Writing in Interactive Fiction

I’ve read a fair bit of writing advice for novels and short stories, but I figure that there are certain specific virtues of good IF prose. As someone who is interested in continuing improving my own writing, I’m interested in reading what people have to say on this matter. It seems that most prose in conventional modelled-world IF revolves around describing objects and letting you know where the exits are. Is this a desirable state of affairs?

That seems to me like a question about world-modeling, rather than about prose per se.

To some extent, I think this is like saying “most of the text in drama revolves around people talking to each other.” Traditional IF is a medium that approaches the world through the lens of setting and of objects. That’s what it’s good at, because that’s what it’s built for. And if you’re working in a medium, it’s reasonable to play to its strengths. I do think that it would be good if there was also more parser-IF in which the primary interaction wasn’t through medium-sized dry goods, but for that to become a standard you need standard architectures.

Perhaps I should have phrased my question as:

Conventional writing advice emphasises things like show don’t tell, go easy on the adjectives, brevity is clarity, avoid the passive voice etc. One might question the usefulness of a lot of this kind of advice, but here I’m asking whether any of that advice is useful for writing IF, and whether there is any IF specific writing principles.

(And, of course, I’d like to see more manipulation of “people, people’s voices, rivers, mountains, flames, rainbows,
shadows, pictures on the screen at the cinema, pictures in books or hung on walls, vapours, gases” in interactive fiction.)

The Prose Medium and IF may be of interest.

Show don’t tell: yes. And (this is the standard line in videogame design) interacting is better yet than showing.

Easy on the adjectives, brevity is clarity: yes. The most common problem I see in new IF authors who think of themselves as writerly is overwriting. Learning to chop down a sentence into something that’s elegantly brief, but still does all the work of a longer sentence, is absolutely invaluable. This goes even more for IF writing than for normal prose, because players want to play: they want to extract information from the text, and then they want to get on with interacting. There are prose writers (Proust comes to mind) who, by force of sheer awesomeness, can get away with throwing brevity out of the window; that’s much harder to do in IF.

The principle running counter to this is that players miss stuff; if you need to deliver important information, you need to do so in a way that can’t be missed. Doing so without bludgeoning the player is one of the tougher tricks in IF writing.

Condensed to cliche length:

Showing is better than telling; getting the player to do is better than either. Describe what the protagonist sees, not what the player would see. Better yet, describe what the protagonist thinks he sees. If possible, don’t describe what the protagonist doesn’t think about. (Ideally, it isn’t that important to the player either.)

Don’t flood the player with objects in a single move. (If you have to rig up some mechanism so that certain things aren’t described until the second “look”, or are triggered by a secondary interaction, do it. Better that than the player TLDRing past them.)

I like to use a system of ropes, pulleys, flashing lights and foghorns to get player attention. Then I force important points into their brain with italics, each point juicier and more italicised than the last.

I call it the 'ol one-two!

  • Wade

Also, and I realise that this is really nebulous, reward the player. Early and often. This is a broader principle than just prose, but prose is one of the quickest, most direct ways to do it.

Another common failure, mostly of male authors: remember to ask yourself am I being a sleazeball? when writing about women. It’s pretty easy to write something that you think is romantic, or perfectly innocent, but which actually reads as sketchy as hell. This is good writing advice in general, but it’s particularly important in IF because IF is participatory; it’s ickier to play a part in someone else’s sleazeball world than it is to merely read about it.

Brevity is good. Authorial laziness, not so good. Consider this made-up example:

These types of non-description, found in all too many games, are the result of laziness. The writer can’t be bothered to actually visualize the chair, the bed, or the chandelier and come up with a description that adds character to the scene.

The description of the chair is a vanilla-flavored cop-out.

The description of the bed is worse, because the author knows it’s a cop-out, and is trying to be clever so the player won’t notice.

