How to Build a Fictional World That Earns Its Place in the Story
I drew a mountain range on a napkin in a diner outside Missoula, in June of last year, because a character in chapter fourteen needed to be six days late to her sister's wedding and I did not know why.
I had tried the easy answers first. A storm. A lamed horse. A checkpoint. None of them earned six days -- they earned one, maybe two, and my protagonist was already the kind of woman who would find a way around a single delay. So I got out a napkin and drew the pass she would have to cross, gave it a name, and put a landslide in it from three years before the novel's opening, one that had narrowed the road to a single cart's width and left the toll station undermanned. Then I did something I almost never do: I worked out, in actual hours, how long it takes one rider to lead a horse through a mile of half-collapsed switchback trail in low autumn light, rather than ride it. The answer was ugly and slow and came out to exactly six days once I added the two nights she would have to spend in a village that did not want her there. The napkin took forty minutes. It fixed a hole in my outline I had been circling for three weeks.
That is worldbuilding working the way it is supposed to work: fast, specific, in service of a scene that would not resolve without it. Most worldbuilding does not work that way. Most worldbuilding -- and I include years of my own in this -- spirals outward from a napkin sketch into a spreadsheet, then a folder, then a private wiki nobody but the author will ever read, most of which never touches a page of the actual book. I wrote about that spiral in an earlier piece, the seven months I spent building a world with three currencies and a thirteen-month calendar for a novel I never wrote a word of. This is the companion to that piece. That one was about the trap. This one is the method that avoids it -- a practical, in-order approach to building the specific parts of a fictional world that a story actually needs, in the sequence that keeps you moving toward the manuscript instead of away from it.
The Pressure Principle, Applied as a Method
The governing idea, if you read the previous piece, is that worldbuilding exists to put pressure on your characters, and nothing else it does matters nearly as much. I am not going to re-litigate that argument here. I want to show you what it looks like as a working method rather than a diagnostic you run after the fact, once the damage is already done and you are staring at forty thousand words of lore and no draft.
Here is the method in one sentence: for every element of your world, before you build it, name the character it is going to squeeze, and the scene in which the squeezing happens. Not eventually. Not "somewhere in the book." A specific character, a specific scene. If you cannot name both, you are not ready to build that element yet, no matter how curious you are about it.
That sounds restrictive. It is restrictive, and it is supposed to be. The instinct that makes people good at worldbuilding -- curiosity, systems-thinking, a hunger to know how a place works all the way down -- is precisely the instinct that, left unmanaged, builds you a world with no book attached to it. The method does not ask you to be less curious. It asks you to spend the curiosity on the parts of the world your specific story is going to touch, and to defer the rest, on purpose, without guilt, until a later book earns it or it simply stays in the drawer with my napkin and my dead-god calendar.
The Iceberg Approach, and How Deep to Go
Every worldbuilder has heard the iceberg metaphor, usually borrowed, a little loosely, from Hemingway's theory of omission -- the idea that a writer who knows a subject thoroughly can leave things out, and the reader will still feel their presence, the way an iceberg's true mass is felt even though nine-tenths of it sits below the waterline. Gene Wolfe is the writer people cite most often for pulling this off in a secondary world. In The Book of the New Sun, he uses real, obscure, archaic English words -- fuligin, chatelaine, peltast -- for invented things, without stopping to define them, and lets the reader assemble meaning from context and confidence. The world feels bottomless because the prose never once apologizes for not explaining itself.
The trap in the metaphor is that writers hear "nine-tenths below the waterline" and conclude they need to build the nine-tenths. They do not. What they need is for the one-tenth on the page to behave as though the rest exists -- which is a question of consistency and implication, not volume. A place name reused correctly three chapters later implies history. A character's offhand reference to a war nobody explains implies depth. Neither requires you to have actually written the war's causes, casualties, and peace treaty in a separate document. The test I use is this: does this fact need an answer, somewhere in a file, or does it only need to feel like it has one? Most of the iceberg can stay unanswered, because a novel is not an encyclopedia and nobody is going to check.
