Request for an example

I am working on the second edition of my book on interactive storytelling. This new edition will include a chapter devoted to interactive fiction. I claim in this chapter that IF does not offer interactive storytelling; it offers only noninteractive stories. I would happily reverse that claim if provided with sufficient evidence to the contrary. I’m here seeking such evidence from the readership.

The evidence I seek would be provided by a transcript from any work of IF demonstrating a dramatically significant interaction between the player and an NPC. By “dramatically significant interaction”, I mean a sequence of actions taken by the player and the NPC that evoke a dramatically significant response from the NPC appropriate to the player’s actions, and could have provoked a different dramatically significant response had the player chosen a different course of action.

For example, it’s not adequate if the NPC carries out a dramatically significant action that is not dramatically responsive to the player’s actions. If the player calls Floyd an ugly toad, and Floyd then sacrifices his life to save the player’s, that’s a dramatic action on Floyd’s part, but not one that is appropriate to the player’s action.

To put it more succinctly, the transcript should demonstrate some emotional intelligence on the part of the NPC. If you prick that NPC, he should bleed. If you tickle him, he should laugh. If you wrong him, he should revenge himself upon you. He should act like a character in a story.

I think that it’s necessary to have the interaction between the player and the NPC extend over a sequence of actions. I would not be convinced if a player, upon first meeting an NPC were to say “Yer mudder wears Army boots”, causing the NPC to break down in tears, while saying “I love you” triggers much hugging and kissing. No magic-coin-in-the-slot stuff – I want to see convincing interaction.

This is a tough specification, and I realize that it cannot be satisfied – not just by IF but by almost ANY interactive work. Still, how close have we come with IF? What’s the best piece in this regard? I would be greatly indebted to anybody who can offer a transcript from a publicly-available work of IF that demonstrates such behavior.

Why not Galatea?

You should probably understand that from an IF perspective, the conflation of ‘genuinely interactive storytelling’ with ‘highly responsive NPCs’ is, if not totally wacky, certainly not an obvious conclusion. (IF is not centrally dedicated to the development of highly responsive NPCs. It’s more of a cool feature that some people do some of the time.)

The state of the art in IF NPCs, and one that satisfies at least some of what you want, is probably Alabaster. Progue in Blue Lacuna is also supposed to be highly flexible, though I can’t speak to that myself.

Matt, I did look at Galatea, but it did not meet my requirements.

“…from an IF perspective, the conflation of ‘genuinely interactive storytelling’ with ‘highly responsive NPCs’ is, if not totally wacky, certainly not an obvious conclusion.”

Interesting idea, Maga. Are you suggesting that the dramatic interaction could come from sources other than the NPCs? If so, what might those sources be?

I’ll go look at Alabaster.

Sorry, my question was non-rhetorical; what is it about Galatea that didn’t meet your requirements? That would help me know what you’re specifically looking for.

First, I had some problems just getting through it; the real issue, though, is that Galatea doesn’t really respond to me; she has her little speeches she delivers, and the writing is excellent, but I can’t interact with her as a character. I can push a few buttons to make things happen, but that’s about all. I was not disappointed by this; my impression is that interpersonal interaction in IF just doesn’t happen. However, I don’t want my impression to go into print before I at least search for something to contradict that impression.

I looked at Alabaster and was surprised to see the conversational tree directly available, so I studied that. It’s quite impressive, but it makes clear that there’s no substantial interpersonal interaction. It’s all quite boolean in style: push the button and get the response.

I’m not expecting IF to provide interactive storytelling; it’s not that kind of medium. However, I want to verify this belief before publishing it.

Well, hrm. If the issue is that the speeches themselves are more or less predefined, then you won’t find what you’re looking for in IF or much of anywhere else; sensible procedural text generation is hard as nails, doing it as dialogue for a character even more so. (I tend to agree with you that “the authorial voice in procedurally generated IS is expressed at a higher level of abstraction,” or at least that we can aspire to that, but I feel like procedural generation of text that you can interact with in a way that reaches into the world model* is incredibly ambitious even if it’s just at the level of room descriptions.)