The description of the chandelier refers to its function, not its appearance. This is rather common in IF. It’s another cop-out. The author doesn’t want to be bothered to write separate objects for the sub-parts of the chandelier, or even to add synonyms (such as “crystal” or “dangling”) and therefore takes the easy way out, which is to tell you what the chandelier does rather than what it looks like.

I’m sure I may have been guilty of all of these at one time or another, so please don’t bother quoting my own games at me. The point is, a player who thinks to examine an in-game object is primarily asking for more detail with which to construct a visual representation of the scene. A description that provides no further visual details makes the model world seem flat and uninteresting.

Since prose is the only thing that IF has to work with (at least practically), it can be argued that this is a prose-issue. The big question is how to reward the player with nothing but prose. Can a description itself be rewarding? Maybe if the writer is able to anticipate how to further the player’s goals with each description…

I’m struggling with this in the game I’m writing right now. Since IF is participatory, the safest thing would be a very objective description with concrete details. However, not only are those kind of descriptions difficult to write, but they’re often boring to read. In the specific case of describing a woman (with the PC being male), is an emotional word like “beautiful” a cop-out? Shouldn’t the reader ideally construct the mental image from the details in the description and figure out that the woman is supposed to be “beautiful”? Here, the principal of showing instead of telling seems to conflict with the principal of conciseness and relevance, with the added problem of the sleazeball factor.

I’m sure it is a cop-out, and it probably has been abused to cover lazy implementation. However, I think it can be an effective, stylish cop-out at times. It can also help to build a certain mood of austere mystery that an objective, detail-heavy description could destroy.

I’d say, unequivocally, yes. A reward doesn’t have to be something that’s useful to the player in game terms: it can be a tasty turn of phrase, a neat little insight into character, the summoning-up of a beautiful or evocative image, a good joke. The main reason that anyone finishes a Robb Sherwin game is because, even though the mechanical aspects of his games are often really frustrating, he hands out this kind of prose reward so often.

Moreover, there’s no such thing as an objective physical description of a person. All the words we have for the body carry some kind of baggage with them. If nothing else, you have to start your description somewhere, which places an emphasis on whatever it is you start with, and…

“Beautiful” on its own is pretty boring, yes. I don’t think that the process is precisely “construct the mental image from the details in the description”, quite so much; the tone of the writing is as or more important than the actual details. Get the tone right, let the writing show how the viewpoint character feels, avoid overspecifying, let the reader’s imagination fill in the details.

Literal physical appearance isn’t always all that important; if looking at the chandelier leads the narrator off on a tangent about the time when the PC’s father installed the chandelier, that can be pretty valuable even if we never really get a direct description of how it looks. And, again, overprecision in description can be dangerous; if someone describes the chandelier in intricate detail, with implemented parts to match, that suggests to the player that the chandelier is Really Important. Which is a problem if it’s just scenery. The terse description is valuable as a signal to the player: this is not the focus of the piece. (Of course, plenty of half-assed authors do this in an unsubtle and lazy way – “This is a fridge like every other fridge. It is pretty boring and you don’t want to waste your time on it” – but if it can be rendered as a gentle nudge, that’s a perfectly valid technique.)

Now, this might just be my very own personnal taste, but when descriptions of objects are involved, I definitely prefer a wisely placed cop-out than a wall of shovel-text that the author put there just for the sake of showing how strong her writting skills are (and failing most of the times!)

I think some of the subtleties of the show and don’t tell principle also apply here. Experienced writters will tell you that showing will be always prefereable than telling, but… showing is meant to emphasize key moments in your tale. If you’re always showing, the most dramatic passages in your novel might not stand out from the rest…

In a similar way, if for every examining action I do in an IF work I get a ton of text, when something actually relevant happens my knee-jerk reaction tends to be the “oh no, just another load of unsubstantial stuff!” and face that potentially key moment with the totally wrong attitude.

