Petter Häggholm’s blog

It’s been observed before, notably by Brandon Sanderson in the context of magic systems but probably before then, that it’s better to worldbuild in depth than breadth (as an old programmer will naturally phrase it). I would argue that this applies not just to the magic system (though that may be important in a fantasy story), but to worldbuilding as a whole.

For one thing, of course, you can only go so broad. At some point, you end up with a character template asking so many questions of each minor character, and location sheets with so many details of geography and botany and ecology and textile industry and who knows what, that you cannot fit it onto a page without the reader mistaking it for an encyclopædia and putting it down. For another, the author’s research and development time is no more infinite than their page count.

But I think the most important reason is that worldbuilding details that don’t go into depth are mere set dressing—cosmetic elements—‘reskins’, if you will, of the familiar world. Members of this fantasy race all have tattoo-like patterns on their skin, shifting as they age? Okay, neat, but so what? It’s only interesting (rather than just nifty) if it has consequences.

As a trivial example, a story I’m working on has a society heavily influenced by an ancestral, martial honour culture (I had the old Norse slightly in mind, among others), nowadays to a significant extent encoded in law. This means there are laws around things like what crime entitles someone to challenge you to a duel you can’t licitly decline, what happens if you decline anyway, and so forth. Cool (I think)…but so what? In a sense, I’ve done little more than provide a historical justification for the legal code. That’s good for flavour, but it’s still just cosmetic.

Unless you think more deeply about it. I find that most of my productive worldbuilding comes not from inventing details, but analysing them. What would it mean to enshrine a martial, duelling honour code into law? Well, the two ideas—honour code and legal code—differ in concept. An honour code is primarily about intent, whereas a legal code is about specifics. Inasfar as an honour code has specifics, they are likely illustrations of broader principles, whereas a legal code consists of specifics.

But wait! This has implications (which is precisely what I want). It strongly suggests—I think plausibility demands—that the two can never match completely, honour and law. If they did, everything would be extremely smooth: People would be guilty of crimes precisely when people thought they deserved it. But life is rarely that simple, and if there’s a mismatch between the honour code as the spirit of the law on one hand, and the letter of the law on another, what then? Well…the sanctions for violating a legal code are legal sanctions; for an honour code, social sanctions. This carves out two areas of mismatch: Maybe some things are illegal but not considered dishonourable—for example, the ‘justifiable’ murder of someone who harms a loved one might be punished legally but socially approved. On the other hand, some behaviours might be legal, even to the degree that the honour code is legally enshrined, but skirt the boundaries of the law. Now you might be completely immune to legal sanction, and yet suffer all the more social sanction for it—for example, if you refuse challenges that you are legally entitled to reject, you’ll eventually become seen as a coward. I’m sure you can think of other consequences. For example, I have a character who only ever refers to another character by an extremely insulting nickname that basically means ‘cowardly rules lawyer’ in this context—that is, cultural concepts, categories, and labels arise to address the tensions.

This is much better. The honour code origin of the legal system is no longer just cosmetic—though it still does the job of adding flavour no less than it did at first—but also a potential source of both internal and external moral conflict. It also adds cohesiveness, where things in the story aren’t just one damned thing after another, but clearly and comprehensibly connected causally: It allows the reader to feel that it makes sense that B happens because, on reflection, that does logically follow from A.

Of course, I don’t think there’s anything wrong with a fantastic setting having aspects that are purely cosmetic in this sense. Probably most details one mentions will be. However, I think that the details that are explored in more depth are vastly more interesting and memorable; and I think that a setting that doesn’t have a decent range of ideas explored in greater depth will tend to feel very superficial. A story that reads like you could search and replace “wizard” with “gunman”, “lightning bolt” with “.303 bullet”, and “fireball” with “hand grenade” is ipso facto a story that has no need to be in the fantasy genre, and accomplishes nothing with it. If the story functionally might as well be set in the real world, then why not go easy on the reader’s suspension of disbelief budget and set it in the real world?

I realised the other day that I had been thinking all wrong about the function of the narrator in delivering description to the reader. Quite probably influenced by role-playing game manuals telling me that the game master serves as the players’ eyes and ears, I thought that the author of a narrative was supposed to deliver, transliterated into prose form, the viewpoint character’s sense impressions: what they see, hear, smell, and so forth.

