April 12, 2004

Electronic Writing Jam

by Andrew Stern · , 12:26 pm

GDC has inspired one lengthy discussion on GTxA so far, perhaps we can get another one going. As I mentioned earlier, the presentation of this year’s Indie Game Jam was a highlight of the conference for me and others (Robin’s writeup, more 1 2 3 4). I love the idea of a group of experienced practitioners getting together in a single location for a few days (pics 1 2 3 4), set loose with a shared, novel, expressive development platform, to quickly jam out an array of little interactive works, ending with an group exhibition shortly thereafter.

This got some of us thinking about doing our own version of a Game Jam. I recall Michael was immediately thinking about the idea of doing a similar event at his new Experimental Game Lab.

Wandering around SOMA a few hours before I flew home to Boston, Noah and I had a bit of time to think about the concept of a Jam. We thought, what would a Jam of electronic writers look like?

Following the model of the Indie Game Jam (IGJ), there could potentially be a series of Jams in which the authoring platform changes each year. The chosen platform for a Jam obviously greatly influences the nature of the works created in it. IGJ Year 0‘s platform was “100,000 guys”; Year 1 was “Shadow Garden”; this year was “2D physics”. What could be a novel, expressive platform for a Jam of electronic writers?

Of course what immediately came to mind was the Facade platform. Now, the full platform for creating story beats, character behaviors, natural language rules, character and object models and animations is big and cumbersome — way too large for use by newcomers in a short workshop. But if we took the time upfront to fashion a subset of the full platform, which would include an existing library of primitive bodies and basic behaviors (walking, speaking with text balloons, gesturing, emoting), a basic set of NLU rules (for general understanding of agreement, disagreement, refer-to topic, praise, criticize, flirt, etc.), a small contained 3D world (a room with interesting props), and some example code of how to create a new character with a collection of simple but interesting behavior and dialog — what sort of Jam could that be?

Imagine a Character Jam, where a group of electronic writers got together for a few days, each authoring their own simple (shallow), reactive, language-capable animated character, designed to be set loose to freely interact with all the other characters and a single human player character.

To give this a chance to be at all cohesive, upfront we’d probably want to decide on a scenario that supports a collection of disparate characters forced to interact, such as a group of estranged family members reuniting at a funeral, or a set of strangers trapped in an elevator, or a hospital waiting room. Perhaps the scenario would have two or three plot points, such as “the elevator starts moving again then stops”, or “the doctor comes in to tell a character their loved one has died”.

We’d have to come up with a common set of behaviors and reactions all characters must have, such as the ability to answer common questions, to respond to a common set of basic topics, to give and receive provocations, criticisms, flirtations, etc., to react to the common plot points, and so on. We’d have a common framework for how characters interact, dialog with one another and share the stage — e.g., only one character can be speaking at a time, with a policy of how a character gets to “own the floor” (probably using joint behaviors). But once this basic set of competency is covered, authors are free to create custom, quirky behaviors for their characters, that other characters, with no a priori knowledge of what those behaviors are, will attempt to interpret and react to. This will certainly lead to absurdist moments, but also hopefully some entertaining emergent narrative.

Like the IGJ, there could also be a musician participant to score the experience, perhaps creating a few character-specific refrains that get sequenced in real-time as the characters interact.

One question is, how possible will it be for participants to actually create something in such a short period of time, using an unfamiliar platform? Perhaps we should allow participants to get familiar with the platform before the Jam starts, and think about a character design, allowing them to immediately begin authoring at the Jam itself. We’d have to supply a full set of animations, low-level behaviors, etc. so that authors only have to do high-level writing and direction. Like the IGJ, Character Jam participants would need to be experienced practitioners and fast workers.

Another question is, like the game research thread discusses, how would we fund this? To prepare the Facade platform in this way, even though it would certainly re-use many existing low-level behaviors and NLU rules, would take 1-2 person-months of work mostly by me and/or Michael. And we need to host the Jam somewhere, requiring a PC per participant (maybe it should be BYO-Laptop), and the hosting of a little public workshop / exhibition at the end.

If the Character Jam worked out, a really nice by-product would be that the platform could be made available to others for their own Jams, and for interactive character research in general. Hey, perhaps the latter would be the avenue to get funding…

17 Responses to “Electronic Writing Jam”

  1. Jill Says:

    Sounds like a brilliant idea. The other week when I was doing my Nordic funding blitz (well, it felt like that, I only sent in one funding application actually) I came across this site which certainly makes it look as though there’s heaps of funding if you write a few grants. Though obviously I don’t really know a thing about American arts funding.


