
The discussion across both days was very wide ranging and all very enjoyable. It ran the range of outrage to laughter. A splendid time.
A few themes and conversations came away with me.
The input problems:
Many times when discussing interactive storytelling we quickly gloss over how we are to interact with the story. After all heroes in stories are often unusual people and we cannot emulate their skills.
A good example of this occurred within LA Noire, a game I enjoyed but in which the hero is a detective who is good at detecting peoples lies by reading their faces. The facial technology and acting is first rate but the problem is – I am not good at reading peoples faces. This meant that for me the game ranged from the comic (obvious twitches for lies) or random guessing. The subtle hunting for clues on a perps faces was beyond my skill. This problem crops up when discussing the most exciting ideas for an interactive story.
Could you really play a character in something as witty as an Oscar Wilde play? I could only play the role of a dullard, the butt of jokes perhaps, I am not witty or as quick as characters in such a performance. Extend this thinking to an extreme such as a fantastical invention like a holodeck and the problem is writ large, the issue becomes not one of the machines ability to quickly create and support a story, it becomes one of YOU being unable to interact quickly enough or of high enough quality.
Of course the input problem works both ways. Right now when playing a console game for example you can only input into the world by crude spatial movements or pressing a few buttons. Imagine wooing someone by slamming around awkwardly in a room and occasionally barking out canned phrases, or best case very slowly answering simple questions like some strange dullard. That is the situation of how we communicate to the story worlds we currently play.
Really this is just an extension of Chris’s thinking around discrete and continuous space impacting dramatic impact, it just applies to all player actions not just movement.
In other words stories miss out the dull bits of a characters life for a reason and they make characters unusual or interesting for a reason. One solution maybe to simply create new kinds of stories which do include the dull bits, which may become interesting. But if we aspire to ape even the dullest penny dreadful then it seems to me the direction of players choices towards the interesting and the abstraction of ability while maximizing freedom are unresolved problems as yet.
A thorny issue I think and not one easily solved, even of you have perfect speech and physical input into a story world, in fact they make it even worse.
The need for different perspectives:
Even of you somehow manage to overcome the skill issues of the player you now have to contend with player behaviour which breaks stories. An example was illuminated when Emily Short summarized The Bacchae for us. Which player in an interactive story would willingly play along to make such a tragedy? By their nature the characters in Greek tragedy are not acting in a manner that any interactive player would. For a start who would willingly get their character killed to make the story work if the story ended for them so they didn’t get to play through till the end?
Revolving perspectives or the player as director such as in Storyteller and Prom Week are interesting explorations towards solutions.
The fall and rise of text:
It became clear to me (I don’t think I was in the majority, but its my report so hey) that text based interactive stories are no barrier to success, commercial or artistic. The market for text based products is growing. Indeed if you buy into the notion that ebooks are changing the way we read for all future generations then text based interactive storytelling is not just hopeful, it could be a genuine new mass market art form.
That’s the good news.
The bad news is making text based interactive stories is much harder than using graphics, in particular 3D. At least if you fail in 3D you get some points for a pretty world or a “ooooh” if you can make the damsel walk across a room realistically and say something non moronic. You get no points for that in a pure text product, your story is naked, with nothing to show but the quality of the writing and interactivity.
Good luck with that.
My mindless optimism:
I came away from the whole event very upbeat and confident that some form of text based interactive storytelling will be hugely successful and it feels closer now than it did in the past.
A note of thanks to the Crawfords and all the participants.  A very engaging two days!

Alexis Kennedy
For me, most people with specific approaches fell pretty clearly along a procedural <--> scripted continuum (eg Storytron <--> Prom Week <--> Cotillion <--> StoryNexus <--> Choice of Games). Anything too procedural falls into the Turing tar-pit and anything too scripted is trivial or brittle, but those are extreme cases. When someone expressed an opinion of preference ('more choice is better', 'hand-building a story is probably cheaper') it was very often a statement of position about where on that continuum the author was comfortable. There is enough common ground along the continuum that we can share practice, but attempts to collapse the continuum to one best-practice point seem futile; I think this is the single most useful taxonomic dimension.

