First of all, I have to admit to having a problem with the term Interactive Drama, as contrasted with the term Interactive Storytelling. To me, drama is the at the root of all interaction, in that I regard it as the output of a character: the character brings the drama. A character’s contribution to the storytelling – the list of its dramatic outputs – is a function of that character’s internal conflict with regards to the story problems. So before the character can bring the drama, he/she/it has to generate the drama.

Let drama generation start as an “interaction” internal to a character. As J.W.Goethe’s character Faust has it: “Zwei Seelen wohnen, ach, in meiner Brust” – “two souls dwell in my bosom”. This ur-conflict inherent to a dramatic character drives the type of story I’m interested in here, because it causes the character to generate dramatic outputs.

Such an output expresses a character’s current internal state, a state which, by ways of this output, impacts the story, and which is, in turn, impacted by the story, by ways of inputs by other characters. That means there’s a feedback loop: following an initial event that starts an interaction between two characters, each of the two takes the other one’s output as input, processes it, and generates an output that is taken as input by the other character. A list of all dramatic outputs of all the characters involved in a story, each list item representing one atom of drama, is, for my purposes, the definition of the documentation of one instance of Interactive Storytelling. Interactive Storytelling, then, is the process that results in such a list.

Species don’t evolve without selection pressure. The dinosaurs remained pretty much static for millions of years because their system worked. Hominids didn’t develop because some simians decided that it might be interesting to see what would happen if they developed bipedalism, larger brains, and a hunter-gatherer lifestyle. Evolution happens only because there’s a crisis that’s killing off the species, and some lucky critter has the weird genes that provide a solution to the crisis. Toynbee had the same idea with his study of history: civilization are responses to challenges. No challenge, no change. The games industry has no incentive to change, because it’s too comfortable making money the way things are.

A second arguement: you don’t get feature evolution without market evolution. In other words, you don’t just add some new trait to a game unless there’s some part of the market that will respond to it. Suppose, for example, somebody wanted to add a shopping feature to their shooter. The market for shooters doesn’t give a damn about shopping, so the game will fail. Sure, you can evolve the feature set, but if the market doesn’t have evolutionary potential for that feature, it won’t work. Monkeys aren’t going to start growing webbed feet until after they come down from the trees and find deep water nearby.

Technology is full of examples of revolutionary development. It’s true that the vast majority of technological development is evolutionary, not revolutionary. But there are some things that simply can’t be done without a revolutionary approach. Airplanes did not incrementally evolve out of railroad engines. Personal computers did not incrementally evolve out of mainframes or minis. A-bombs did not evolve out of firecrackers.

I did read about Siboot in your “On Game Design” and was intrigued by it. In fact, it was one of the things that was formative of my thinking on these issues. However, I must admit I haven’t gotten beyond reading about it to playing/seeing it.

The thing is, I think the game industry would be a lot more responsive to a seemingly marketable attempt in that direction than they would be to the kind of complexity that, say, bots seem to want to introduce, without a compelling example. I’ll grant that game publishers are terribly conservative about what they consider, but I really just don’t think we’ve found the example game concept that demonstrates why interactive narrative matters.

When there are “Katamari Damacys” and “Icos” and “Animal Crossings” out there, and when “Puzzle Pirates” can make huge cash as a garage operation, I feel like somehow quirky and innovative ideas manage to make it to market when they are eminently ‘playable’. And I just haven’t seen an ‘interactive narrative’ or something that is even approaching that which seems playably compelling.

I’ll admit that my own perspective is limited; in fact, one of the reasons I’m looking forward to school is to have a chance to see and discuss all the work that’s going on that I’m missing (without cash/time away from my day job to go to the GDC, etc). However, I have a few game industry contacts, and they say the same thing about ‘interactive narrative’: where are the examples that make me want to pick up the controller/keyboard and never put it down? That make me say: “I’ll never live without interactive narrative. It’s just so compelling, I can’t resist it.”