But if the idea is that Galatea will respond the same to the same stimuli regardless of what’s gone on before, that’s not true, is it? The game tracks her state in various ways, as I understand it. Another thing that might be worth looking at is Deirdra Kiai’s The Play, which is a multiple-choice game, but which tracks the attitudes of the NPCs – if you step on someone’s foot, he won’t react nicely to you later.

If your point is that this sort of interaction is much more coarse-grained than what you’d find from a real character in a story, that’s certainly true.

*As opposed to something like Hunter, In Darkness, where I think you can examine some of the words that appear procedurally – which in itself is impressive – but you can’t do anything else to them, I don’t think.

At least at the immediate level, IF tends not to be about Man vs. Man; it tends to be about Man vs. World, occasionally Man vs. Himself and often Man vs. Thorny Technical Problem. So, yeah, you can absolutely have dramatic interaction with medium-sized dry goods and simple NPCs. It limits the kinds of stories you can do rather sharply, of course – but most IF innovation is about working out new things you can do within that space, rather than trying to expand the territory. (But I suspect that the issue here is that your definition of ‘genuine interactive storytelling’ is much stronger than what most people would understand by the phrase, so I doubt this will satisfy either.)

(Oh, and Make it Good should probably be mentioned if we’re talking about reactive NPCs, too. The problem is that it’s very difficult, and I can’t find a transcript off-hand.)

It does, yeah.

Chris, I have no special interest in pushing Galatea on you, and as I’ve said in our earlier conversations, I do not think it represents any particular paragon of accessibility for IF. It’s not something I would use to defend the strengths of the parser or put forward as the pinnacle of IF development.

However, from a purely understanding-how-it-works perspective, it’s not the case that you always get the same text out for the same text in. Galatea does track not only factual knowledge and what you were talking about last, but also mood states and relationship strength with the player. All of these factors determine how she responds when you bring up a particular conversation topic; and in a few circumstances they set her up to make follow-up comments or ask questions of her own.

This may be difficult to detect if you’re struggling with the parser, but here are a bunch of walkthroughs here that lead to different outcomes, along with a couple of cheat commands that expose the numbers under the hood, if you want to see them:

Alabaster also tracks moods, though it does so for more cosmetic effects, as described here: … versation/

Many IF works do this. Underneath there may be a conversational tree or some other programming device. That doesn’t mean that to move through the tree you are only pushing buttons. Read Alabaster, read Blue Lacuna, read Make it Good.

Exemplified by a certain babel fish dispenser?

Hi Chris-- I wrote Blue Lacuna, which a few people above mentioned. There aren’t any transcripts online that I know about, and it might not be what you’re looking for (for the same reasons as Galatea), but here’s one quick example from the first conversation with the main NPC, with some debug info turned on. Most conversation in it isn’t this yes/no based; typically some words in the conversation are highlighted, indicating you can type that word to move the conversation in that direction.

The character keeps track of how he feels about you along three axes, which the system uses to categorize the relationship into one of about twelve archetypes (i.e. “bitter father,” which is positive paternalism, negative affinity, and positive submission). The values and archetypes are used to control which conversations scenes from a library are possible to occur. Each conversation has opportunities to change the values. This is all mostly hand-scripted, but there are also always opportunities to break off the script: walking away from someone mid-conversation or ignoring what they’re saying, for example, can make them temporarily annoyed, which with repeated antisocial behavior becomes a permanent affinity loss. You can also take more complicated actions, like agree to meet someone at a certain time and then not show up, that cause relationship value changes. The hand-authored text is varied with certain mix-ins to do things like insert references to prior conversations, the location/time of day, and also changing things like posture and gestures based on the relationship values (so a paternal Progue might put his hand on your shoulder while speaking, but a negative affinity Progue will cross his arms and look at you disdainfully).

