In the simulationist approach to writing fiction, rules are defined and then outcomes are decided upon. These rules can be fundamental rules far removed from human/story scales (“there is no up quark”) to something that can be intuitively grasped and whose first order effects are somewhat obvious (“people die of an aneurysm when they reach the age of 50”). Once the rules have been established and the outcomes determined, the simulationist writer then goes looking for a story to tell within the set of those outcomes.
In the narrativist approach to writing fiction, outcomes are defined and then rules are created to match those outcomes. These outcomes are almost always on the high level and almost always deal with either narrative convention (“everyone in the group should have a unique role”) or literary analysis (“I want to write about depression and how it affects people without literally talking about the mental disorder”). Once the narrative has been established, the rules are created largely in order to support that narrative, and rarely extend beyond what is necessary to create that support.
I don’t think that it’s particularly useful to focus on the failure modes of these approaches, except insofar as it’s important for writers to avoid them. From both sides, I feel that it’s uncharitable to define either approach by its failures, especially where those failures can (in my opinion) be avoided. When I say “failure mode”, I don’t mean “this is why it fails”, I mean “if it fails, this is one of the likely ways that it does”.
The primary failure mode for simulationism is a story while sacrifices enjoyment, thematic cohesion, or some other narrative element in order to adhere to the rules as they have been set out or details as they have been pinned down. If you ever see a writer claim that something boring or anticlimactic happened that way “because that’s how it would have happened”, that is, to me, a failure of approaching writing as an exploration of the outcomes that result from a set of rules.
In part, I think this happens because the simulation is not done “in full”; a writer might start writing in order to see what’s going to happen to the plot, and not realize that the plot isn’t satisfying until they’ve invested dozens or hundreds of hours into that plot. At that point, costs have been sunk, and the story has to either be dropped, have an unsatisfying ending, or go through a rewriting process to produce a different outcome.
The secondary, and more easily avoided pitfall of simulationism is giving extraneous details. There are a few forms that this takes. The first, and easiest to correct, is simple over-expositing, giving description or exposition that does not advance the plot or characters or otherwise have an impact on the story, or the reader’s enjoyment of the story. There are lots of ways to do exposition and description wrong, and if you’re focused on the world as it exists rather than the story, it seems to me that you’re slightly more likely to do it wrong — but this is something that can be fixed by gaining skill as a writer.
The other half of “extraneous detail” is a little bit more insidious. Removing details from the reader’s view while still keeping them as part of the whole is fine, until those details actually do impact the plot and characters … and sometimes, that added complexity will be to the detriment of the work. If characters are informed by an invented religion, the grammar of an invented language, the mores of an invented culture, the canon of invented stories … this can all get to be very complex, and not so easily removed as simply not mentioning it, because it does inform the character, and the background is necessary to make sense of what the character does and why. Unfortunately, this can be at odds with creating a satisfying narrative.
The primary failure mode for narrativism is the plot hole. Because the writer is making up the rules to fit the outcome they desire, it’s possible to create rules that are insufficient to produce the outcome, or which produce outcomes which should logically occur but don’t, or where the rules are unintentionally contradictory. There are also less egregious plot holes of omission, where a reader might fill in the gaps on their own, but are nevertheless annoying for readers.
It can be difficult to create rules to fit outcomes, especially if you’re concerned with ensuring that those outcomes are the only outcomes. If the narrative says “X admits to something she didn’t want to admit to” and the rule that you create to allow that is “there exists a truth serum”, then that has a lot of undesirable knock-on effects. If you’re a narrativist, you might be tempted to not do the work to see those knock-on effects to their conclusions, especially because you might be only concerned with that one particular outcome. This is one of the ways that plot holes happen.
Less seriously, some rules are pretty and some rules are not. Even if you make up rules to fit the narrative after the fact, and these rules properly “create” the plot without contradicting themselves or creating undesirable impacts on the plot, they might be ugly rules, too transparently obvious in what they’re trying to set up, too beneficial to the protagonist, and over-fitted to the story. Sometimes explaining those rules would, by necessity, result in an infodump or other awkward exposition, and narrativists will sometimes leave them out as a result, as it’s more aesthetically pleasing to offer no explanation. (This is another way that plot holes happen.)