The exception is anything your plot is actually going to stand on. If the war's outcome determines who has authority over the mountain pass in chapter fourteen, then yes, you need the answer, because a character's life depends on it and an inconsistency there will eventually surface as a plot hole. That is the whole discipline in one distinction: build real answers under the ten percent that touches the page, and let everything beneath that stay a rumor, a texture, a shape glimpsed and not confirmed.
I have started keeping two separate lists for exactly this reason, and the separation alone has saved me weeks. One list is "answered" -- facts I actually worked out because a scene needed them, with the chapter number next to each so I can find my reasoning again later. The other is "implied" -- facts I have gestured at in the prose, a war, a dead king, a border dispute, that I have deliberately left unresolved because no scene has asked me to resolve them yet. The second list is not a to-do list. It is a permission list. It exists so that when I feel the pull to go answer one of those questions anyway, I can look at it and remember that the pull is curiosity, not necessity, and the manuscript is waiting.
The Physical World: Geography That Decides the Plot
Geography is the worldbuilding with the biggest payoff for the least effort, and it is the layer most writers under-invest in relative to magic systems and political maps, because it looks like the boring part. It is not boring. It is causal. Where the mountains sit decides who can reach whom and how fast. Where the river bends decides which town got rich on trade and which one got poor. How far the harvest has to travel before it rots decides whether a famine is three weeks away or three years away. Geography is the one layer of worldbuilding that generates plot logic almost by accident, because distance and scarcity are conflict engines whether or not you intend them to be.
The land decides who arrives late.
I draw distances before I draw coastlines now, and it has changed the order in which I build almost everything. A map that starts as a pretty coastline invites decoration -- an island here because it looks nice, a second mountain range there because the page felt empty. A map that starts as a table of travel times between the handful of places your plot actually visits invites plot. How many days from the capital to the border garrison, on foot, in winter, with a wagon slowing the party down. Once those numbers exist, the coastline can be drawn around them in an afternoon, because now it only has to be consistent with facts your story already depends on, rather than invented first and reconciled with the plot later, which is the more common order and the one that generates the most contradictions in revision.
Frank Herbert did not set out to write an ecology textbook. In the early 1950s, a magazine assignment sent him to Florence, Oregon, to report on a government project using European beach grass to stop migrating sand dunes from swallowing the coastal highway. The article he was supposed to write never ran in the form he pitched it, but the reporting stayed with him for a decade and became the seed of Dune -- a planet whose entire civilization, from the stillsuits that reclaim a person's own sweat and breath moisture to the worm-riding culture of the Fremen to the economics of a single valuable resource controlling galactic politics, is downstream of one physical fact: there is not enough water. That single scarcity, worked through with a journalist's patience for how systems actually function, generates more plot, more culture, and more character psychology than any invented magic system I can think of. Herbert did not build seven schools of desert magic. He built one unforgiving physical constraint and then followed its consequences all the way to the throne room.
You do not need a planet's worth of ecology to use this lever. You need to know, for your specific story, what the land makes hard. I think of the nonfiction writer John McPhee here, oddly enough -- not a novelist at all, but the finest living chronicler of what geology actually does to human plans, in books like Basin and Range and Annals of the Former World. McPhee spends hundreds of pages on how a fault line or a flood plain quietly rewrites the fate of the people who build a town on top of it, and the lesson transfers directly to fiction: decide what the land refuses to allow, and your plot will start generating itself. A siege that cannot be resupplied because the only road floods every spring. A wedding delayed six days because a pass collapsed three years ago and nobody has fixed it. Draw the geography that creates the friction your story needs, and skip the geography that does not.
Cultures: What People Believe Under Pressure
Culture is where most beginning worldbuilders go straight to invention -- a made-up religion, a made-up holiday calendar, a made-up system of honorifics -- without first asking the question that makes any of it cohere: what material condition produced this belief? Custom is rarely arbitrary in a well-built world. It is usually the fossil record of a problem the culture solved, or failed to solve, generations ago, still being performed long after anyone remembers why.