For me, the conversation kept coming back to the intentional gap - every choice about what to put in is complemented by a choice about what to leave out. I think the gap is even more important at the procedural end. Chris said that Storytron never should have been so general; Emily chose a very specific social context for Cotillion; the Sims relies on player interpretation. There was some conversation about how the authorial voice manifests in a procedural narrative, if the player can shape the story substantially, and I think that voice is most strongly present where the author has forbidden the story to go - in the gaps.

I am using 'gap' to describe at least two things here - the spaces in a story that encourage engagement by requiring the player to insert their own imagination, and the absence of implementation of choice or supported mechanic. But I think they have something in common.

I was pleased to hear so much positive comment about text, and to read Rod's summary. I do want to riff on this:

"The bad news is making text based interactive stories is much harder than using graphics, in particular 3D."

This is all true, and I'm not going to disagree - pictures snag attention - but I am going to elaborate. Good writing *does* count, and also text has

- infinite SFX. You can say, as the leviathan rises from the sea, you see cities caught between its teeth. Iain M Banks 'the writer has access to the greatest special effects budget imaginable: the human imagination.'

- personalisability. The second person singular is as powerful a means of promoting identification as anything short of a first person viewpoint, and it's easy to use cheap powerful tricks to customise text to make it about the user.

- versatility. 'It is my seven hundredth day at Mather & Krake, and the bones of my legs have begun to fuse with the legs of my chair.' OK, I'm going to tell a twenty-year-arc story about a corporate enforcer confined to her desk, interacting with the world through letters, visitors and memories. You *could* do this with a graphical interface, of course you could - but you can prototype and implement unusual, exotic, experimental stories very quickly. 'Unfilmable novel' sounds much more likely than 'unnovellable film'.

Chris Crawford
1. The single most significant event, from my perspective, was the atrocious and unprecedented weather. Late June in Jacksonville is always sunny and almost always warm. But we had overcast skies, cold, and rain! I couldn't believe it. Fortunately, we were able to fit comfortably inside the house. In my defense, a much larger event that same weekend, the National Parasailing Fly-In, was caught even more flat-footed than I was.

2. The second most salient observation concerns the near-wasted Sunday afternoon, with my attempts to pose the question "How can we best proceed in the future?" triggering answers to the completely different question "What is the best, highest, and worthiest form of interactive storytelling?" My own flustered responses to some of the comments contributed to the confusion. The conclusion we came to was pretty much a shrugging of the shoulders and the statement, "Anything goes".

3. Next in salience came the fascinating discussion on Saturday initiated by the question, "Can a computer calculate love?" This question didn't really hit the core issue, which we developed during the course of the discussion. The main question broke down into two secondary questions. The first of these was, "Is it possible to create computer algorithms that calculate love?"  I found this question easy to answer in two different ways. First, there's the matter of approximation. We don't need an algorithm that captures that complete essence of love; we need only produce something that yields an approximation close enough to satisfy the needs of the player. Second, and more important, we don't need to worry about a "hidden variable" such as "love"; we need only produce behavior that the player interprets as representative of love. In other words, if two characters hug, kiss, laugh together, hold hands, and so on, in the right proportions and the right sequence and timing, we really don't care whether there's actually an internal value that we call "love". It's the behavior, not the label, that matters.

The second secondary question was much more interesting: are the spiritual and emotional aspects of the human condition necessarily unreachable through rigorous analytical techniques? This again split into two parts: a) can it be done; and b) even if it could be done, SHOULD it be done? Rod Humble argued that the application of logical, mathematical thinking to artistic matters was objectionable; he used the striking phrase "bringing math to an art fight". I myself believe that there is nothing demeaning about the application of math to artistic matters. After all, we eviscerate cattle, dry their intestines, and then use those dried intestines to make catgut, from which we make strings for violins, cellos, and other musical instruments. Is Vivaldi's work dehumanized because it utilizes dead cow intestines? I think not.

As to the question of actually pulling this off, I see no problem. We'll never attain high fidelity, but we can get close enough to communicate powerful ideas with this medium.

4. We spoke briefly about nomenclature. I urged that we should embrace terminology taken from drama and literature in preference to terms coming from computer science. The group rejected my overall position but accepted "character" in preference to "actor", "agent", and "NPC"; and "storyworld" to describe the thingamajig that people actually play with when they're playing interactive storytelling. But my specific suggestions regarding "Prop" instead of "Object" and "Thing", and "Stage" instead of "Position" and "Location" were rejected. There was also an agreement that we need some special way of differentiating the human player from the algorithmically-driven characters. This caused some people to hold out for "NPC" (a term I revile). The use of "protagonist" didn't generate any excitement.