    Sounds fun. Like those demo parties. The Gathering just finished here in Norway, and no, I wasn’t there. 5000 people were, though.

  2. michael Says:

    I’ve always had the intention of publicly releasing the ABL compiler, and eventually perhaps other interactive drama architectural pieces (the Facade NLU, the beat-based drama manager, an adversarial search-based drama manager and so forth). I’ve had several students working the last 8 months or so to connect ABL to the Unreal Tournament engine, and to author simple character examples in ABL. This Summer a couple of students are going to make a more elaborate (but of course not remotely as elaborate as Facade) story experience in the ABL/UT infrastructure. Potentially this code base (since it will be publicly released anyway) could serve as a basis for a Character Jam.

    One problematic aspect of the Character Jam as it’s propsed is the independent authoring of characters that then somehow coordinate together to make a story experience. If one forgets about using the beat manager and having a progression to the story, perhaps this would work – it would be like having a roomfull of chatterbots talk to each other and to the player (kind of Sims-like experience). But as soon as you want progression, then you need the beat manager to coordinate the activities of multiple characters into story units – but if the characters are being authored independently of each other, how can beats do this coordination? How would the beats “know” about the details of the characters and the dramatic potential of coordinating different characters in different situations? This works in Facade because Grace and Trip were authored to work together, with a specific collection of story beats in mind.

  3. michael Says:

    Regarding the idea of art funding, there are many art grants of the order of a few thousand dollars here, few thousand dollars there variety. But to do this kind of work, you really need NSF levels of funding, on the order of a hundred thousand here, a hundred thousand there (before overhead) variety. Well, that’s to fund the longer term work… OK, to literally just pay for someone like Andrew’s or my time for a couple of full time months doesn’t take an NSF grant, but it does take more than your typical art grant.

  4. andrew Says:

    the independent authoring of characters that then somehow coordinate together to make a story experience

    I’m imagining not using the beat manager, and in fact not thinking of it as a story; instead, focusing on character, thinking of it as almost-plotless experimental theater. It’s a bunch of quirky characters essentially bouncing off one another in a tense situation. But, we can come up with ways to give it potential for emergent narrative, to be more than a bunch of (state-less) chatterbots talking to each other:

    – Characters are designed to reach out and provoke others — to ask pointed questions, make bold statements, to flirt, cajole, rant, etc.

    – Each character, as the author wishes, can be designed to have their own internal progression, progressed in some fashion by the stimulus from other characters. Each character could have its own simple internal arc(s), which get traversed in different ways depending on who they bounce off of in any run of the experience. If each character is a bit of a self-absorbed drama queen in their own way, it should be interesting.

    – Upfront we all decide upon 2-3 possible “plot points” that each character will need to support.

    Things like that could provide enough loose structure to make sure the experience has a feeling of progression. It would probably be a chaotic, sometimes incoherent and absurd but inevitably amusing thing to watch and play. Again, the focus is on character.

  5. Sean Barrett Says:

    About half of the IGJ participants used laptops instead of the supplied machines, so it’s probably not outrageous to go that route, although you may possibly be limiting your audience somewhat. One advantage to using a standard provided PC is you can make sure all the software is set up properly in advance, but the laptop users have never been a problem for us in practice.

  6. andrew Says:

    By the way, this discussion should be not just about this particular incarnation of a Character Jam, but also new ideas for platforms, formats, etc. for an Electronic Writing Jam or Jams… If anyone out there has ideas, please suggest!

  7. noah Says:

    Part of what I like about the idea of a Character Jam as a type of Electronic Writing Jam is that it embraces both the writing of behaviors and the writing of text. (By the second of these I mean text experienced by the reader/player.) I’d like to see all EWJs (in this imagined series) integrate procedural and textual writing practices. A jam in which the participants just produced text for a given tool or situation would be much less exciting to me.

    As for the lack of a beat manager in the CJ, I do understand that we want to give people a certain amount of room for decoupled development. At the same time, I would like to find some clever way for joint behaviors to still work — and not just by having both characters written by the same author(ing group). Perhaps one writes the joint behaviors so that the other character only has to respond using things from the pre-specified generic repertoire. Or maybe there’s some way of saying, “My response to seeing the casket is to faint with a priority 10. Catching me is a priority 7 for anyone close to me. If you’re not doing anything higher-priority, you catch me.” Hmmm… I know there are others out there who have studied these sorts of systems more than I have. Michael, Andrew, Athomas — are there solutions to this sort of problem from systems you’ve read about (or created) that seem appealing for use in this situation?

  8. nick Says:

    Sorry to hop in so late here.