I don’t know if evolutionary approaches “from games” will work. But I do know that any progress will be incremental, whether we take games as our starting point or not. I just don’t know how it could be otherwise with any sufficiently complex human endeavor. I also don’t see what’s wrong with using some of the lessons learned from games to inform how we go about working with narrative, even if we don’t see the financial structures or markets that have evolved to support traditional games as suitable for ‘interactive narrative’.

First, I don’t agree that evolutionary approaches from games will work. The games industry has specialized in a style of entertainment inimical to storytelling. They’d have to unlearn much of what they already know in order to start evolving towards storytelling. It’s not going to happen.

Second, your comments on ’simplified sign systems’ seem unaware of work that has already been done in this field. You might want to look at eeyal, the language I created for use in Siboot, a game, 18 years ago. Even more useful would be Deikto, the language I am developing for my interactive storytelling technology.

With a BA and an MA in English, I’ve learned from the best to fling my high brow views around with an air of entitlement.

Game industry people don’t like writers less because of expense than because they all secretly want to BE writers. Of course, if anyone in the world is capable of producing more crap than Hollywood writers, its game designers and publishers, who tend to think “Ender’s Game” is the height of Western Culture and who think every dramatic moment can be improved by a good inside joke.

What I’d really like is if game designers AND Hollywood scriptwriters AND the money people involved in both industries were a little less ignorant about what makes good storytelling – in fact, less ignorant of culture in general. I’d also like the American people to gain 35 IQ points on average and the ability to stop time while I sleep.

And, BTW, if you’d like to do some more trolling, as you can see I’m more than up to the task as well.

Thanks also for informing me on the unsuitability of “spoken standard english” as an interface. Lately, I was getting those, you know, “ideas”… but you sure shot that one down with your brillant starting questions, and that was that. Looking forward towards your sign systems, which I’m sure will be much more than just reasonably complex.

Just recently, in Knights of the Old Republic, we see Bioware using this kind of ‘english sentences as signs’ model – where choices are put before the player and they clearly indicate a ‘good’, ‘neutral’, or ‘evil’ path. In that case, though, the semiotic ‘atoms’, or smallest significant interactive semiotic units, are sentences or paragraphs, and not words or phrases.

This is a classic example of the kind of ’shortcut’ I think we need to develop. Except frankly, the KoToR example is not very sophisticated or dramatically engaging (kudos to Bioware for actually caring enough about dialogue to make it interactively relevant at all).

How might we refine the Bioware ‘morality sentences’ sign system? How might we give it more interactive AND dramatic interest?

It’s a flash game that uses rudimentary symbols as its interactive medium.

http://www.abc.net.au/gameon/tork/torkgame.htm

I’d like to think I’m saying something a lot subtler than “no stories now!” I’m saying we should first seek to make games (or , more broadly, interactive experiences) that are more emotionally and socially gratifying, and if, in the process of seeking that, we find narrative techniques making their way organically into games, that’s all well and good. But our first goal should be to create ‘authentically interactive’ media that are more reflective.

Instead, we seem to be proceeding from a ’structural’ concern to get the kinds of narratives with which we are already familiar into games (Hollywood scripts being the most popular ‘model’), without any real sense of the specific benefits of those structures to this new medium. And that just feels forced.

The game, ICO, for example, which I, along with many other people, find one of the most emotionally gratifying ‘game’ experiences yet created, has a very minimal, non-intrusive, narrative. However, the folks at SONY focused on integrating that narrative thoroughly into the gameplay. Because the elements of gesture and sound design were so well done and so well integrated into the overall art direction, ICO ‘felt’ like it had more real ’story’ than far more ‘narratively complex’ games, such as Metal Gear Solid 3.

In other words, by focusing on the holistic experience and integrating just a small number of ‘narrative rudiments’, ICO created more ‘interactive narrative’ than anyone has ever produced by churning out thousands of lines of dialogue. And the same for the ‘bot’ community who are so focused on the intimate details of gerundives that they can’t see that ‘Alice’ (or ELIZA or what have you) is basically a parlor trick, without very much real human interest.