I wasn’t advocating over-precision. I was certainly suggesting that one of the purposes of a description property is to stimulate the reader’s ability to visualize the scene. This can be done in a single sentence, and without the need to add subsidiary parts. It’s quite common (and not a cop-out) for subsidiary parts to be subsumed into the vocabulary for the main object:

The armor is not a separate object, and that signals that it isn’t important. All the same, mentioning it is far preferable to something like this:

That’s a non-visual description. It’s a cop-out.

You’re setting up a false dichotomy in order to justify the use of cop-out descriptions. There’s a HUGE gulf between those two extremes.

I would also point out that there are many amazingly talented writers (though not, unfortunately, in IF) who provide lengthy, detailed descriptions of scenes, things, and people but who are NOT including said descriptions “just for the sake of showing how strong [their] writing skills are.” You seem to be implying that you think good writing is always done for the sake of showing off one’s writing ability. That’s certainly not the case.

I repeat: a non-visual description is not necessarily a cop-out, because (as in other text media), the appearance of something is not always very important. Thus:

No physical description whatsoever, but it’s certainly more useful to the player than knowing how tall he is. Now, YMMV: some readers value visualisation more than others, and (since IF is typically about manipulating physical things) visual description is typically more important than it is in straight prose fiction. But ‘non-visual descriptions are always cop-outs’ is flat-out false.

Well, I didn’t say, “Non-visual descriptions are always cop-outs.” All I said was, “One of the purposes of a description property is to stimulate the reader’s ability to visualize the scene.”

I like your description of Marbolc, BTW, and it does lead the player to think about how she will need to interact with Marbolc, which is very much to the point. But I can’t help thinking that if the author were to include that much prose about Marbolc, it would be useful to know that he’s muscular and carrying a battle-axe and a dented shield. Maybe mention his mustache, maybe leave that up to the player’s imagination. No need to drag in extraneous detail. But the muscles, scars, and armament can hardly be irrelevant. Unless, of course, Marbolc has forsworn violence and is now serving as a butler, which is certainly not ruled out by your description. In the latter case, “A white towel is draped over his right forearm” would take care of it nicely.

Obviously, the real difference between regular and IF novels is in the participation. The doing, rather than following.
The prose should adapt to IF keeping in mind one particular thing, imo. In a novel, the writer can move you around and force you to read any detail about a given object, scene or whatever as he wants: he can turn away from the said detail when he wants. In IF, that is not the case and giving false hopes (in terms of things one could do with an object) with glossy text is a mistake. I made this mistake too many times already.
What can save the barracks, though, is that one doesn’t really have to give cop-outs for any single object in a certain room. Just give description (the best description you can) of what you want and leave the rest out of the game.

A living room has a big table, chairs around it, a cupboard, a cat in a corner, windows on three sides of the room, a painting, an armchair, a carpet, a candelabra, appliques on the walls for lighting and at least three-billion insects of every shape and size hiding somewhere. You don’t REALLY WANT to add all of this to the game if it is unnecessary. Just REMOVE the content you are not wiling to let the player interact with. Leave what you want to describe (the gazillion insects is scenery, but can be interesting scenery–for a joke if anything else) and DROP DEAD the rest. No one will think “oh my god how UNFURNISHED this room is”.

What you put in, by the way, should have a description.
In a game I’m lazily trying to produce these months, I’m having an opening scene in a certain place with 2 characters having a conversation. It is all pointless: it is just there to give an unexpected beginning to the story and to propel the atmosphere with a CSI-like final-sentence-before-the-titles. The surroundings are barely described. BUT. Once you wander around a bit with your “sight”, it happens some object lead to others, and others, and others. You can even do unnecessary things like interacting with the ground, or looking in a direction. This doesn’t change the game, it’s just there for eye candy. If you WANT to interact, it rewards you with some bit of (inglisc, atm) prose. If you DON’T WANT you can skip with no problem at all. When “looking west” you discover a fistful of birds circling a rocky formation. You can examine the birds and they are not “just the same old bunch of birds circling a rocky formation”. Had it be that way, there would have been no birds circling no thing.