5. One issue seemed to underlie much of the discussion without becoming the explicit focus: can we define or even characterize interactive storytelling? The general attitude seemed to be that the state of the art is still too undeveloped to permit us to reliably narrow our thinking; we must instead pursue all possible avenues of discovery before we can say what interactive storytelling really is. At one point I quite undiplomatically let slip a comment along the lines of "Well, maybe YOU don't know enough about interactive storytelling to attempt to characterize it…"

6. At one point, I introduced a rather convoluted line of thinking involving a series of universes in each of which God planned out the life of each person according to one of the common schemes for interactive storytelling: narrow branching tree, choose-your-own-adventure, bushy tree, and procedurally generated. I then challenged the others to contemplate the degree of free will afforded an individual in such universes, my point being that the greatest free will was offered by the procedurally-generated universe, which in turn suggested that the greatest or best interactivity would be supplied by procedurally-generated interactive storytelling. One of the participants flatly rejected the notion that free will corresponds to agency.

7. We got ourselves into another silly kerfuffle over the concept of the Holodeck. I had posed the question, does the Holodeck offer us a Holy Grail? Here we got confused over the distinction between the Holodeck as a sensory experience and the Holodeck as a near-perfect drama simulator. A number of people rejected the idea of the Holodeck being a Holy Grail, but when we disentangled the sensory side of the Holodeck from the dramatic simulation side, there was convergence toward the idea that a Holodeck as drama simulator was desirable. Nevertheless, a number of people refused to endorse the notion that the Holodeck can be regarded as a Holy Grail, because they felt that there were lots of other undiscovered directions in which interactive storytelling might profitably go.

8. There was disagreement over the role of the authorial voice in interactive storytelling. Several participants were quite seriously opposed to procedural generation, on the grounds that it robbed the medium of an authorial voice. I was unable to convince them that the authorial voice in procedurally generated IS is expressed at a higher level of abstraction.

Kevin Gliner
My main takeaway from last weekend was that it’s highly likely there are multiple viable solutions for interactive storytelling. 
I base this on the simple observation that if you create a category called “interactive storytelling”, there must be a category of everything else called “non-interactive storytelling” which includes such wide ranging solutions as “novels”, “music”, “movies”, “comics”, “poetry” and so forth.  While it doesn’t logically follow that “interactive storytelling” must also have multiple answers, it does seem more than probable given the really broad macro level categorization (interactive vs. non-interactive, neither of which is actually descriptive of what someone might experience).  Whether those interactive solutions all look like mirrors from the other side (e.g. “interactive novels”) or are sliced a completely different way I don’t know.
Another way of saying this:  interactive storytelling and non-interactive storytelling are comparable categories.  Interactive storytelling and novels are not, because novels are a specific type of non-interactive storytelling.  Similarly, interactive storytelling is not comparable to literature because literature is a specific qualitative categorization of non-interactive storytelling.  Given the diverse range of successful approaches to non-interactive storytelling, there are probably an equally diverse range of successful approaches to interactive storytelling (never mind qualities, if something like a literature equivalent is your concern).
It was a great weekend, well worth the travel to the remote destination.  Look forward to seeing everyone at future gatherings and at more traditional conferences throughout the year.

