July 24, 2006

Sandcastle Construction

by Andrew Stern · , 2:00 pm

Tadhg Kelly over at particleblog (whom I recently linked to) has posted a good rant about the “interactive storytelling community’s” misguided notions of the nature and feasibility of interactive stories. In particular he points out that stories are delicate structures, so how can they be made interactive? The age-old question, but I like Tadhg’s take on it.

Please read his full post before reading my comments below, which I also posted on his blog.

Stories are certainly structures, and I like how you point out that they’re exceptionally delicate and brittle ones, where good stories (not derivative formulaic ones) each have their own particular custom structure.

I also agree with you that labeling gameplay as story (e.g. gameplay in soccer or Counterstrike), even games set in fictional worlds, is mistaken.

I don’t see why you resist the idea that two parties, i.e. the human player and the system, can each make meaningful “moves” (action and dialog), either symmetrically in a turn-taking fashion or even asymmetrically, to achieve a collaboration that, albeit loosely and messily, builds a structure over time.

Take the metaphor of two parties building a sandcastle together. It requires collaboration. The sandcastle is brittle. There are millions of interesting castles that can be built. If one person destroys part of the castle, it is possible for the other to work to repair the damage; the overall castle will be blemished, but it can still hold. Certainly it’s easy for one party to destroy the castle quickly, in one or two bold strokes. It’s even possible to inadvertently destroy the castle. That’s the nature of the beast. To alleviate this fragility, one party could be observant and warn the other party if they’re about to destroy the castle (purposefully or not).

But it’s too pessimistic to deny the possibility that complex structures can be built collaboratively. It takes cooperation, and forces responsibility. But control and responsibility is what agency is all about.

Further, in the case of a structure that is temporal, i.e. a story being created in real time, it’s not as easy to nitpick over the structure of the story as whole; you can get away with a little more messiness in its construction, I think, especially if the process of building the structure has frequent moment-by-moment rewards.

In a nutshell, the term “story” has too much baggage. Fuck terms. The point is, there are interesting systems we can build here that are capable of collaborating with the player to build a structure, if the player feels like working with the system. The structure may be haphazardly built, blemished, limited, weak in some spots and strong in others — but a structure nonetheless, and one co-created by the player, rewarding the player with a sense of creative accomplishment and personal expression.

Tadhg wrote, Interactive storytelling, on the other hand, tries to square the circle, regarding players as actors, games as sets of stories, and lots of other misguided notions that derive from linguistic tricks, misunderstandings of the principles of both games and stories, and a healthy invocation of the Technology God as a lodestone of possibility.

I have a few responses here. First, the term interactive storytelling is a poor one; an interactive story that offers true agency, which is what we care about here, is more of an interactive story generation experience, with no “telling”. There’s no perfect term (“movies” was a nice, short catchy invented term for moving pictures); “interactive storymaking” is better, at least.

I agree there are many misguided or ultimately unfruitful approaches to structuring an interactive narrative experience.

As regards to the Technology God, it’s true that technology is a key part of solution to creating a creative, collaborative AI system. Design and interface will play major roles too. Some of us engaged in such efforts write up detailed descriptions of the capabilities and limitations of our technology and design, which can help demystify or debunk the hype that surrounds the work.

Mostly, however, the notion continues to propagate because it keeps a variety of people in the news and mentioned in magazines. Like many other sources of self-propagating publicity (of which the industry has too many), it should be actively debunked. But that’s a subject for another article.

Yes, hype in general needs to be debunked. But don’t forget one source of hype (such as hyperbole in the press) is the strong desire by so many players for the experiences we’re talking about here. People want this stuff, it’s worth the effort to try to build something that achieves at least some of what people want.

10 Responses to “Sandcastle Construction”


  1. David Elson Says:

    Interesting thread. The sandcastle analogy makes me wonder whether the users of an ideal interactive narrative game can be conflated with players of FPS and similar games.

    In other words, when a story is co-created by the player, rewarding the player with a sense of creative accomplishment and personal expression, it sounds more like the kind of improvisational game I played in drama camp, or a Maxis world-building game, than the kind of game that tests the player’s skill at accomplishing specific goals within a world defined by rigid rules. Both are fun, but one seems to exist in an agency sphere subject to the unchangeable rules of the story-world, and the other in a distinct creative sphere operating outside the story-world (but possibly subject to rules of expression).

  2. Malcolm Says:

    I think this whole debate is misguided and asking the wrong questions. It’s not a matter of “Can computers tell interactive stories?” (whatever they may be), but “What new creative works can we build given these new tools we have, one of which is a vastly increased possibility of interaction?”.

    Consider the invention of moving pictures. Can you tell ‘stories’ with movies? Of course you can. Do they look like written narrative… well, no not really. Consider adapting your favorite novel to film and you’ll see what I mean. Access to the characters’ inner worlds is suddenly lost. You can write “John thought longingly about Mary.” but filming that just shows a guy staring into space with a silly look on his face.