I now think that is all wrong. Certainly it takes a text away from how you would normally tell a story, and it isn’t even how people think about the stories of their own lives, in the real world! No, if I recall an event—whether to myself or to somebody else—I am recalling a few salient details. For example, I ordered lunch from DoorDash yesterday. Drivers come in many forms—male (mostly), female (occasionally); short and tall; white, East Asian, Middle Eastern, occasionally something else. Yesterday’s delivery driver was a black man, and he was very tall, almost a head taller than me. Those details stuck out and lodged in my memory, probably because (though I’m not tall) few people are that much taller than me, and because relatively few delivery drivers here are black. So if I were to write a story about my life yesterday, and if for some reason it is relevant to describe the delivery driver, then those and only those are the details I should mention. It would be unnatural to describe the guy in more detail than I (realistically, or in this case: actually) remember, and whatever I do not remember clearly did not strike me as a very salient detail at the time.

Physical descriptions in prose, at least when told from a viewpoint character’s perspective, should reflect their recollection of physical features, not the physical features.

I suspect that a useful exercise in figuring this out would just be to iteratively shorten a description. One needn't do immediate violence to one's darlings. Imagine having a prose description and wanting to summarise it for some reason—for a blog post title, a Bluesky post, or an email subject. Whatever you find yourself striving hardest to preserve, the things you fight hardest to keep out of the “&c.” at the end—those are probably the most salient details.

A framing narrative imposes narrative duties and obligation on its writer. For example, we all ‘know’ that The Hobbit, The Lord of the Rings, and The Silmarillion were translated from the Red Book of Westmarch, which was chiefly written by Bilbo and Frodo Baggins, expanded by Samwise Gamgee, and amended by his descendants, the Fairbairns of Westmarch.

But I would argue that many writers implicitly adopt a metanarrative identity even when they do not provide an explicit framing narrative. After all, a writer is a craftsman of prose—responsible for choosing their words with intention—and if an impression is clearly conveyed, does it really matter whether it is conveyed through explicit statements or via contextual clues?¹ For instance, suppose the author of a story set in mediæval times (or fantasy pseudo-mediæval) goes out of their way to avoid terminology that rely on modern knowledge in favour of circumlocutions, using descriptive language rather than words like “electricity”. The effect is one of a storyteller limited to knowledge of the world, presumably because they are situated in it. Albeit by implication rather than explicit statement, hasn’t this author taken on as much of a narrative poise as Tolkien with the Red Book? An anonymous and unknown poise, to be sure, but that’s just as true of many real-world books (we don’t know who wrote almost any of the books in the Bible, for example, yet many have distinctive voices).

By implication, then, the author is logically limited to the knowledge of a narrative world that is available within that world. That doesn't mean they can’t break the convention from time to time—we may or may not like it, but when Tolkien compares the firework dragon to a locomotive in The Fellowship of the Ring, we read it as an editorial or translation device: even if Bilbo wrote the original, the locomotive part was clearly the editor/translator, Tolkien, using imagery familiar to us—himself and the readers. But it does imply that the a author is less free to break the illusion on the other side of the divide between worlds: if they tell you things about the invented world that the anonymous narrator they pose as could not have known, it breaks the illusion. They may be a third-party ‘omniscient’ narrator, but such a narrator of a real-world story is only inferred to be ‘omniscient’ about the story, not about our actual world!

It gets even worse when characters break the same constraints, which is terribly common. For example, consider an uncharitable reading of Brandon Sanderson’s Mistborn trilogy.² The magic system is—I would be remiss if I did not use the word “Newtonian”. Magically metabolising one metal will let you magnetically attract things; another, push them away. One will make you strong and tough, another more sensitive and hence better able to see and hear. —And so on. It appeals directly to my materialist heart;³ this is not a problem. What is a bit of a problem is the fact that I understand it so well. Why do I know the logic of how magic works in Sanderson’s books? Well, obviously it’s because he tells you…quite explicitly, repeatedly, and at length.⁴ Although there are some missing facts in the characters’ understanding of the magic system (“allomancy”), they seem to be pretty clear on the system, and certainly the narrator knows everything.

Let us pause to think how weird that is. (Leave the specifics of Mistborn behind—the very ‘hard’, Newtonian feel to the magic makes it a good, vivid picture, but I’m not here to harp on it.) Your average fantasy world is vaguely mediæval, usually a sort of quasi-‘high’-Middle-Ages thing, with the occasional bit of Early Mediæval (like Deverry) or pre-Industrial (Mistborn, kind of; Discworld, some books). Imagine a writer in that setting writing a story from the poise of an omniscient narrator. Think how they would describe, not magical forces, but physical ones. Not mythical beasts, just animals in faraway foreign lands. Not curses and blessings, but ordinary diseases. Think how poorly understood all these phenomena were in the year 500, or 1000, or 1750. People have always had some notion of things, but even that may be flawed in ways difficult to imagine (consider physics going from Aristotelian to Newtonian, then Einstenian, then quantum). No one could have guessed that the hardness of a rock and the flash of lightning are both fundamentally down to electricity; and if here the same effects turn out to have a unifying cause, in other cases they may inappropriately group things—think how many diseases lurk behind the word “fever”. We may scoff at mediæval writers imagining griffins or unicorns, but to someone who would never come within a thousand miles of Africa, is a unicorn more absurd-seeming than an elephant?