Ursula K. Le Guin approached this with more rigor than almost anyone in the genre, and it is not an accident. She was raised by an anthropologist father, Alfred Kroeber, and a mother, Theodora Kroeber, who wrote about Ishi, the last survivor of the Yahi people, in a house where culture was studied as a living system rather than invented as set dressing. It shows. Always Coming Home is written as something closer to speculative ethnography than a novel -- songs, recipes, and field notes of a people called the Kesh, presented the way an anthropologist would present a real culture's material. And in The Left Hand of Darkness, the Gethenians' fluid, ambisexual biology is not a costume detail. Their entire social structure, their taboos, their politics, their concept of shame, all follow from that one biological fact, worked through with the seriousness of someone who was raised to ask what a custom is actually for.
The practical version of this, for your own draft, is a habit rather than a system: for every custom you invent, write one sentence naming the material problem it originally solved, and then ask which of your characters is now caught between the custom's dead original purpose and its living, current enforcement. A harvest tithe that once prevented famine and now just enriches a corrupt council is not lore. It is pressure on the farmer who still pays it out of habit and the daughter who has started asking why. That gap -- between why a custom exists and why people still obey it -- is worth more dramatically than any amount of invented vocabulary for greetings and holidays.
I use this on something as small as a mourning custom. In the same manuscript as the mountain pass, I had a village that painted a door white when someone died in the house, a detail I invented purely for atmosphere on a Tuesday afternoon. It sat there as texture for months, doing nothing, until I asked the material question and realized the custom only made sense if white paint was expensive and locally rare -- which meant a poor family painting a door white was announcing, publicly, that they had spent money they did not have to mourn correctly. That single material fact turned a decorative detail into a scene: my protagonist's mother refuses to paint the door, and the whole street knows what the refusal means before a word of dialogue explains it.
Politics: Who Holds the Lever, and Why
Political worldbuilding fails in a specific, recognizable way: a chart. Four factions, twelve named conspirators, a line of succession going back nine generations, none of it connected to a person the reader has met. I have built that chart. I built it for the novel with the dead-god calendar, and nine of my twelve conspirators never appeared on a page. The chart was not politics. It was an org diagram wearing a crown.
Le Guin again, because she solved this problem cleanly: The Dispossessed is subtitled An Ambiguous Utopia, and its politics -- an anarchist society on the moon Anarres set against a propertied, capitalist Urras -- never sit as background. They sit inside one physicist, Shevek, who is a citizen of one world and a visitor in the other, and who is squeezed, specifically and personally, by the contradictions of both. The politics is legible because it has a body to press against. N.K. Jemisin's Broken Earth trilogy does the same thing with more violence: the caste system that governs orogenes -- people whose geological powers could save civilization but who are legally property, trained by "Guardians," some of them literally wired into the earth as node maintainers to keep quakes from destroying cities -- is not explained in a paragraph of exposition. It arrives as a single image, a child strung with tubes and left in a box beneath a city, and that one image indicts the entire political structure more efficiently than any chart could. Jemisin, who worked for years as a counseling psychologist before she published fiction full time, builds systemic cruelty the way a clinician notices it: specific, embodied, and never abstract.
So the rule for your own political worldbuilding is the same rule as everywhere else in this piece: before you draw the chart, find the body. Who, specifically, is crushed by this law, this succession rule, this trade embargo? Build the law only deeply enough to explain what it does to that person. The rest of the chart -- the eleven other conspirators, the six generations of minor kings -- can stay a rumor your characters mention in passing, if it needs to exist at all.
Magic or Technology: One System, Fully Loaded
Magic and technology are the same worldbuilding problem wearing different clothes, and the rule for both is identical to the rule for politics: every capability needs a cost, paid by a specific character, in a specific scene, or it is not a system, it is a wish list. I wrote a whole piece on the three sentences a magic system needs before chapter one, including where writers like Brandon Sanderson and Jemisin sit on the hard-versus-soft spectrum, so I will not repeat that argument here. What I want to add is the technology half of the equation, because writers reach for magic systems far more often than they reach for the equally powerful move of restricting technology.