    New techniques had to be discovered. Who discovered them? Not the people who were debunking the new medium. It was the people who were passionate about the possibilities. And they learnt by trial and error. Criticism was an important part of that process, but constructiveky not destructively.

    I firmly believe that computers can provide us with wonderful new ways of telling stories. They already have. Hypertextual fiction exists. Interactive fiction exists. Do these things look like traditional narratives? Yes and no.

    This whole area is still in its infancy and we are only just learning which techniques work and which don’t. It seems to me that a good way to start is to explore how structural concepts from traditional narrative can be transferred into this new medium, since structure is one of the things the new medium does best. Whatever tools we create, we will still need artists to use them to produce great works.

    Malcolm

  3. andrew Says:

    David, yes, I think so; I certainly don’t envision interactive stories to test the player’s skill per se; they should be entertaining dramatic/comedic play, not too much work, not onerous challenges.

    Malcolm, the reason we think about traditional stories is because there may be techniques of expression, and structures there, that are extremely powerful and potentially applied / adapted / at least serve as inspiration for creating work in this interactive medium. Perhaps we can cherry pick techniques from linear storytelling, but you’re right that it behooves us to sometimes take a more bottom-up approach of figuring out what we can make.

  4. andrew Says:

    Tadhg responded here; below is my reply, saved here for the record.

    Tadhg wrote: Because in that context there are only two kinds of move that you can make. One is the move that faithfully follows the pre-ordained path, one is a move that breaks it.

    If you work with my sandcastle collaboration metaphor for a moment, there is no pre-ordained path. You’re building a castle, which can take many forms, yet still be a structure. Such a system allows you to play, experiment, and build a structure in many, many different ways. I say allow, not force, because offering true agency gives the player the control and the responsibility to contribute or not to building this structure. (Again, in this example the sandcastle is a metaphor for a series of events that occur over time, that have a holistic structure.)

    I’d suggest letting go of the unfruitful approaches to story you’ve brought up, and think of scenarios designed to handle a wide array of player action, and still have some focus. Then you can avoid the problems you’ve brought up. This is a design problem. Many scenarios are not this robust — say, Hamlet, as well as most finely crafted stories we love from linear media.

    A classic example being in roleplaying games when players wilfully veer off the path or GMs sart dictating long paragraphs from an adventure with no interruptions. Neither is anything good.

    Again, imagine a scenario where there is no “path”, but instead a space of dramatic play in which the player’s collaborator, the system, can work with a broad range of action. In response to previous player moves, the system is making moves in an attempt to build a structure that is robust enough to take many forms. At every moment the player has many, many moves that can make progress in constructing this structure, because the structure allows for so many types of moves. Even moves that seem to “willfully veer” from contributing to this structure — say, destroying part of the structure — the system is free to consider this an intentional move and try to work with it.

    For example, if the player begins acting boorish, rude, talking about topics out of the broad domain of the scenario (e.g., investing in futures in pork rinds, or how to sail a boat), the system can interpret such moves in ways that allow it to believably make forward progress. One example (and there are many more): interpret boorish behavior from the player as disagreement with what the NPCs are talking about, and move forward from there. Improv techniques suggest to never say no, to work with what your partners do.

    There are limits; if the player does their best to completely destroy the structure, they can, and it’s probably best to just end the experience at the point and start again. Collaborative creation does require some degree of cooperation, no doubt about it. It may even be a skill that players learn, akin to how they’ve gained skills with manipulating an avatar with a controller (newbie game players can’t get Laura Croft or Prince of Persia very far without some practice and learning).

    This is the kind of dramatic space we attempted to set up with Facade: a social and psychological situation, in which it’s valid for the player to act in all kinds of ways, and still progress the scenario forward. I’m not saying we did a particularly good job of it, because it’s very difficult to implement this procedurally. The technology to understand the player has to be built and tuned, the authoring needs many iterations and playtesting to get right, etc. etc. But there are glimmers of success among the array of failures. Facade was research, an experiment.

    What does work, and works for computer games too, is when the world of the game is loose enough that the players can cut loose, but also triggered enough that they can discover more about the fiction at their pace. This isn’t any kind of structure, it’s a recognition that this is a game first and foremost, and therefore robust systems are the most important thing.

    Sure, that works an experience, but like you say, there’s no cohesive structure.

    Players don’t like to be guided ro structured. They like to play. They may think that they want an interactive story, but in practise they much prefer s GTA.

    Collaboratively building a structure (eg kids on the beach building a sandcastle together) is play. Open-ended make-believe for that matter can have structure. The process of building the structure just has to be robust enough to allow for play and experimentation.

  5. andrew Says:

    again, here I’m keeping a copy of my half of another exchange between Tadhg and me.

    But Facade didn’t work. … it’s interest value as a game is limited. All you do in it is push and see what the response in, and then just muck about it with it to see what the results are. It’s not really like you care or have any real stake in it as such, and the cracks appear quickly enough.

