I do not have a good definition of IF. This is not just because the community has many conflicting opinions about what IF is. (Although the community does). It's because we have a lot of conflicting opinions of what the important aspects of IF are. I'm going to give my opinion about that. I don't expect to convince everybody -- but hopefully it'll strike a chord with people, and spark some more interesting discussion.
A good definition does not cover every single weird case. (Platypuses are mammals.) It covers most of the important territory, and makes people say "Yeah, that's basically it." The weirder and more hypothetical the case, the less a definition should worry about it.
In fact, it's worse than platypuses. Platypuses are an odd example of mammals, but at least we can easily tell a platypus when we see one -- and a platypus is unquestionably a mammal.
But when I start thinking "Is this game IF? What about that game?" I immediately run into cases of "Yes, mostly." Or "Not much, but it has some IF elements." Or "Parts of the design are handled in an IF-like way, but others aren't."
It's almost completely useless to categorize games as IF or not-IF. Instead, I want to characterize aspects of games. Some games have more IF elements; some have fewer. In some games, the majority of the gameplay is IF. In others, only a minority. We will see a lot of continuums in this essay.
(You might say that any game which is at least 50% IF counts as being really IF, but of course that's always going to be a fuzzy and subjective measurement. I don't care whether a particular example is at the 49% level or 51%; I want to talk about the characteristics that 49% or 51% of it are like.)
This is what I think interactive fiction is:
A program which reveals a story (or related stories), created by an author (or authors), to a player (or players); such that the range of action available to the player is only partially known to him, and must be understood in terms of the story world; and such that the majority of important results of the player's actions are unique results, specifically created by the author to support that part of the story which the player is experiencing.
IF happens between a computer and a player. Live role-playing (with a human game master) is just a different category of thing, and I'm not including it in this definition.
(A human may undertake to game-master an IF work to another human -- I've done this with the Enchanter trilogy. If he sticks strictly to the program, he's just taking the role of a computer in executing that program. If he allows the player as much freedom as is customary in live role-playing, he's doing live role-playing. In between is a continuum.)
The difference between a single-story IF work and a work with multiple endings (or multiple stories in some other sense) is an interesting division within IF. But from the standpoint of defining IF, we take no notice of this distinction.
A totally open system, in which all the story comes from the player and not from an author, is not IF. It's a toy which you can make up stories while playing with.
A system where the program creates the story... is hypothetical. (It's probably also true artificial intelligence.) Erasmatron notwithstanding, we don't have it. I'm not going to distort my definition to include examples which don't exist (and have no prospect of existing soon).
This is one of my two keynotes of IF. In a Colossal-Cave-style text parser, you know a lot of the verbs which are available, and (when you read the room descriptions) you know a lot of the nouns which are available. But you don't know all of each, and you can't discover them in a blind or mechanical way (treating the program in front of you as a program). You must think in terms of the story world: what do I have available to me? How do these things behave? If I were really there, what would I do?
In a Myst-style graphical game, your "verbs" are completely known to you: you can click. (Perhaps there are a couple more, like "open the inventory window" and "drag one object onto another", but they're all in the manual.) But the range of "nouns" available to you is everything depicted on the screen, and it's not obvious which are interactive -- which are within your range of action. You must study your screen and make sense of it as an environment. Thus, it satisfies my requirement, although in a completely different way than Colossal Cave.
"Hunt the Wumpus", on the other hand, only has two commands: move to an adjacent room, or shoot into an adjacent room. (The adjacent rooms are listed for you.) The range of action is entirely known -- even though the results of an action may not be known until you try it.
This requirement encapsulates my impulse to categorize choose-your-own-adventure games, or other purely-menu-driven systems, as "not-IF". My problem with CYOA games (or books) is that the range of action is never uncertain -- you can progress mechanically, by trying every menu possibility. In fact, this is what usually happens when I play a CYOA game. There is never any sense of thinking of something new to try.
(Of course, one could imagine a CYOA which has so many options on each page that you can't try them all, or even read down the list looking for ideas. I reply: that's absurdly hypothetical. A CYOA book with a million options per page, like an AI story generator, is an imaginary trick -- at least this decade. Of course my intuitive definition of IF doesn't account for it. If it ever becomes real, my categorization of the world will change, in the face of new evidence.)
(Contrariwise, you might say: aha, the range of action for Colossal Cave is perfectly well known -- you type on a keyboard and hit Enter. I reply, don't be a silly person. That does not characterize the range of action in the way that the player thinks about it... unless the player is entirely unfamiliar with IF.)
(And such a novice player will in fact not succeed in playing the IF game. Thus my point.)
This is my second keynote of IF. It distinguishes IF from a CRPG (computer role-playing game).
In an IF game, many of your actions cause "simulationist" results -- according to wide-spread, easily-learnable rules of the model world. In fact these results are probably a requirement of making playable IF. You type "get rock"; the rock moves to your inventory. You type "north"; you move to an adjacent room to the north.
But these common responses in the simulated world are not the important part of the IF work. They are tools you use to get to the important parts. Your attention is focussed on new text or images, new scenery. You've solved the puzzle not when you make the block slide according to the rules, but when the block slides to its goal and you read the author's hand-written description of the door unlocking.
The characteristic of a CRPG is that most of what you do is in the results-of-rules sector, not the hand-written-results sector. There will certainly be some of both. But if you spend most of your time thinking "How do I marshall my spells and weapons to kill the next monster? How do I get enough gold or experience to get more spells and weapons?" then you're playing a CRPG. The hand-written plot scenes (and the specially-coded puzzles) are interludes and payoffs -- they're not the point of the game.