I totally agree with you on the need for engineers and storytellers to begin speaking to one another about their respective crafts (in fact, I’ve found myself saying much the same in other forums). However, I think one of the hard lessons storytellers will take from this ‘meeting of the minds’ is that they must start at the beginning, and that means dropping the very notion of ’spoken standard english’ as an interactive medium at all. It’s just too complex to integrate into interactive experiences at this stage.

Now, if we wanted to do really valuable (and incremental) work ‘on the way’ to spoken standard english as a medium, we could start with simplified ’sign systems’ as a means of interaction, and in fact, that’s what we already have in current games. When you push a button to fire a gun to ‘kill’ an enemy, you are already participating in a reasonably complex sign system. The question is, how do we make the signs for doing –other– things that are not physicalized experiences, interactively interesting? And how do we design goal-directed challenges areound those ’signs’? That, to me, is a better ’starting question’ than ‘how to make interactive narrative’?

We may find that ‘narrative’ or ‘fiction’ are not good words for describing what actually ‘works’. We may find that the “structured experiences” that are really successful (aesthetically and/or financially) are more different than ‘narrative’ than we at first imagined.

To be frank, I’d be very surprised if this turned out to be true. It’s not the first time that I hear this suggestion, either; maybe the “Why Johnny need not storytell” meme is the flipside of “Why Johnny must program“. Meanwhile, I still have to find people from the “engineering” side of things who are able to discuss when to use, say, a Do-er vs a Be-er approach for a literary character with the same level of competence that they seem to expect from people from the “storytelling” side when discussing, say, the use of Nodes vs Edges for frequency measurements.

As for “the tasks ahead”, I’ve already discovered mine a while ago: find a way to write “interactive characters”, defined as “texts that can be reasoned with”. By “reasoning”, I don’t refer to the idea of a “canonical” model of reasoning, as was found in GOFAI, but to a “contextual” model, as is used by storytellers (though the implementation may well rest on GOFAI techniques, and I’m not in any way dissing GOFAI’s many accomplishments). I think that this use of the “story” concept has an analog in the use of the “scenegraph” concept by the CG folks – it’s what you refer to as a “shortcut”.

Stories, to me, are a well-known “shortcut” for the communication of meaning, involving the exclusion of all the stuff that’s not relevant at the moment (stuff that’s “not in the frustrum”, so to speak). In short, I believe that the general cause of “interactive storytelling” could be much furthered if engineers learnt more about what’s what in “non-interactive storytelling”. Like, how characters only become characters by interacting: in reality, there ain’t no such thing as “stories without interaction”. It’s only that, in “old” media, the audience most often has no way of being “interactive characters”, too (there are exceptions, like Augusto Boal’s “spect-actors”).

I’ve heard engineers suspect that I only insist that they learn this mushy-faced storytelling stuff to get them to do the hard work of developing the infrastructure needed to build interactive characters, so that all I need to do is to fill in the blanks they provide with some lofty prose, and get all the props for “inventing” such wonderful creatures. Not so. I’m pretty independent these days from people developing new tools, because I found that the basics of what I need already exist (not that the existing tools couldn’t be much improved to make me more productive, but that’s optimization I’ll gladly defer until I know exactly what I’ll need in the long run). So my motives for promoting this kind of learning are not that egotistical (OK, so I expect a rising tide of storytelling competence to lift my boat, too, but I think this is legit).

From how I see it, to begin developing digital “interactive characters”, a storyteller needs to know how ELIZA works, how SIR works (for the abstract of Bertram Raphael’s thesis, scroll down to the section “AITR-220″; there’s a link there to FTP the 168-page PS/PDF document), and, to get something of a bigger picture, how LISP works (this is to learn how to implement self-reference through recursion; both ELIZA and SIR are not “characters” that can refer to themselves, and thereby put themselves “into the story”). Yep, those ideas are, like, forty years old and older. But the rest of the needed ideas are 400 years old and older – Shakespeare was (at least in the context of “Western” literature) the first storyteller who “got characters right”, after many generations of storytellers trying to “get there” since before Aristotle (whose “Poetica” is, like, 2300 years old, but still suggested reading for screenwriting novices). You can see that I don’t expect engineers to come up with any technical breakthroughs which will make an “interactive storyteller” out of me. Anybody who ever took a class in improv acting knows that “interactive storytelling” ain’t nothing new at all. All we’re trying to do is to get the “digital” part of it right.