    Facade is certainly limited in its pool of responses, albeit quite large relative to the status quo of conversations in games; its biggest problem, I think, is its limited feedback for where the player is in the state space, i.e., it’s hard to tell precisely what effect you’re having on them. Without showing sliders of the characters’ internal state (a la The Sims), giving feedback purely in dialog and emotional expression, it’s a difficult feat to pull off well.

    Re: how much you care for the characters, that’s more a function of the quality of the writing, duration of the piece, and personal taste.

    That said, even in the spirit of co-operation, you can’t help but feel that you’re basically just filling in a set of pre-scripted responses every now and then, and it’s just not that fun after a fairly short space of time.

    To the extent players feel that, I see that as a shortcoming of this particular implementation, i.e. the system needs to be more generative, or at least have much more authored content; it’s not a fundamental problem with the concept of interactive drama.

    I didn’t come away from it with any real sense of a story built or told, nor of anything accomplished. It was and is a curio.

    I beg to disagree, of course. While Facade is a short, loosely-plotted drama, it’s a drama. A play in one-act.

    I’m guessing that until you see a more successful interactive drama built, you’ll deem the stepping stones that get us there as useless. Hey, it’s a free country. (er, countries)

    If anything, it reinforces most of what I’m saying. I can see that you might think that it’s an implementation issue, but to me it belies a conceptual issue far more.

    I’ve yet to hear something to convince me that it’s not possible to design a scenario that works how I’ve described. Bringing up examples of previous unfruitful approaches (tabletop RPGs, GTA3) does not prove there is no solution.

    Perhaps I should point to an example form that does work (I and many others have witnessed it first hand): long-form improv, e.g. the Harold. Now, while I think this requires too much training and skill on the part of the participants, it does show a proof positive example of interactive, generative, collaborative storymaking.

    The great thing about sandcastles is that they are very simple and made out of a component that is firm enough to mould but soft enough to shape, the shapes are easily made and that’s all there is to it. Add a set of pieces, architectural plans that need to be followed and so on and watch how the fun drains out of the experience almost immediately.

    Again, I’ll refer to the situation used in Facade (maybe “situation” is a better term than “scenario”), where the player is confronted with friends whose marriage is falling apart, and they try to suck her into the battle and force her to say and do things that will have irreversible effects on them. I’d argue that, from a conceptual perspective,

    - the events that unfold can be relatively simple, yet compelling
    - the situation is easy for the player to mold with a few well placed utterances
    - there are no pre-set “pieces” or “plans” that “need to be followed” that would drain the fun; it’s pretty open-ended how this situation can proceed and unfold. Lots of ways such a situation could play out and still be interesting.
    - the system can attempt to shape the overall experience by raising the tension at a good pace, cooling down the tension if the player raises it overly quickly. There’s a variety of topics that can be discussed, to keep things fresh from playthrough to playthrough.

  6. Malcolm Says:

    Andrew wrote:
    Malcolm, the reason we think about traditional stories is because there may be techniques of expression, and structures there, that are extremely powerful and potentially applied / adapted / at least serve as inspiration for creating work in this interactive medium. Perhaps we can cherry pick techniques from linear storytelling, but you’re right that it behooves us to sometimes take a more bottom-up approach of figuring out what we can make

    I totally agree, we need a starting point, and structuralist analysis of traditional stories seems like a natural one – computers are good at structure. And then we explore from there. I know little of the history, but I imagine that movie makers did much the same thing: imitating theatre at the outset, and then developing their own style and set of techniques suited to their medium.

    Malcolm

  7. Malcolm Says:

    Tadhg wrote:
    Players don’t like to be guided ro structured. They like to play. They may think that they want an interactive story, but in practise they much prefer s GTA.

    Maybe game players don’t like to be guided (although that is debatable, a game must have at least some structure, otherwise its just stuffing around). But in any case, not everyone is a game player. Novel readers obviously do like to be guided, in fact, completely railroaded without any choice in the plot whatsoever. Is there no middle ground? Are we to assume that no-one would be willing to participate in an interactive fiction in which their activities are constructively guided?

    I think the existance of improv games proves this assumption to be false. Everybody who does improv learns the golden rules: Accept offers, Don’t block. You are of course free to do what you want, but you will have a more enjoyable experience if you play in a constructive fashion. Why wouldn’t people be willing to accept the same rules in their interaction with a computer?

    Malcolm

  8. Grand Text Auto » String of Pearls in the Sandbox Says:

    [...] ve the player hop from little sandbox to little sandbox, but to have the player and system collaborate to make sandcastles. Also worth linking back to is my debate w [...]

  9. Grand Text Auto » Shadow Monsters and Revealing Light Says:

    [...] blank”>Camille Utterback and Zack Simpson, Worthington’s Shadow Monsters has players collaborating with the system to create real-time animated, vocalizing monste [...]

  10. Grand Text Auto » GTxA Symposium: Future Directions Says:

    [...] nts to do, to go where the player wants to go, to be a capable improvisational partner and collaborative writer with the player. This requires so much story content, that it [...]

Leave a Reply

Powered by WordPress