    I think The Unknown was created in an electronic writing jam of one sort (at least if we believe its origin myth.)

    I have to admint that interactive fiction is the first thing that comes to mind for me, so it may be that everyone would like an e-writing jam in their own preferred form. That have been some e-writing jams in that form, such as the one that produced the curious and wacky Pick Up the Phone Booth and Aisle. I’d certainly be willing to work in another form, however.

    I like the character jam, which builds on the sorts of work that Janet Murray has had students do in class, building chatterbots. Collaborations could really improve the outcome there, though. With accessible bots as an end product, and people worthing together to fill out dialog and round out personalities, it could be particularly fun and interesting, even if the bots didn’t all sit in the same virtual environment at the end, even if they were only represented in text.

  9. andrew Says:

    There have been very successful collaborative e-writing platforms for a long time, of course — bboards, MUDs and MOOs, IRC, blogs.

    One of the primary features of the Indie Game Jam that appealed to me was the novelty of the platform they chose each time — putting 100,000 guys on screen, the full-body interface of Shadow Garden, the fully-procedural 2D physics engine. (I realize that for most readers, without having seen these experiments, my mere descriptions of them aren’t as inspiring.) I’d like to imagine e-writing forms that extend beyond what we’ve already seen — to go beyond state-of-the-art chatterbots, IF, etc., or at least use them in new interesting ways.

    A reason I suggested a 3D reactive language-capable character platform was, not just because it’s a form I’m used to, but because such a platform has never been used in the way I described.

    Had the collaborative network novel The Unknown not yet been made yet, that form would have been a great new idea for an e-writing Jam. Now it’s old hat. (Just kidding. I’m sure there’s plenty of uncharted territory there, of course, that I’d love to explore.)

    In the design space of text-based IF, how well do you think it lends itself to a collaborative Jam, or how could it be extended / used in new ways? Can you point us to anymore past examples of collaborative IF?

    I notice I’m pushing for a collaborative Jam, which doesn’t have to be the case — the Game Jam created a series of small independent pieces. But there’s something very appealing about having the writing of the participants co-mingle and intersect.

  10. andrew Says:

    are there solutions to this sort of problem from systems you’ve read about (or created) that seem appealing for use in this situation?

    Noah, I think in fact we’d find joint behaviors useful to do any coordination at all between characters in a Character Jam, whether it’s supporting a pre-defined generic repertoire of behaviors, or for specific behaviors that we devise at the Jam itself. Michael wrote the joint behaviors technology with multi-character teams in mind (although so far we haven’t fully exercised this capability beyond just teams of two, Grace and Trip). To keep things sane, we’d probably want not much more than a Breakfast Club-sized group of characters.

    Perhaps the Jam could include a process where authors can suggest joint behavior ideas to the authoring group, that they will want the other characters to support; perhaps each participant is allowed to suggest two of their own devising, that everyone must support, and can suggest more that other authors have the option of supporting if they have time.

    The truth is, we’re going to have to keep these behaviors pretty simple, to allow us to pull of a Jam in 3-4 days. I’m imagining the primary activity of each character is saying provocative stuff in a theatrical way — yelling in someone else’s face, or expositing to the group, or mumbling to self (and audience) in the corner or whatever, and having the other characters react in character. This is more sophisticated than chatterbots because characters would be more motivated, have more internal state, and of course, be animated with gesture and facial expression. The primary progression of the experience would occur as each character internally progresses, due to the results of the interactions with the other characters. (The other characters include the player.)

    I’m interested to hear from e-writers (the GTxA folk, anyone else out there), what do you like or don’t like about the concept? For those accustomed to writing e-lit, does writing e-theater turn you on or off? Does the process described sound too cumbersome?

    On the difficulty issue, to make it more feasible, it could be that I and/or Michael (assuming we are among the authors, which I think we’d want to be) would need to do most of our authoring before the Jam, in order to be facilitators / tech support during the Jam. Also, writers who feel less comfortable with minimal programming could bring along a programming buddy…

  11. scott Says:

    I’m down with the jam, and would actually probably rather do something less Unknown-like. It would be dynamite if we could do something where we could mix the characters in the end / have them interact with each other in some way. Anyway, given that the timing is right, I’d be glad to make it up to Boston or wherever for a few days of writing.

    My students in hypertext are doing something like an Unknown writing jam right now. We generated a set of characters using the “pull characteristics out of a hat” technique, and four teams are each using the same set of characters to write in different “genres” — cyberpunk, noir, over-the-top romance, and reality tv/horror. I should be interesting to see what the end product looks like.