But fantasy stories seem to be full of people who either know exactly what magical creatures are out there; or stubbornly refuse to believe in anything until it literally bites them; or they may have some wild superstitions instead of real knowledge of magical creatures, but any given character has either the one or the other. In reality, history did not see the study of the little-known neatly divided into categories according to what would eventually be discovered centuries later! There weren't proto-chemists on the left staring daggers at the woo-woo alchemists over on the right; woo-woo alchemists were the proto-chemists. Operating at the far reaches of knowledge, they had no way of distinguishing between a true rumour of the far and strange, and superstition.

So what happened in the world of Mistborn? Was there an allomantic Newton and a Maxwell of Metals who discovered, catalogued, systematised, and published on the laws and interactions of magic? Did all of our knowledgeable characters go through the school system to learn the Theoretical Allomancy?

Even if only the narrator seems to know too much, explaining everything about a magic system with no fuzziness or uncertainty around the edges has an uncanny effect. If I read a book and come out understanding exactly how the magic system works, it strikes me as unrealistic. What are the odds of someone writing a novel set in the real world and only ever mentioning things with which I am already familiar or which the narrative will explain to me? They seem rather slim to me; the moment your novel hints at the existence of, say, banking, or metropolitan transit planning, I know that the world of your story holds deep mysteries to which I shall never be privy. But a lot of fantasy novels feel rather sharply divided into things I understand perfectly well, which are common in the story; and…well, the things I don’t understand are on the outside, perhaps on the other side of the fourth wall. It is uncanny. If you blindfold me and put me in a wooded area, and I recognise every single species of plant around me…then I know I must be in a park, its flora picked from the very narrow selection of things I can immediately recognise. If I were in a real forest, I’d see some pine trees and some mosses I can’t name and an awful lot of plants I cannot classify even to the superfamily taxonomic level. Real forests are full of mysteries. There are plenty of things even the experts can only identify as “little brown mushrooms” or “little brown birds”, and nobody is an expert at everything.

All of this is to say that verisimilitude in world-building requires the presence of the mysterious and the unknown, because the real world is full of the mysterious and the unknown (and the moreso the lower the technology level of a story’s setting). Characters who know too much outright break the story’s logic, and in a looser and more implicit fashion, the same is true of a narrator whose omniscient poise is too omniscient. The forces and creatures that exist in the real world are imperfectly understood, so surely magical forces would also be misunderstood and surrounded by myths and misunderstandings; and a story that mentions only things fully and accurately understood feels unreal like a park is unreal, artificial, and stilted. There’s an obvious temptation for an author show off the world they spent so much time, energy, and creativity developing (do I ever—I’ve never written a story worth reading, but am an inveterate world-builder: I wish you could publish that and nothing else!), but that’s a temptation best resisted. There’s nothing wrong with the author knowing exactly how all the fantastic elements operate that they introduce into their world: in fact, it is much better if they do. But that does not mean that the narrator must know everything, inasmuch as the narrator is a character, a story-teller played directly by the author.

¹ Of course, the narrative fabric may not be entirely in harmony with the story, and that’s also fine as long as it is intentional. Careful phrasing may hide a twist (I haven’t read The Planet of the Apes, but I imagine it must be phrased rather deliberately), and a tone at odds with the text may do so for the sake of effect—telling a silly story with a somber tone can heighten the comedy (cf. the movie Airplane!); telling a dark story in a light-hearted tone can lend an edge of satire. But in all these cases, the author is choosing words intentionally for effect, not being misleading through sloppiness.

² I use them as a familiar example that has some of the aspects I want to discuss, but some of my concerns are ameliorated and some are arguably logically resolved. I know I’m being a bit unfair, but using a well-known story is too useful to give up. Sorry, Mr. Sanderson. I do like the books! At least enough to re-read twice so far.

³ Or more precisely, to the neuronal networks in my materialist brain responsible for the

⁴ Not in a way that’s bad prose. I don’t mean that he writes very repetitively, only that the information is repeated a few times across the several books to make sure it is clear to the reader. In terms of expository prose, I think that’s a positive.