The resolution of the character’s story arc is dependent on where the player ends up positioning the character in this space. The goal was to make the player feel responsible for how the character turned out: a tragic suicide or dramatic change of heart wasn’t predestined, it was brought on by how you treated him throughout the course of the story. There’s some more info about all of this in a paper here.

Of course, this is all very different than the more procedural approach you’re interested in-- but I think it’s a decent example that with sufficient work from an IF author, effects like the ones you want to create procedurally can be invoked. I think of Blue Lacuna now as sort of like the clockwork automaton in Hugo: it’s not a general-purpose machine, but fragile and delicately crafted to do one specific thing reasonably well.

On a side note, we looked at “Balance of Power: 21st Century” in a class I taught last summer at UC Santa Cruz, and it produced some of the best conversations in the course on the intersection between process/simulation and narrative. I was sad to see with the revamp of the StoryTron website you’ve taken it down-- it was doing some really cool stuff, I thought!

I think we might be using different senses of the word ‘interactive’. On one level, all written stories are interactive: the reader must imagine the events of the text themselves and the results of this imaginative interaction with the text are going to differ from person to person. In two more senses, many visual computer games are interactive: you not only get to make choices as to how you interact with the game world, but in several games your actions can actively shape events within the game (see for instance Planescape, Dwarf Fortress, Mass Effect).

Interactive fiction at its best combines all these forms of interactivity: you must creatively engage with a work as a text and as a game, and in many works your actions can direct the course of the narrative. While it’s a different (and more limited) kind of interactivity than that which you’d find in a typical table-top roleplaying game, playing interactive fiction has the capacity to be deeply interactive.

This. I always had a problem with Stron’s online-only status, in part out of a fear of something like this happening. There were some interesting and complex works made in SWAT over the years, and now none of them can be experienced by anyone because there was never an offline interpreter.

At a minimum, I’d say if your definition of the word “interactive” includes passive consumption of books and movies, it’s too broad to draw a meaningful distinction in media classification.

What I’m saying is precisely the opposite: that reading is never passive consumption. In a film the sign and what is being signified are typically the same thing: Sam Spade is a definite man with features strongly resembling Humphrey Bogart. But when you read the description there is a gulf between the sign and what is being signified that the reader must cross themselves. For instance:

From this rather definite description you might conjure an indefinite number of images of Sam, and the image you create is yours. In this way, distinct from visual mediums like film and typical video games, reading is an interactive experience in and of itself.

So I pulled out my copy of Chris Crawford on Interactive Storytelling. It has a two page section (In chapter 20, “distant relatives”) dedicated to interactive fiction, and Chris manages to mention both Crowther and Woods, Infocom and Sierra, and goes on to mention the development beyond two-word parsers and contemporary development systems like TADS and Inform. I feel a little like Arthur Dent (“Harmless, that’s it, one word?”) but let’s face it, there is a limited amount of space in the book, and it’s accurate, as far as I can tell.

This is an actual quote from page 337 in my paperback edition:

Let’s break this down.

Chris defines “interactive” in chapter 2 (and this is the best definition I’ve found):

The brilliant part here is that it defines interactivity as a process but puts the constraints not on the process itself, but on the agents - they need to do the thinking, they don’t just need to exchange thoughts. Which is also very much in line with the rest of the book emphasizing processes over data, verbs over nouns etc. (And Joey, you know I respect you, but here I’m going with Chris: you may be thinking about the book, but it’s not thinking about you). Fiction, I guess, is simply that which is “made up”. And so, interactive fiction is not just a string concatenation of “interactive” and “fiction” but also does live up to be both those things. Well, unless it’s about non-fiction.

But then Chris goes on to say that this is “certainly not storytelling”.

Chris, I’ve read your book and I’m still not sure what that assertion really means, and I can’t say for sure whether or not I agree. I think what this means is not simply what it looks like (that either there is no story, or it’s not being told, or maybe there is a no-telling no-story going on, very zen) but what you mean is that interactive fiction does not constitute what you mean when you say “interactive storytelling”.