What I do think would be helpful to further the cause of “interactive storytelling” in general would be engineers who not only demanded (rightfully) that, in order to play game, storytellers must learn their methods, but who also took the effort of learning ours. From reading all those conference papers, and seeing them develop “authoring tools”, I get the impression that many of them think they know how to do it, and don’t. In addition to discussing procedural literacy courses, why not also discuss character development techniques? It’ll happen anyhow; it might as well happen here (yeah, you’re right – this is a great blog).

So, onward with my thoughts:

The desire to “create interactive story” is just so monolithic. In the early days of computer graphics it had to be fairly unlikely for, say, J.J. Gibson to set for himself as a goal: “to create the visual field”. The state of hardware had to have made graphics pioneers too humble for that. With the hardware we have today and the wild claims of AI theorists about what machines do/should do in 20 years based on Moore’s law, etc, I think we have a tendency to miss that software advances in many areas that affect game design have come very incrementally in conjunction with hardware advances. If we are to have ‘interactive story’, its likely to come just as incrementally. Unfortunately, the things our hardware is already accomplishing in other areas is impressive, and it can make us feel pressure that wouldn’t have been there if we’d been entertaining these ideas in 1970. No one person is going to ‘invent’ interactive story, either, any more than one person ‘invented’ computer graphics. In fact, it is a bit surprising to me that you all, who are so devoted to the task, seem to think it is so ’small’ a task to be bitten off by one or even a handful of people, in a single generation.

If for no other reason, the incremental nature of the progress made will result from each advance needing to have its own kind of market/aesthetic benefit. Not only is ‘interactive story’ unproven in the market, but few of the ‘components’ of what might one day contribute to ‘interactive story’ are proven in the market either. For example, more sophisticated facial animations. How much did the admittedly fine work of the Half-Life 2 team to make the faces of its characters convincing really contribute to the bottom line? Are there ROI analyses or focus groups that could prove this to a publisher? (I frankly, was disappointed by how little ‘interactive leverage’ HL2 got out of their face work.)

What’s more, though graphical innovations do have a long track record of directly benefitting the bottom line (inflated a great deal in the minds of publishers, granted), we often underappreciate the poverty of even the best representations of visual fields in computer games. As I know you are all well aware, there are huge shortcuts that we can take when representing space because of the nature of the human visual system.

Before we can “create interactive story”, don’t we need to ask ourselves “How subject is the human cognitive apparatus that deals with linguistic and narrative ‘fields’ to the kinds of ’shortcuts’ that are so commonplace in other representational models?” and “How sophisticated is the level of interactivity offered by games when placed next to real human experiences in the world, like say, that of a painter with a brush?” I think we will find that, despite the radical success of Gibson’s “surface and incident light” model of visual perception, and despite the intial wonder evoked by the use of input devices to navigate 3d space, on the whole the experience is really quite poor. The simulations offered by games are quite unconvincing, and what’s more, they are highly conventionalized. Gamers perform huge acts of ’suspension of disbelief’ when they play games, from physics to vision to ‘enemy AI’. As we pursue and encourage the industry to pursue innovations which support complex emotional/social/intellectual experiences in games, we need to be reasonable about the quality and kind of experiences that are realistic, and not set higher standards than those already at play in other areas.