  12. andrew Says:

    It occurs to me that Mark Bernstein’s Thespis system would be a really appropriate platform for a Jam. I wonder if Thespis been used in this way yet?

    (I suggested similarities between the Thespis system and Facade system in my response to Mark in First Person.)

  13. noah Says:

    I believe Mark did some Thespis-related collaborations with the Equator folks, who also had a narrative project around a tool they called “Auld Leaky” (see Building Narrative Structures Using Context Based Linking and On Writing Sculptural Hypertext).

    It strikes me that I’d like to do an EWJ around the rule-based natural language generator for Terminal Time. Michael, what would that take, do you think?

  14. scott Says:

    Maybe the writing jame could happen on multiple occasions with different modalities — one idea that Nick and I chatted about briefly on Sunday was the possibility of developing a common set of characters and then working with them using different types of electronic writing techniques — hypertext, chatterbot, IF, etc. I think the key to making it work would be to scope the project(s) in such a way that at the end of a “jam” we have something that works on its own, whether as part of a larger whole or not.

    It would be cool to write something for a Facade style work, but I’m guessing that it’s probably difficult to produce something over the course of a weekend that you could just “plug in” and have it working. But if we were able to sort of modularize a project across different modes of electronic writing, we could probably make something interesting that would build incrementally.

    Nick also mentioned the problem of certain kinds of projects relegating one or the other of us (or more probably one of you than me given my comparatively minimalist technical skill set) to being “tech guys” or instructors. That would probably be the problem with a Facade style thing, or even an IF. But, on the other hand, if we could take turns being project director in different sessions, maybe that would work. I’d love the opportunity to learn more in that kind of hands-on way in the process of writing an interesting project.

    I always like the metaphor of the Band, who during the Big Pink sessions worked with a lot of guest artists and spent a lot of time playing each other’s instruments.

  15. andrew Says:

    Yes, I think a requirement is that a Jam should always produce something(s) that works on its own, that others can play.

    The idea of working a common set of character through different e-writing forms / modalities sounds interesting, I like it.

    While I agree a Facade-style Jam would be ambitious, I’m going to think about how to make it possible. Yes, it would be difficult, but accomplishing difficult things can be worthwhile :-). With enough up-front work, it should be possible to make an authoring environment where an author with no programming experience can at least specify a collection of dialog and basic direction for a “smart” virtual actor to perform. Then, if the author has some programming experience and/or teams up with a programmer, you can access relatively easy hooks for customizing existing behaviors. The next level of complexity, should you dare accept that mission, is to create your own brand-new behaviors.

    I’ll admit it sounds complex, but if it’s set up to allow participants to wade in as shallowly or deeply as you wish, it could work. And if we pull it off, it could be a very interesting piece to play with.

    Also I think the idea of a series of Jams, where each participant is the director of one of the Jams, makes logistical sense.

  16. Michael Says:

    It strikes me that I’d like to do an EWJ around the rule-based natural language generator for Terminal Time. Michael, what would that take, do you think?

    The Terminal Time generator, on its own, just takes a semantic representation of what you want to say and turns it into natural language. So, to make an interactive something-or-other with it, you’d first have to embed the generator in some interactive framework. What kind of interaction are you picturing supporting? Alot of the work of using the generator involves developing a semantic language (read, predicate calculus representations) for talking about whatever it is you want to generate text about. For Terminal Time, this was representations, based on the Upper Cyc Ontology, of historical events. To use it in a Jam, you’d probably want to provide the participants with a ready-made representational framework for whatever the topic of the Jam is.

    One thought that Paul, Steffi (TT collaborators) and I tossed around a couple of years ago was the idea of holding workshops where people could develop representations of their local history (with multiple points of view), throw them into Terminal Time, and make site-specific versions of Terminal Time. This is a kind of Jam, using TT as a mirror enabling people to explore their own history and ideological bias. But we’d definitely need some nice authoring tools that we don’t have right now. As Andrew points out above for an ABL-based jam, an authoring environment that allows one to peel away multiple layers of complexity in authoring, starting simple, but with a nice path to more complex authoring, is key. Neither ABL nor TT has such a (presumably GUI) authoring tool.

    One issue for developing a jam is deciding what level of programming background the jam assumes. I think the EGW jam assumes a game programming background (a C/C++ programmer who’s done real-time event-based programming). What kind of authoring tool you need depends on the participants’ background.

  17. andrew Says:

    The Unknown (Scott et al) and Rob Wittig had an informal writing jam of their own last weekend, using SubEthaEdit, which looks great. Rob describes it a bit on his blog.

Powered by WordPress