Herbert's Dune is instructive here too. The single most consequential piece of worldbuilding in the entire Dune universe is not the spice or the sandworms. It is the Butlerian Jihad, a holy war fought millennia before the novel begins that outlawed all "thinking machines." Removing computers from the setting is what forces the existence of Mentats, humans trained from childhood to perform computation the way a machine would, at the cost of a strange, flattened humanity, and Guild Navigators, who mutate their bodies on a diet of spice to fold space without machine assistance. A restriction generated two of the most memorable character types in science fiction, and it did it more efficiently than adding a new gadget ever could have. The absence is the worldbuilding. What your world's people cannot do is often more load-bearing than what they can.
Susanna Clarke builds the same kind of absence into Jonathan Strange & Mr Norrell: English magic has not vanished by accident but has quietly died out, leaving a whole profession of "theoretical magicians" who study and footnote a craft none of them can actually practice, which is why the arrival of two men who can do real magic upends institutions, politics, and friendships that had organized themselves for a century around magic's absence. The scarcity is not an incidental fact about her world. It is the entire premise, and every character's ambition, rivalry, or fear is calibrated against how rare and how threatening real power has suddenly become.
Whichever direction you build in, the test is the same one I gave you for politics. Take your magic system or your technology, and for every rule, ask: who pays for this, specifically, and when? A fire mage whose cost is abstract -- "magic is dangerous" -- is a worldbuilding diagram. A fire mage who, fifteen years after paying the price, flinches when someone she loves takes her hand, because the cost she paid was the ability to feel warmth, is a character. Build toward the second kind and stop as soon as you have enough rules to produce that specific, embodied cost for the character in front of you.
Language: The Smallest Worldbuilding With the Biggest Return
Tolkien built two complete languages, with grammar, phonology, and centuries of internal history, because the languages were the actual project and the novel was built to give them somewhere to live. That is an almost unrepeatable set of circumstances, and I want to say plainly what most worldbuilding advice dances around: you almost certainly do not need to do this, and trying to is one of the most efficient ways to spend six months producing zero pages of manuscript.
What you need instead is a short, deliberately incomplete lexicon -- twenty to forty invented terms, each one attached to a scene where it actually gets used. A form of address that marks class. A curse word specific to a trade. A place name whose etymology you know even if you never explain it on the page, because knowing it will keep you from accidentally contradicting yourself in book two. Ted Chiang does more with language, in a single novella, than most epic fantasy does with a thousand pages of grammar. In "Story of Your Life," the heptapods' written language is non-linear -- a whole sentence exists as one simultaneous shape rather than a sequence of words -- and learning to read it changes how the linguist protagonist experiences time itself. Chiang, who spent years as a technical writer producing software documentation before he became one of science fiction's most decorated authors, builds the entire emotional architecture of the story out of a linguistic idea, not a glossary. You do not need the glossary. You need the idea the language is for.
It is worth knowing where the actual professional ceiling of this work sits, so you can calibrate against it honestly. David J. Peterson, the linguist hired to build Dothraki and Valyrian for the television adaptation of A Song of Ice and Fire, produced full working grammars with verb conjugation, case systems, and thousands of root words, on an HBO budget, over months of dedicated, paid, specialist labor -- and he has written an entire book, The Art of Language Invention, on how to do it properly. That is what full conlanging actually costs in time and expertise. You are not being asked to compete with that. You are being asked to invent forty words that a reader will half-remember and never look up, because they never needed to.
The Procrastination Trap, Named Plainly
I covered the psychology of this at length in the earlier piece -- why worldbuilding feels like writing while carrying almost none of writing's risk, and why the satisfaction you feel after a good worldbuilding session should worry you more than it reassures you. I will not re-run that argument here. But I want to give you the fastest version of the tell, because you need something you can check in thirty seconds, mid-session, before the spiral gets three weeks deep.
Ask yourself: could I open my manuscript right now and use what I just built in the next five hundred words? If the honest answer is no -- if what you just built is interesting but does not touch tomorrow's scene -- you were probably avoiding something. That something is almost never the worldbuilding. It is the scene.