What’s more, we need to be open to an evolving and iterative process. It’s not just an issue of proving to publishers that innovations which lead directly to better emotional/social/intellectual game experiences are marketable. It’s an issue of evaluating their genuine aesthetic benefit to ‘players’ and ‘bringing players along’ on the many ‘conventionalizations’ and ’shortcuts’ we make as we try to boil complex real-life experiences down to biteable interactive chunks. When the the Lumieres showed their train footage, French audiences ran. When Welles first screened Citizen Kane, the angular shooting was extremely off-putting. When David Fincher started blasting quick cuts and microphotography at audiences, it was stimulating to young people, but it gave my parents headaches. Not only is innovation incremental, but so, naturally, is the ability of audiences/players to accept and engage with new techniques.

Instead of ‘creating interactive fiction’, we need to go about finding out what kinds of interactivity yield positive emotional/social/intellectual experiences. We may find that ‘narrative’ or ‘fiction’ are not good words for describing what actually ‘works’. We may find that the “structured experiences” that are really successful (aesthetically and/or financially) are more different than ‘narrative’ than we at first imagined. We must be –open– to discovering the tasks ahead, and not foreclose them so much by convincing ourselves that we already know what it is. All issues of mathematics aside, this is also a major difference between artists/writers and typical developers. The former often set out making something without a concrete idea in their head of what they will make, finding the open-endedness of the approach more conducive to really outstanding results.

Sorry about that ambiguity. The sentence should have read “Joe asks Mary barn how much dirty?” which makes the meaning clear.

BTW, the name’s Dirk. Not Dick. Dirk. Ok?

Sorry about that. I suggest that you retaliate by calling me Chirs. May I call you Irchard?

After much pain at coding, I have decided to tear apart Deikto grammar a second time and resort to an even more abstract system that introduces formal parts of a sentence and dispenses entirely with clauses. It is certainly disheartening throwing away a ton of code, but that’s what is necessary.

So defining every single concept and the circumstances of its – possibly varied – use in a story is pretty much it for me.

Could you provide an example of what you mean by this? I’d love to see an example of what you mean by defining every single concept. Is this a strictly verbal definition, like what you’d get in a dictionary? Or is it a more extended definition?

At the highest level of my model is the story; the smallest independent functional unit in my view of “story” is the individual character. The “characterizing” feature of a single chraracter is its ontology. A character has at least one definition for each concept contained in its ontology (like in a dictionary, where a concept can have several entries, or “senses”, there may be more, dependent on the context the story provides/needs). The character also knows at least one purpose, one reason, one method, one time and one location relating to each concept, linking the concept to other known concepts.

There’s a pragmatic reason for this. The minimum ability I require from a character before I consider it to be “interactive” is that it can plausibly – i.e. not contradictive to the current story as a whole – answer the questions “What is it?”, “What?”, “Who?”, “Where?”, “When?”, “How?”, and “Why?”, refering to any of its acts (including, of course, all speech acts). I find that I get that covered by using the six predicates I mentioned. I’m not trying to say that that’s sufficient; rather, it seems like that’s the absolute minimum for me. So that’s why I start there.

And oh, yeah, different characters use different ontologies, of course. Conflicting ontologies – e.g., conflicting definitions of the same concept -, to me, is an obvious way of designing dramatic conflict between characters.

Writers see language primarily in terms of its richness and nuance; the notion of a stripped-down language is abhorrent to them.

In all fairness, but this seems like a bit of an over-generalization to me. I very much like “little” languages – as long as they’re recursively extendable, that is. That way I get a chance of sticking to simple principles, while adding refinements in a top-down, iterative way.

I am looking for storytellers, not writers. A writer lives and breathes the beauty of language; a storyteller is more interested in the dramatic dynamics of the storytelling process. Storytellers first encountered the separation of these two passions with the cinema. For thirty years, they were not permitted any but the most limited language. Some objected that it was impossible to tell any coherent story without recourse to language, but they were disproven. It just took some rethinking of the nature of story and the means of expression. We face the same problem with interactive storytelling.