And basically this boils down to “IF writers aren’t doing what Chris Crawford does”. Which is pretty much true: there is not that emphasis on modelling intrapersonal relationships, and yes, most of the IF out there is very linear, and a lot of it is puzzle-driven. So, yeah.

But maybe “certainly not storytelling” is a little bold?

The section ends with

This is s not a conclusion, but a prediction. Absence of proof is not proof of absence. And certainly, absence of historical evidence is not in it self a solid ground to predict the future.

It sounds like you’re implying that interactive fiction is somehow fundamentally incompatible with your vision of interactive storytelling, and I dispute that. What would that be? The command line? The text-only interface? Some limitations in the underlying world model? In the authoring systems? No, I don’t buy it. I think it’s just that nobody’s really doing it. It’s hard, and most people are not even trying, they have other goals.

Also, what you say in chapter 3, What Can’t be Part of Interactive Storytelling is not necessarily true for interactive fiction.

The argument is about a scene from the movie Bless the Child (IMDB - I haven’t seen it), where a devil-worshipping cultist (really?) presents a girl with a choice - to leap off a building to prove her faith in God, or reject God and stay on the roof. But…

the girl makes an unexpected move: she tells the cultist “After you.”

I’d say that one of the key features of (parser-based) interactive fiction is exactly that is capable of handling “unexpected” options like this, because the player is just presented with a command prompt, and not with a menu of options - not even with a large but limited number of combinations of actions and items, but just a command prompt, where you can try anything. Yes, of course it’s an illusion of complete freedom to enter anything, but the thing is not really that you can enter anything, but that as a player, it’s not obvious which commands will work, and as an author, you can make any command available. Responding to things the player character says is one them.

Here’s my lesson:

Interactive fiction is a medium that could very well work for creating interactive storytelling, maybe even doing things leading experts say can’t be done, but so far it hasn’t been done.

And finally, Chris, it’s great of you to drop by here. Respect.

I think Blue Lacuna is probably the best example I’ve encountered, though obviously Aaron’s more detailed description above illustrates why.

But I think Chris has a point. The vast majority of IF really is just a non-interactive story with the pace of storytelling set by the player’s ability to solve puzzles. However, I don’t think that means that IF is intrinsically restricted as such, with stories like Blue Lacuna hinting at the possibilities. Perhaps there are still ties to strong, long-standing conventions that end up limiting the tools used to produce IF, making it difficult to properly create the kind of interactive storytelling that Chris envisions. I wouldn’t know. But it’s certainly not impossible to produce such interactive storytelling in an IF medium.

The thing is, if you try and actually create the sort of storytelling I think Chris means, you come up with emergent narratives, emergent gameplay. That’s fine - but not to the extent where you lose the author altogether. “Emergent” is fun and good and awesome without a narrative structure. Lose the structure, lose the author’s agency, put it all in the hands of the player and of the program - and well, a narrative will emerge, but it will never ever be as good or as fulfilling as the narrative an author could deliver. Interacting with Progue is fantastic - but not nearly as satisfying as reaching the end of the story, and arriving at the end of the narrative with a Progue that, although shaped by the player, was always entirely foreseen by the author.

I’ll have more thoughts on this when I play “The Last Express”… the one game that apparently managed to surprise its authors. But even that one doesn’t forego the authorial voice and the narrative.

If I’m rambling about things unrelated to this thread’s purpose, then I must have taken a wrong turn somewhere, and I apologise.

Frankly, I fail to see Chris’ point. If we call any narrative with a beginning, a plot and an end a “story”, and define “interaction” as “changing the development of the story as consequence of the player’s (reader’s …) actions”, IF is certainly “interactive storytelling.”

Methinks Chris arbitrarily restricts his definition so that IF would be excluded from his criteria, contrary to the common sense meaning “interactive storytelling” would have. For example, I have no idea why NPCs should be the touchstone to IS. If a door drops irreversibly shut due to my actions, that certainly alters the course of the story.