There are a whole host of “bad writing” tropes that I think happen because of the dictates of narrative. “Idiot Ball” is one of the more common ones that gets called out, but that’s just a subset of characters acting in a way contrary to what’s been established because they need to act that way in order to advance the plot. Flat characters are another, and happen because the character is only filling their role within the plot, and doesn’t exist beyond that role. These failures are failures of writing, and not endemic to narrativism, but are ones that a simulationist would probably not fall into.
Approaches in Practice
With perhaps a few rare exceptions, I don’t think that anyone actually follows “pure” narrativism or “pure” simulationism. Instead, people use both approaches in creating a story, going back and forth between the two in order to provide a balance between them; a story should naturally follow from its premise, and that story should do everything right in order to satisfy the reader. The primary point of contention between the two camps, such as they exist, is in which approach should be the primary one, and beyond matters of personal preference, this comes down to what you get out of the time and effort you spend.
The traditional simulation first workflow is to first ask a question, big or small, and then get a general feel for what the answers to that question look like. At the very extreme end, that means starting with alterations to particle physics, while the somewhat more reasonable extreme end starts with variations on land masses and works forward at whatever level is feasible to create a world. The weakest version of simulation first would simply be to start with a cast of characters, each fully realized before they start interacting, because that starts as close to the drama as possible.
The problem then becomes one of “finding the story”. If you have a world, and that world has people, there are stories in it, and part of the joy of simulation is that the stories have a wellspring, which can lead to finding stories that are relatively unique and more grounded in their own internal reality. The question is, given the premise, who do I expect to be in conflict? What are the interesting conflicts that arise? Much of this is an iterative process; you explore the ramifications of the rules you’ve set up, then see whether those ramifications result in anything interesting, then if they don’t create something to latch onto, explore more ramifications of the rules, or tweaks to the rules to result in different effects, and so on.
Eventually, the story is found, and then the writing process actually begins, with some amount of simulationism and some amount of narrativism.
The big risk of making a narrative first is that you’ll run into problems with your rules, facts, or precursors somewhere down the road. Sometimes a narrativist will start with a specific structure (the monomyth being a classic, as is the three-act structure found in most Hollywood movies), but other times there will be specific scenes that the writer thinks are cool or impactful, and wants to hit.
Usually writers have some instinctive ability to avoid direct contradictions. All but the worst writers in the world won’t craft a story in which the main character is established as a doctor at the start of the book and referenced as a fisherman at the end, or have a child whose age is first referenced as seven and then four, or other basic continuity errors. Where the problems for the narrativist usually crop up are the first order effects, which is when the switch to simulation usually needs to happen, at least briefly.
Even outside the realm of speculative fiction, most writers will find themselves looking things up in order to find a workable scenario for their planned plot beat. If the writer is lazy, they might just go for technobabble, which will fail if seen by an expert but otherwise probably pass the reader by, but a diligent writer with a narrativist bent will take some time to figure out something that’s plausible. One of the keys to plausibility is in saying as little as possible though, so if the plot beat is “and then he cracks open the safe” a narrativist would probably learn just enough about the safe-cracking process that they can get through it with a handful of lines.
At some point in the process, the narrativist writer will come to a place where the rules or facts don’t line up with the plot that he has planned (or, if he’s not planning out ahead of time, don’t line up with any satisfying plot). At that point, it’s usually time for some editing; since the writer has control over the past, they can insert an early reference to whatever mechanism is going to unstick the plot, and it then retroactively becomes a Chekov’s gun.
Where simulation comes into play for the narrativist is that they’re still constrained by what’s shown on screen, meaning that if it’s established that a car can’t go more than 60 mph, it can’t be shown going 70 mph later on unless there’s some sort of justification for that. I think that in a work that went into enough detail, or went for long enough, eventually the writer would approach a point where they began approaching simulationism. If everything is pinned down, then the story can only follow from its consequences, and wiggle room disappears with every new detail that’s put into place.
Points of Contention
The biggest point that the (more) narrativist camp makes against the simulationist camp is that simulationists spend a lot of time on things which won’t end up making it into the story. The strawman extreme of this would be something like figuring out a name for every person who attends a high school or writing down the shoe size and blood type of all the principle characters. I doubt that many people have actually gone that far, mostly because of the tedium involved, but the argument as I’ve seen it is that a simulationist will inevitably be producing far more than the story calls for. (To some extent, the fact that no one actually does figure out irrelevant things ahead of time is evidence that we’re already engaging in pruning away the things that we know aren’t going to be story-relevant.)