Suppose you’re a modern-day Grimm, utterly in love with folklore—perhaps traditions you grew up with; perhaps traditions you fell in love with because they are new and strange to you in a place you moved to as an adult. As you talk to people in the rural area about old stories, you come across a truly great one—perhaps it involves moose and moonshine—which you hear about from two old codgers who heard it from their respective grandfathers.

You love it; it’s brilliant; you immediately decide to write it down and publish it (whether as a short story or novella or academic treatise on folklore, you may decide)…but you’ve got a bit of a problem: You have two sources for your story, but they don’t quite agree. Each contains details missing from the other; and, worse, there are points on which the stories are incompatible, or flatly disagree.

How will you present these two stories, given their differences? I can think of a number of approaches, most of which I’ve seen in actual use. For example:

You might simply give both versions. I don’t think that needs much explanation: You can publish one story after the other.

Or, since they’re mostly the same story, you might annotate differences by giving one story as the main text and indicating differences in footnotes, endnotes, or marginal notes.

If you want the story to flow better and make it more of a narrative than an academic read—basically, if you want to be the final product to be a single, uninterrupted narrative—perhaps you’d rather supplement one version with the other by basically telling one version from beginning to end, but taking elements from the other that the first was missing. In this case, when the stories disagree, you might either prefer one over the other, or you might take a more archival view of things and want to preserve both versions, thus including anecdotes from the secondary story even when they introduce tension or contradictions.

You might take this one step further and harmonise the stories. In this case, you’d create your final version by including elements of both stories but making sure to explain away any tensions and reconcile any contradictions.

What you probably wouldn’t do—at least I wouldn’t—is to chrologically interleave the narratives. This peculiar approach works as if you took the first story and cut it up into short chunks—paragraphs or sometimes even sentences—and annotated them each with its chronological position in the story; then, did the same with the second version of the story; and finally, included all the pieces of both versions by taking them in strict chronological order, even if that sometimes means interleaving the stories: A/B/B/B/A/B/A/A/… And if you did this, surely you would at least make sure not to include all the many times that both stories agree, repeating the same details twice.

But if you were a biblical redactor, this is precisely what you might decide to do. In fact, it is what the writers who produced the final biblical text did with a number of well-known and important stories. When Joseph is sold into slavery in Egypt, that’s two stories, or two versions of one story; in each case the basic gist is that he’s betrayed by his brothers, except that one plans to save him, but he is sold into slavery in Egypt before he can be thus saved; but the two versions disagree on why his brothers hated him, which brother wanted to spare him, or by whom he was sold. The story simply tells you that it was Reuben who decided to save him and also that it was Judah, and the editor does not appear to have done anything to either resolve or explain the contradiction: the text is a mess—the product of chronological interleaving. And the same is true of several other stories: for example, the Flood myth famously doesn’t make sense, and David’s introduction to Saul and fight with Goliath has similar doublets and contradictions (including the trivial but striking fact that David kills Goliath twice in immediate succession, first emphatically without a sword and second emphatically with one).

To me, this editorial decision is so profoundly puzzling that I cannot stop thinking about it. I am aware of some common ingredients in standard answers to why the text looks this way: A great value placed, perhaps due to political necessity, on preserving both traditions rather than prioritising the one over the other. The fact that while some doublets can be preserved by repeating similar stories (someone passing off his wife as his sister logically could happen more than once), other things can’t be repeated like that (even if it’s awkward to have the Flood last only forty days and one hundred and fifty days in one story, you can’t tell one after another within the same narrative if there’s to be only one deluge). And yet, as explanations, they feel as fragmentary to me as the preceding sentences might strike the reader; I still don’t really understand how someone could find this a satisfying solution, on purely literary grounds; it simply isn’t how a story works in my mind. (And this has nothing to do with whether you treat this as putative history or recognise it as pure myth; my problem here has to do with literary structure, not historicity.)

I think what I’d really like to see is a book that explores this topic by giving a broader overview of narrative construction in general and in the Ancient Near East in particular, compares it to the development and (if ever applicable) recombination or reconciliation of parallel narratives in other cultures, and so on. Unfortunately, if such a book exists, I have not been able to find it. If I had infinite time and energy, I would go back to university, earn another degree or two, and write that book as a monograph. If I instead had infinite money, I’d endow a grant or scholarship somewhere in return for somebody writing it. As it is, I suspect I may go to my grave many years from now still wondering why, oh, why those biblical redactors had such a profoundly weird editorial approach.

How is this the way anyone decides to combine two stories?