“Writers vs storytellers” – hm, that’s a novel one. How widely did you test this? I think I can get at what you mean, but since I’m coming in from TV/movie writing, the distinction doesn’t seem to make that much sense in my surroundings. When I’m considering language, writing, and coding, my primary question is: How can I expose the largest possible “surface area” of story meaning for the client to interact with, while using the smallest possible amount of code/text? So to me, brevity is most beautiful. Besides, if I weren’t “interested in the dramatic dynamics of the storytelling process”, I don’t think I would be interested in algorithmic solutions to storytelling problems, either.

I am currently plowing through the complexities of implementation of the language. The most recent problem has arisen from the grammar of the verb “ask”. There are several types of clauses that can attach to this verb: fact, opinion, or gossip.

I see here that your procedure is indeed very different from mine. Apart from the precedence which I, as mentioned, grant to the class of “wh-questions”, I would define and explain the concept of “asking” in some thematic range, probably “questions vs answers”, depending on how a particular character would interpret and use it in a particular story. I could introduce refined concepts, like “asking for facts”, “asking for opinions”, etc, each with its own set of answers to the “what”, “where”, “why”, etc, but wouldn’t introduce any particular grammar problems with this. Not that I don’t have my own major problems; it just seems that using “verbs vs nouns” as a shaping dichotomy leads to quite different problems than using “predicates vs concepts”.

Joe ask Mary barn how dirty?

Is that supposed to be read “Joe asks Mary how dirty the barn is”, or “Joe asks Mary how the barn got dirty”?

To take this as an example: if I needed that in a story, I would first write the concepts for “barn” and “dirtyness”, so that I could know what a “dirty barn” is. The “method” for “dirty barn” could be used to explain how the barn got dirty, and to rate degrees of dirtyness, I would use a simple quantifier system, quite similar to yours.

I’m still digging through the mechanics of all this to insure that it really can be said in a clean fashion. It’s slow, grinding work checking out every possibility to make sure that it all works.

I do think that I know what you mean…

BTW, the name’s Dirk. Not Dick. Dirk. Ok?

There’s a ton of documentation, but I am keeping it close to my vest for the moment, not out of any sense of paranoid secrecy, but rather a healthy appreciation for the likelihood of major changes. Indeed, the dictionary posted on the website has already been altered considerably.

I am currently plowing through the complexities of implementation of the language. The most recent problem has arisen from the grammar of the verb “ask”. There are several types of clauses that can attach to this verb: fact, opinion, or gossip. Here are some examples of each type:

Fact: Joe ask Mary Tom height how much?
Opinion: Joe ask Mary Tom nice how much?
Gossip: Joe ask Mary Tom nice how much as per Sean?

The ask-fact is further complicated by the fact that it can also be applied to stages and props:

Joe ask Mary barn how dirty?

I’m still digging through the mechanics of all this to insure that it really can be said in a clean fashion. It’s slow, grinding work checking out every possibility to make sure that it all works.

This raises one of the interesting disputes we have had in designing the Erasmatron. Writers see language primarily in terms of its richness and nuance; the notion of a stripped-down language is abhorrent to them. I respect this feeling; after all, any writer who has spent the many years necessary to master the intricacies of finely-wrought language would be loth to abandon so fine a tool.

Yet that is exactly what I think necessary for interactive storytelling. I am looking for storytellers, not writers. A writer lives and breathes the beauty of language; a storyteller is more interested in the dramatic dynamics of the storytelling process. Storytellers first encountered the separation of these two passions with the cinema. For thirty years, they were not permitted any but the most limited language. Some objected that it was impossible to tell any coherent story without recourse to language, but they were disproven. It just took some rethinking of the nature of story and the means of expression. We face the same problem with interactive storytelling.

So defining every single concept and the circumstances of its – possibly varied – use in a story is pretty much it for me.

Could you provide an example of what you mean by this? I’d love to see an example of what you mean by defining every single concept. Is this a strictly verbal definition, like what you’d get in a dictionary? Or is it a more extended definition?

Not to say that the concept of a verb as an action within your storyworld would not be a central concept, but the use of words throws up lots of natural language problems that can be (somewhat) conveniently sidestepped when one takes a “The Sims” approach, representing communication and emotion symbolically.