Rules can be created at a low level, with emergent properties at the story level, or at a high level, close to where the story takes place. An example of a low level rule would be that vampire skin contains a molecular structure that’s hyper-sensitive to ultraviolet light and begins a chain reaction resulting in combustion on sufficient exposure, while an example of a higher level rule is that vampires can’t go out in sunlight. These rules might have the same effect on the actual plot, or they might diverge in some significant way — such as the fact that sensitivity to UV light would theoretically allow the creation of an anti-vampire UV gun. However, if the results of both those rules are the same (because no one in-story knows them, or because they aren’t discoverable, or because they’re discoverable but don’t impact the plot or characters), then I think there’s a strong argument that the higher level rule is preferable so as to spend less effort, avoid confusing the issue within the narrative, and to strengthen the link between narrative rules and in-universe rules.
I think a fair number of arguments between narrativists and simulationists are a result of disagreement over what level the rules should be constructed on, in part because to the simulationist view, there’s not really a “should” there; the rules are constructed first, and the story comes after, so it doesn’t matter whether the rules as viewed by the characters and expressed to the reader are base rules, first order effects, second order effects, etc. To the narrativist, this likely points back in the direction of the wasted effort argument.
Evidence of Process
With regard to the failure modes above, there is a somewhat common claim by the simulationist side that not pinning things down, or creating rules after the fact to explain desired outcomes, will be visible to the reader. In the weaker version of the claim, this goes back toward the argument of effort; at a certain point, patching holes will require so much effort that the narrativist will throw up their hands and declare that the holes don’t matter, or that they’re necessary to the story, or some other such thing. Alternately, these patches will be visible to the reader as such. In the stronger version of the claim, no narrativist process can possibly cover up for the fact that the narrative was crafted as a narrative, rather than being a narrative plucked from the world. Most of the disagreement on this one is, I think, hypothetical; if someone pointed out to me that it was clear where narrative had paved the way, I would probably say that was just an example of an author not really caring to cover up their work — though that might still be an argument against pursuing narrative-first.
Narrative-first and simulation-first approaches produce different works with different qualia. That statement might be a little debatable, because in theory the processes might arrive at the exact same words on the page … but I think that’s unlikely, and that each approach leaves its own imprint. A lot of simulationists simply like the features that they find most often in simulationist stories, like a focus on processes, second-order effects, miscellania, deep dives on complicated topics, etc. A lot of narrativists, by contrast, might take pleasure from thematic resonance, recurring motifs, parallel narrative structures, etc. This shouldn’t be a point of contention, merely a question of differing tastes, but some people get bent out of shape about different tastes even when there’s no philosophical/practical backing to those tastes.
Most works of fiction are created through narrativist processes , therefore simulationist works are rare, therefore they are valuable, even if they require more effort on the part of the author, or even if they only produce a different (not better) qualia when read. This largely comes down to definitions of “best” and what an author is actually trying to do when writing a story.
These will mostly be using works that I wrote, because I’m intimately familiar with the writing process for them. Might add more later, but writing about writing sometimes grates on me.
Shadows of the Limelight
I believe the process went something like this:
I like superhero stories, but given powers, there’s usually absolutely no point in actually putting on a costume to go out and fight crime, and especially no point in doing so to commit crime, and there are so many knock-on effects from powers that you’d spend a mammoth amount of time creating a consistent world just to focus on what would probably be the least interesting aspect of that world.
What must be true for costumed superheroes to exist? Here I was working backward, trying to find a sufficiently clever or simple rule to explain the outcomes. Eventually, I settled on “fame gives you superpowers”, which seemed like an interesting starting point.
The basic rule of “fame gives you superpowers” presents a few natural questions, like “what is fame” and “what are superpowers”. I (mentally) tried a few things on for size, which is where a lot of the back-and-forth of worldbuilding comes from, most of it never committed to paper. Much of this process is either thinking about interesting outcomes (e.g. “there are feedback loops” or “people wear ornate costumes” or “there are large, set-piece battles”) and then working backward, or thinking about the effects of a rule (e.g. “all politicians are famous, therefore they all have superpowers” or “the face of warfare is radically changed by the prominence of powered individuals”) and building out from there, which sometimes results in a rule getting adjusted or amended in order to provide more interesting outcomes.
Note: This was originally written two years ago, and hasn’t been substantially revised since then, but I thought I would put it on the blog rather than continuing to link to it as a Google Doc.
Narrativism vs. Simulationism