Joseph Limbaugh
The things from the conference that made my brain whir the most:
1. The concept of a computer Gamemaster or Story Director.  This was something that I had encountered previously in the Storytron forums and in Chris' book on I.S. We had several moments where the discussion touched on this with people concerned whether there was an auteur at all in an interactive story and how great it would be to recreate the pen-and-paper rpg experience with a computer GM.  There was very little discussion about HOW such a piece of software might be created, or the algorithms that it might use to spin narrative.  Some had the opinion that this was an impossible or unnecessary goal.  Most of plans for I.S. so far seem to involve putting some interesting characters together with some dramatic verbs and letting the stew bubble.  My (gut) feeling is that creating an I.S. Director is a pretty important problem to solve, and that there is a complicated algorithm that determines what comes next in a story, but I'm still not totally certain where you would begin.
2. There are a lot of different people creating a lot of new interactive technologies and most of them aren't working for the big companies.  When I.S. emerges, it will probably be from some unknown coder working in his or her garage, in fact there will probably be several different creators, each approaching the problem from a different angle.   Long live indie game development!
3. Laura recounted her experiences with Storytron, and this resonated with me since I was also one of the people working on a storyworld using the technology, nothing as grand as Vanity and Vexation, but I did very much enjoy the authoring tool.  I have to disagree with Chris here and say that I never thought the tool was too complicated to use.  It was elegant, well-designed, and chock full of potential, but not in my opinion too complicated.  I think the thing that was (and is) complicated was trying to wrap your head around just how a storyworld was supposed to work.  It was like having all the parts of a B-57 and the tools to build it, but no blueprint. I honestly believe if we had had more time and maybe more people working on the problem, we would have come up with some amazing things.  Maybe not a holodeck, but I think it would have been something incredibly engaging and unlike anything else currently available.  If Chris ever does put up a Kickstarter page (and he should) for Le Morte d'Arthur I hope he gets a ridiculous amount of cash, so much so that he has extra money to bring back Storytron.
Other random thoughts:
- Some one should start a game company called "Grindy Walking" or "Skinner Box."
- Hardcore Farmville.
- Don't bring Math to an Art fight (or the opposite)
- Choice is good, especially if the choice is a dilemma
- Food 4 Less is rather disturbing inside

Emily Short
Phrontisterion refocused my attention on the importance of stepping back from one's tools periodically and thinking about them from the perspective of how they relate to the structural framing of interactive stories. The discussions about what people wished to do with Storytron (especially Laura's thoughts on Vanity and Vexation), and war stories about first contact with Storybricks and some of the other languages and tools we talked about, reminded me that interactive storytelling tools tend to be documented and presented in terms of features one level (at least) lower than the introductory user typically will understand.

That means a lot of eager creators have trouble breaking down the concept they have into terms that the engine is able to accept. People spend a lot of time trying to create something in an engine that isn't ideal for that kind of product (for instance, the number of people who try to write combat-based RPGs in Inform, only to give up sooner or later because it just is *not* really designed for that purpose). Or they spend a lot of effort, months or years of implementation, before realizing that they made a fundamental design error early on. Or they get overwhelmed by the complexity of making all the details to support their grand vision, and they never get further than the first scene. These are things that we see people in the IF community wipe out on all the time.

If we were better at building "magic crayon" storytelling tools, tools where you got a result that felt like it was actually going somewhere from the very beginning (even if the story/storyworld then needed a lot of refinement), that would be an incredibly helpful and energizing thing for many would-be authors who otherwise quit before they release any completed projects.

There are a couple of components to achieving this, I think. One is faster feedback, being able to see what you're doing: see Bret Victor's stuff on tools that more immediately show you what you're doing. ( Some interactive storytelling tools are better than others at letting the author quickly see the results of new work.

Another component is a system that lends itself to sketching before you have to lay on any paint -- creating a prototype that is functional enough to demonstrate the strengths and weaknesses of the structure you've designed, and than can then be built out more fully.

I think this is something a lot of skilled authors have already learned to do; at least in the IF community, many authors have talked about methods they use to prototype important system functions or build from a desired transcript or in some respect create a scaffold before filling it out. But this is not necessarily something that the tools themselves help to support or encourage. It's especially important, and especially lacking, on the level of plot structure. IF languages and the more recent group of choice-based tools generally don't go very far with helping an author visualize a structure or implement a scaffolding rapidly that then gets filled in.

If we can achieve a tool that lets people articulate a rough vision in more or less the terms they conceived of it ("I want a romance about a young woman who isn't wealthy and has to choose between several suitors...") -- and then refine, elaborate, or override aspects of that vision in order to get it closer to what they really want -- that might go a long way to closing the coder/writer divide.

The Attendees of Phrontisterion 7 from left to right:
Kevin Gliner, Stephanie Spong, Rod Humble, Laura Mixon, Dan Fabulich, Veronica Zammitto, Alexis Kennedy, Joseph Limbaugh, Emily Short, Lyndon Goodacre, Chris Crawford