We can, theoretically, “synonymize” almost any word in Deikto, based on individuation by Actor, Stage, Adjective, Verb, Role, or even special combinations of the above. However, this could end up being a monster mess, and so we don’t want plunge into it unless we are absolutely certain that it’s necessary.

I really am afraid that it’s necessary for any interactive story I might be interested in writing. I feel that I just have to live with this potential monster mess, and see which kind of tools we can develop to reign it in, and make it somewhat less messy. There are some ideas.

Apart from the dictionary, is any other documentation of Deikto already available? What are your plans for this language?

When I started out a good four years ago, I didn’t know diddley about computer programming. I just slipped through its cracks in kind of an odd way, suddenly convinced, by the logic of a certain progression, that now I had to figure out how these technical languages work, and how to encode a story’s meaning in them. About half a year in, I hit the big wall: “Gee, if you want to do that ‘interactive storytelling’ thing, you’ll have to define each and every single concept you use in a story. And if you use a concept in two different ways, or meanings, in a story, you’ll have to define it twice! And man, there are words, like “why”, that can mean thousands, even millions, of different things in a story! And then you have to figure out how to get this machine to not confuse all those meanings, and say the right thing at the right time! Noooo!”

Then I started studying how you engineers are trying to do interactive storytelling. Learned the lingo to understand all those papers about planners, story directors, drama managers, dialog acts, and similar software objects. Built my own little models, etc. In short, I couldn’t crack it that way. How are you people getting any text integrity while writing those systems? You have one chunk of text that’s supposed to “manage” another chunk of text, but the “managing” chunk has no idea of what the “managed” chunk does, what its effects are, why, etc. I worked hard at trying to understand how you mean to do this for two full years, but it still does not make sense to me. After two years of drowning in nested infinite regresses, I gave up. It’s the single big incompatibility I’ve experienced between me as a writer and the engineers.

So defining every single concept and the circumstances of its – possibly varied – use in a story is pretty much it for me. The game industry, and other software industries interested in solving the NL problem, hate this idea. I sometimes hate it, too. It’s the most difficult way of writing ever. And no tool support; toolbuilding is part of the process.

But since I do not see any other approaches working, and can’t even understand how they should work, I’m sticking with this. We’ll see what drops.

A good (old-fashioned, linear) story – the type where the problem seems bafflingly complicated and unsolvable in the middle, but where everything “suddenly makes sense” in the end – often makes use of a certain algorithm. Two witers, Melanie Anne Phillips and Chris Huntley, have laid out this algorithm in their “Dramatica” theory book. An application example for their theory is in showing that even Shakespeare (perhaps unconsciously) used this algorithm, by analyzing and factoring plays like “Hamlet” and “Othello”, using the Dramatica system.

The Dramatica software is quite popular amongst screenwriters, story editors and (some) directors, and since “Harry Potter IV” easily factors into the system, I suspect that Joanne K. Rowling (and/or her editor(s)) is using it, too. I’ve been using it for about ten years now, and for the past four, have been trying to figure out how to use it for direct interaction with the audience, instead of as a structuring tool for writers and a communication tool for people working together producing linear stories (which is how it’s commonly used).

So that algorithm is the bridge I have crossed to meet you here: hello, engineers. I want to tell you something, regarding mathematics: If and when we have a good day, we writers use verbs and nouns with the same precision in reference to an algorithm and its production rules as engineers use numbers (on their good days :-).

And here is where my interest lies: if we could find a translational algorithm that corresponded to the algorithm that we already use to encode a story’s “meaning”, then we might be able to develop a new class of systems where the writers could just write, using verbs and nouns, and where the machine could translate the “meaning” of the “word” symbols into number class symbols, which it could then process, and after processing, translate back into the “word” class of symbols. Please note that I’m not trying to work on a “Cyc-like” problem here – “meaning”, for my current purposes, is not “meant” to exist in an “open” context, but restricted to the “meaning” in, and of, a particular, “meaning-coherent”, story.