Wow, I really unleashed an avalanche of ideas; I doubt that I can adequately address them all, but I’ll take a stab at it. In no particular order:

Definition of Interactivity: Let’s not conflate ‘interactive’ with ‘active’. I can be active upon a brick wall, banging my head against it, calling it dirty names, and asking it the meaning of life, but that doesn’t mean that I’m interacting with it. Interactivity is a two-way street.

Definition of interactive storytelling One correspondent suggests that “changing the story in response to the player’s actions” constitutes an adequate criterion for interactive storytelling. I accept that only in the most minimal sense. If the software dumps some text upon the player, who responds by asking “What the hell does that mean?”, to which the software responds “42”, I suppose that constitutes changing the story in response to the player’s actions, but it falls short of my requirements because it’s bereft of dramatic significance. The ideal I seek – which I acknowledge to be unattainable as yet – is dramatically significant response to the player’s actions. Forcing the player to choose from a skimpy menu of responses precludes dramatically significant responses. If the torturer tells his gagged victim “Just say the word and I’ll stop!”, that’s not a genuine offer.

My book I was especially displeased with the cursory treatment I gave IF in that edition, which is why I am expanding my coverage to a separate chapter. First, I want to make certain that I am giving fair treatment to the medium by including all arguments, and second, I want to take the space to fully explain my complaints. While I’m not disowning the claims made in the first edition, I am acknowledging that they were too cursory for the task.

Story versus storytelling This is a finer point lost when taking quotes out of the context of the book. I make it clear in the book (well, I thought I made it clear), that story is data while storytelling is process. You cannot interact with a story, but you can interact with the process of storytelling. Most IF is little more than story, but the better works have a stronger element of storytelling in them, making them what I would now call “weak interactive storytelling”. More on this later.

Balance of Power 21C and procedural text The system used there is the highest refinement of a system I started developing with the original Balance of Power back in 1984. Believe it or not, it was inspired by the National Enquirer; I realized that their headlines could be generated with random permutations of “flying saucer”, “Elvis”, “reincarnated”, “love nest”, and [currently popular female celebrity]. I continued to develop the system over the years, culminating in the system used in Balance of Power 21C, which uses a scripting language that’s pretty good. I’ll be writing up that system in the second edition of the book. However, I’m also abandoning the entire idea; I have come to the conclusion that it’s not the right path.

my statement that “Interactive Fiction will not lead to interactive storytelling” First off, whether we label it a conclusion, a prediction, or a lesson is insignificant. It’s a proposition I claim to be true; let’s consider its truth and falsehood. The ‘absence of proof’ argument doesn’t much work here, because I’m not claiming absence of proof for support: I’m using what we do know about IF to draw a conclusion. That conclusion is supported by my observations, but I acknowledge that the level of support I offer is weak; hence my decision to expand my coverage of the medium to a separate chapter.

The ‘unexpected option’ using the Bless The Child example The argument here is that IF can indeed provide this capability by making the answer invisible, thereby requiring the player to divine it. This is true, but it brings us back into the classic parser puzzle problems that have, IMAO, bedeviled IF in times past. I accept the point that modern systems with lists of available terms and suggestions have finally driven a stake through the heart of the parser puzzle problem; should we be pulling this monster out of the grave and reviving it?

the argument that IF might still generate decent interactive storytelling. Nothing’s impossible, I agree – but is there not a role for reasoned assessment of the likelihood of success? I was on the losing end of an intense disagreement at Phrontisterion 7 (here’s the report: When I say ‘losing’, I don’t mean to admit that I was wrong – Shiva forbid! Rather, my claim {that reasoned analysis could help us concentrate our efforts on directions that are more likely to succeed} was rejected by everybody else on the grounds that we really don’t know enough to draw such conclusions. My catty response was “Well, perhaps YOU don’t know enough…” HISS! PFFFT! You might find the discussion there worthwhile.

One last concept I’m struggling towards: a visual metaphor for my reservations about IF, based on its deeply boolean nature. I attach the image. Does it communicate the concept?