The scene is always the thing you are avoiding.
Worldbuilding is safe because you can always be more thorough. A scene has to actually be written, badly at first, by you, today, and there is no version of "more thorough" that substitutes for sitting down and doing it.
When to Stop Building and Start Writing
The stopping point most writers look for does not exist. There is no moment when a fictional world is finished, and waiting for that moment is how novels die in the research folder. The stopping point that does exist is smaller and much more useful: you have built enough when you can answer, for each of the next five scenes on your outline, every worldbuilding question those five scenes are actually going to ask.
Run this checklist before your next writing session, not before your next research session:
- List your next five scenes, in order, from your outline.
- For each one, write down the single worldbuilding fact it depends on -- the pass, the custom, the law, the cost of the magic -- and only that fact.
- Build those five facts, and no others, even if a sixth fact is tempting you.
- Write the five scenes.
- Repeat, with the next five, once you get there.
This will feel like you are leaving the world unfinished. You are. The world is always unfinished, the same way a real country's history and geography and law are, at any given moment, unfinished. Readers do not experience your setting as a completed object. They experience it exactly as fast as your characters move through it, one scene at a time, which means the five-scenes rule is not a compromise. It is simply matching your building pace to the pace your reader will actually consume the results.
I ran this checklist properly for the first time on the novel with the mountain pass, about four chapters after the napkin. I had a list of nine worldbuilding questions I was itching to answer -- the pass's geological history, the toll station's founding, a second pass on the far side of the range I was fairly sure existed but had not needed yet. I ran the five-scenes test against all nine. Two survived: the pass itself, because the next scene crossed it, and the toll-keeper's name, because a character was about to say it out loud. The other seven went into a single-line list titled "later, if it matters," and none of them has mattered since, across four hundred pages. I do not think about them as failures. I think about them as the correct amount of world for the book I was actually writing.
Building While You Write, Not Only Before
The deepest change I made to my own process, after the seven-month failure, was to stop treating worldbuilding as a phase that happens before drafting and start treating it as a habit that happens alongside it. I no longer sit down with the intention of "building the world" for a week before I start chapter one. I sit down to write chapter one, and when chapter one asks a question the world has not answered yet -- what does this village smell like in the harvest month, who actually enforces the toll on that pass -- I answer it in that moment, write it down in one place so I do not contradict myself later, and go straight back to the sentence I was writing. The building happens in the gaps the writing opens up, not in a separate, earlier, safer room.
This requires one piece of infrastructure: a single running file where those in-the-moment answers get recorded as you generate them, searchable, so that when you invent the toll-keeper's name in chapter fourteen you can find it again in chapter thirty-one instead of quietly renaming him and hoping nobody notices. A running lore file that your project's other content -- the map, the family tree, the flowchart tracking cause and effect between political decisions and consequences -- can all reference is worth more than a pre-written bible, because it grows at exactly the rate your actual manuscript grows, never faster. If you want the fuller step-by-step version of setting this system up before you start, the guide to worldbuilding walks through it in more depth, and the worldbuilding bible template gives you a structure sized for exactly this just-in-time approach rather than the encyclopedic one.
If you want the practical, in-order version of the argument I made in the earlier piece -- why worldbuilding has one job, and what happens when you forget that -- it is here, and the two pieces are meant to be read together: one tells you why to stop, this one tells you how to build in the first place without needing to.
The napkin from the diner outside Missoula is still in my desk drawer. I never redrew the mountain pass properly, never gave it a full geological history, never worked out what minerals ran through it or who had fought over it four centuries before my novel opens. I did not need to. I needed six days, a landslide, and a village that did not want my protagonist there, and the napkin gave me exactly that and nothing more. The book got written. The mountain pass appears in one scene, briefly, and does its job, and I have never once needed to know anything else about it.
Plotiar keeps a running Lore knowledge base alongside your chapters -- the place-in-the-moment answers, the map, the family tree, the cause-and-effect flowchart -- so the world grows at the pace your manuscript actually needs it to. If you are ready to build only what the next scene requires, start a project for free.