So I’m working on an interpreter that takes any natural language input string, relates it to a story structure, and creates “events”, which are strings written in a language based on symbolic algebra. But since those symbols are words rather than numbers, I seculate that writers can learn that language much faster than they can learn, say, Java. Which means that, given some training, they should be able to produce tiny pieces of content – sentences, or even parts of sentences – that the machine could use to synthesize resonses to user input that “make sense” in the context of the current story. Anyway, that’s how I hope it’ll work – we shall see :-).

I’m currently building a prototype implementing some basic ideas; I’m planning on a GPL/LGPL-licensed release in September now, with public testing starting in late May.

the user – class Writer, superclass of Storyteller – is expected to write numerical equations in their place.

This won’t do for most writers I know, who need to be free to just use verbs and nouns as the story evolves. There’s a flow in a pleasant writing experience, and all successful storytelling tools, starting with Aristotle’s Poetica, are able to work in relation to this flow. An overhaul of both authoring systems might therefore involve a switch of responsibilities: let the writers control their verbs and nouns, and let the programs control the underlying numerical calculations.

I agree that writers abhor mathematics and requiring them to use mathematics will chase away a great many talented writers. However, this is unavoidable, for the very essence of interactivity lies in the algorithm, and the clearest and most expressive way of describing algorithms is in the language of mathematics. To put it another way, if you want the writers to be able to control their verbs, then they must be able to control what the verbs DO. How can you clearly specify an imperative without using an algorithm?

fyi, Espen Aarseth chose to sit in on Ernest Adams’ Interactive Narrative Research with us, rather than on Raph Koster’s Game Design Atoms, if that tells you anything.

More later…

Definition
by Chris Crawford

Process intensity is the degree to which a program emphasizes processes instead of data. All programs use a mix of process and data. Process is reflected in algorithms equations, and branches. Data is reflected in data tables, images, sounds, and text. A process-intensive program spends a lot of time crunching numbers; a data-intensive program spends a lot of time moving bytes around.

Idea takeaway for general system architecture, regarding interactive digital authoring systems to be used by professional writers (i.e., people like me): let the computers do the number crunching, and the writer do the moving-bytes-about part, where “bytes” are “words”.

Destilling Crawford’s text, I might come up with the following illustration:


world = process ,
data
;
process = abstract ,
indirect ,
[ ?equation? ] ,
[ ?principle? ] ,
[ ?force? ] ,
[ ?thought/idea? ]
;
data = tangible ,
direct ,
[ ?number? ] ,
[ ?fact? ] ,
[ ?event? ] ,
[ ?fractal-of-knowledge? ]
;


To generalize over this illustration: process seems dynamic to me; there’s movement from location to location, and it’s usually operationalized through the arcs of a program’s abstract search tree (AST). Data seems static, under ideal circumstances; at the time the process uses it, it is expected to represent very specific names, events, etc., at very specific times. Refering back to Chris Crawfords definition, it’s the “byte-moving” part, which has the writer writing text to be stored in the nodes of the concrete network.

As for now, it is still unclear to me whether useful programs can possibly be written by writers where the programs are their own writers, maybe in the way that the “head-writer/gag-smith” arrangement common in the industry of soap-opera production works. How could a process for that be realized?

To me, this now more than ever seems to be a promising research direction, whatever the popular opinion at the GDC might have been. I hope that pioneers who have chosen this direction, like Chris Crawford, will find the strenght to keep on for the last mile and still keep their loved ones by their side. I (married, 3 kids) am convinced that it can work.

This won’t do for most writers I know, who need to be free to just use verbs and nouns as the story evolves. There’s a flow in a pleasant writing experience, and all successful storytelling tools, starting with Aristotle’s Poetica, are able to work in relation to this flow. An overhaul of both authoring systems might therefore involve a switch of responsibilities: let the writers control their verbs and nouns, and let the programs control the underlying numerical calculations.