One final thing about describing and giving false hopes. I’m not good at all in this, but follow me please.
When describing, try to foresee what you want or don’t want to implement. With the accent on DON’T WANT.
A table with “chairs around it” is a sparse description that can lead to something or can lead to nothing. A table “with six chairs around it” is a puzzle. You will have to write code for every single chair 'cause eventually players will try and do something to those chairs when stuck. Also, “a large dining table” can be enough. What did you mentally summon? A table with no chair or a table with [insert number here] chairs around it? A “large portrait on the north wall”? Is there a frame around it? Isn’t it wooden? Is it a panting, right? A dark, brown-hued one, right?

Check this (quoted) passage from a book On Writing by a very talented writer. That’s what I mean.

As Maga once told me, my reviews lack quotes, so there you go.

This is TERRIBLE prose. On behalf of the author, I must say he was writing it trying to think in English not being english himself. That could lead to awkwardness.

First: every other noun has an adjective before it. It gets boring really soon.
Then: The lot of description turns the attention from the only important things: the quartz ledge (which one had to examine to find a very important object) and the exits.
More: The same things were told in the previous room description–the falling ceiling, the magma river and its “effortless, although slow and patient” flowing (ouch!), the “fiery belly” (god, this is awful), the stony walls.
Even more: There is no sign of an exit back from this room, you must remember where you came from to get back there.
Finally: there is a simile in every sentence.

The same could have been done in one third of the text.

Too bad, I couldn’t find a “good prose” to show by contrast in the same game. Again, forgive the author :slight_smile:

That description does seem useful and characterizes the NPC nicely, but I think the description fails to justify offering so much information. I do think descriptions are lacking without some mention of physical appearance. In Marbolc’s description, it would be best if the PC’s background knowledge were associated with physical appearance. Maybe the PC sees the scar along Marbolc’s face left by one of those Sacari arrows. Simply revealing backstory about characters or objects in the descriptions seems like a form of telling to me.

After all, the command players use to reveal descriptions of objects and NPCs is EXAMINE. An examination of something can certainly trigger associated thoughts and memories, but those thoughts and memories should be grounded in the physical appearance.

You had a lot of great phrases that created powerful imagery; the problem was that the descriptions taken as a whole often carried too much unnecessary wordplay. As you said, there were too many similies, too much qualifying of adjectives and adverbs. There were also vague phrases that didn’t help to form a specific picture, such as “Something like steps” in the passage you quoted. However, all things considered, the prose in Andromeda Awakening was effective for me, at least. I don’t know if it was effective for anyone else, but the activate imagery on the whole is definitely concrete and useful.

That is the usual convention, yes. It’s a standard technique with which every IF author should be familiar.

But it’s important to realise that EXAMINE is enormously central to traditional parser IF; it’s so omnipresent that it’s near-invisible. Experienced players use it, I estimate, as a quarter to a third of all well-formed commands. No other verb comes remotely close. EXAMINE is so big that it has swamped the functions of other important verbs. It covers a wide range of looking-at – from close scrutiny taking several minutes, to a passing glance that takes no time at all – but it’s also the reason that virtually no game uses >THINK ABOUT, precisely because we expect EXAMINE to do that for us. If you wanted a command that more accurately specified what >EXAMINE really does, it would be >TELL ME ABOUT.

So, depending on the game’s focus, I think it’s perfectly legitimate to take >EXAMINE as really meaning >THINK ABOUT, or >CHECK STATUS OF, or >WHAT IS. The convention of using descriptions of an object’s visual appearance as clues to its significance is a well-established and successful one, , and many of the most beautiful works in IF are the ones that really know how to work that technique. But it’s mostly the result of the convention that IF is centrally about particular things – the exploration of space, the manipulation of medium-sized dry goods to solve puzzles. If you’re doing something with different concerns, the primacy of visual descriptions may not be the best principle. It may not even be a good one.