I don't mean to disparage what has been called "simulationist IF". This is indeed a division of IF -- a game in which the rules of the world are more complex, and you spend more time employing them to get to the important outcomes. We certainly see a continuum between IF and CRPGs. A very heavily simulationist IF game would approach being a CRPG, just as a CRPG which was mostly hand-coded puzzles would approach being IF.
(Furthermore, any game can have sub-parts which are of a different form. Zork 1 and Beyond Zork had small CRPG interludes, in which you fight monsters in a strongly rule-based way. Note that these are generally considered jarring scenes -- not fitting well with the rest of the game. This is more true of Beyond Zork, whose fight model had more rule complexity than that of Zork 1. Thus my point.)
Just to be clear, a Myst-style graphical game certainly fulfils this requirement as well. Ask anyone who's worked on such a game. Most of the implementation work goes into hand-designing rooms to explore; hand-designing objects; hand-designing animations and sounds with which the game will respond to the player's actions. Just about everything is a one-of-a-kind.
(A CRPG has lots of hand-designed maps and rooms -- but that level of design doesn't work into the gameplay very much. As a player, you treat most of those maps as "floor" and "wall", which are general categories which behave according to certain rules. Then there are doors (also behaving according to consistent rules), water (ditto), etc, etc. Now, this may be a complex and interesting set of rules -- it may in fact provide for a lot of good gameplay, as you figure out how to use the environment to support your goals in the game. I'm not saying that's an inferior kind of game. I'm saying that it's a CRPG kind of game, as opposed to an IF-style game.)
Up above I said, "Such a novice player will in fact not succeed in playing the IF game." This produced a bit of discussion about the purely relativist view, where you don't say "That game is IF", but rather "That game is IF to this particular player."
It's certainly true that my criteria are subjective, and different players will judge them differently. In fact, they may be different for different players. You may know none, all, or some of the range of action of a given game. Does that make it not-IF or IF for you? Well, I guess so. But I don't want to make a big deal out of that. I'm perfectly willing to take the implied "average IF audience" as my standard. That's what I do when I design a game, after all.
I also got a lot of comments on what the "important" parts of a game are. In an action game, for example, you might fight your way past two dozen demons in order to reach a switch; flipping the switch lowers a drawbridge, and that lets you into a castle.
Lowering a drawbridge is a unique result, in my sense. Flipping a switch to do that is, I would say, an IF element in the game. And getting to the castle is also a unique result -- you get to explore the castle. Perhaps you even get a cut-scene when you enter, which is scripted and animated for that particular point in the story.
But are these the important parts of the game? Not in the sense I intend. Entering the castle is your motivation; flipping the switch is a major goal of the scene, and the motivation for killing the demons. However, in the generic action game, killing the demons is what you're actually doing when you play. That's where the challenge of the game is; that's how you spend most of your time. The game mechanics have lots of rules to manage different weapons, different kinds of armor, health and injury. You've got dozens of controls devoted to fighting. You've got one control devoted to "flip the switch or button or lever or whatever."
This is not insignificant.
Of course, more recent action games do spend more time on environment manipulation, and environment puzzles. They make you think more about accomplishing scripted tasks in the game world, and give you more interesting hand-designed results. The action game genre, in other words, is incorporating more and more IF elements. (As are other game genres.)
Some of the transitional text-with-graphics adventures (for example, Spellcasting 101 and sequels) had menus for choosing verbs and objects -- in addition to the traditional text parser. That certainly seems to violate my criterion about "partially known range of action." Do these games count as IF?
I'd say that interface does weaken the games as IF, but probably not by much. The IF element is diluted exactly to the extent that the player can try every combination of verb and object. If there are enough options, it becomes impractical to deal with the interface in a brute-force mechanical way -- you mostly have to consider the range of action in terms of the game world. So mostly, the game plays as IF.
If, in some particular situation, you stare at the menu and say "Aha! I see a command the game accepts which will help me!" then that's a glitch in the IF experience. That's the case where the menu system becomes a non-IF element.
In Spellcasting 101, that almost never happened, so I have no qualms about categorizing that game as IF.
I've been pretty clear about categorizing choose-your-own-adventure systems as "not-IF". This is kind of harsh, I admit. They're obviously closely related to what I'm calling "IF". Does it serve any purpose to draw the boundary to exclude them?
(I am, of course, speaking of the generic flat-out 100% CYOA game. The CYOA genre, like all other game genres, has evolved and hybridized over time. Some CYOA-based games incorporate IF elements -- inventory manipulation, keyword dialogue systems, etc. In that, they become more IF-like. But I want to discuss the archetypical categories, not specific games.)
I do see a boundary between the two groups, and the boundary is precisely this criterion of "partially-known range of action". And this criterion seems important to me. It was the great innovation of Colossal Cave, and -- differently implemented -- the great strength of Myst as well. These are the sorts of games we call "adventures"; I am fascinated by that structural similarity; I certainly can't dismiss it. And CYOA games distinctively lack that particular feature. It seems to me that this is why I am unsatisfied with them. So: whether you call them "IF" is a matter of terminology, but I think that I am describing an important category, and CYOAs are outside it.
Maybe I should be inventing a new term -- "dynamic-range-of-action games", maybe. Only that's lame and nobody would use it. So I won't bother.
I'll probably wind up using the term "interactive fiction" inconsistently -- sometimes in a general sense which includes CYOAs, sometimes specifically as defined in this essay. But hell -- I already use the term inconsistently. (Sometimes I use it solely for text-parser games; sometimes I including graphical games as IF.)
So what the heck.
Updated July 8, 2002.