Worldbuilding as Foundation

First appeared: How to Write Absurdly Well — Adrian Tchaikovsky Mode: SCHOLAR Domain: Writing craft / story structure

Definition

Worldbuilding is not preparation for the story — it is the generative source from which character, plot, and emotional resonance all emerge. In Tchaikovsky's method: start with a single what-if premise; follow its ripples outward through logical consequence until the world achieves sufficient internal coherence and pressure; then let the world tell you what the story must be and who the characters must be.

This is the opposite of the psychology-first approach (start with a character's wound; build a world that tests it). Here, the world pre-exists the story. The story is the most interesting route through the world's existing tensions. Characters arrive already embedded in the world's web of relationships — they are not constructed from psychological backstory but discovered by walking through the world and seeing who is there.

The method originates in tabletop role-playing game mastery: "When you're creating a world for a role-playing game, you make it very robustly because you don't know what players are going to break." Building for other people's exploration produces worlds that are genuinely inhabitable — they extend considerably beyond the pages of the book itself, which is exactly what immersive speculative fiction requires. [DIRECT QUOTE — lightly edited]

The Stone-in-Pool Method

The core worldbuilding technique:

"The image I usually use is the idea that you're dropping a stone into a pool. And then you have the ripples from where that stone impacts, and you follow them out, and each set of ripples is a consecutive sort of logical therefore. If you just — that is your starting point, well, therefore this must be true. And if that's true, then this must be true." [DIRECT QUOTE]

The what-if is the stone. Everything follows through consecutive logical consequence. The world is built bottom-up, not top-down. By the time Tchaikovsky sits down to formalize the world, it has been "boiling away in the back of my head probably for months, years even." The formal sitting-down phase takes weeks; the background phase takes years. [PARAPHRASED]

What makes a good what-if:

Originality: avoids repetition of others' work; finds a genuine spin on any crowded space or doesn't enter it
Generativity: a seed that grows — you feel the book expanding as you think about it
Combinability: many books happen when two half-ideas click together into one complete idea (Children of Ruin: uplifted octopuses + alien neurology; neither was enough alone)

The Left Wall

"What science says is possible — that is your left wall as you write the book. So you can only expand outwards from that." [DIRECT QUOTE]

The left wall is the constraint you may not cross — the body of established knowledge (science, history, period accuracy) that defines what must be true in your world. The wall is not a limitation but a structure: "working out the contours of that left wall is what your research is." [PARAPHRASED]

Across the spectrum from hard science fiction to epic fantasy:

Hard science fiction: the left wall is as large as possible; many things are foreclosed; credibility requires living within real science's constraints
Space opera / soft sci-fi: the left wall is present but selective; some conventions (faster-than-light travel) are permitted departures
Fantasy: "You build your own left wall by hand. You end up with a world that has the same amount of logic and reality to it as your science fiction world. It's just that you have built it by hand rather than importing large chunks of it from the real world." [DIRECT QUOTE] — magic works like this; this is what magic costs; this is what magic cannot do

The unknown unknowns problem: you will not know what you don't know until you need it. Research uncovers those gaps before they become plot-breaking problems mid-draft. [PARAPHRASED]

The One Big Lie

"You can get away with one big lie. And it's just that one thing which is just convenient to have in the book to make the plot work in the way that you want, and that is your one big lie. But, in order to support your one big lie, everything else needs to be true." [DIRECT QUOTE]

The one big lie is a permission structure, not a recipe: every story is allowed exactly one departure from reality. In Children of Time, the big lie is the nanovirus accelerating spider uplift from hundreds of millions of years to tens of thousands. The lie is the timescale; the evolutionary biology is as accurate as Tchaikovsky could make it. The honesty of everything surrounding the lie is what makes the lie land with weight — readers feel "this is a thing that could happen." [PARAPHRASED]

This principle applies across genres. Historical fiction: the big lie might be a counterfactual event; the period detail around it must be genuine. Fantasy: the big lie might be the magic system's foundational premise; the internal logic of that system must be consistent.

Note: the one big lie governs departures from external reality. It does not license internal inconsistency. A world can have one premise that departs from the real; that premise must then be followed through with complete internal honesty.

World Generating Character

Tchaikovsky's characters are not designed from psychological backstory — they emerge from the world:

"By the time that character has emerged as — ah yes, let's have him in the book — there are all those strings attaching that character to all the other parts of the setting already there. They're not things I necessarily need to tack on by hand because it's inherent in who this character is as to how he's going to relate to everything else that's going on." [DIRECT QUOTE — lightly edited]

The pawnbroker example from City of Lost Chances: he is an immigrant from a nation previously invaded by the current occupiers of his city. His entire relational situation — to the occupiers, to the resistance, to the academic he deals with — is already implied by who he is and where he comes from before the writer has consciously designed any of it.

Characters still surprise in the writing: "They've got a thing romantically for this other character that's not in the plan at all, but it's just like it's a thing you discover when you meet them because there is a gap there that all of the exactly that shape that they then move to fill." [DIRECT QUOTE] Character depth reveals itself as you meet characters in their world, not as you construct them in the abstract.

The key distinction from psychology-first approaches: The Weiland/Herne framework (see Character Arc Architecture) builds characters from a Ghost → Lie → Want/Need/Truth structure. The wound precedes the world. Tchaikovsky's method inverts this: the world precedes and produces the character's situation. Both can produce compelling arcs. The methodological question is whether you start with an internal psychology or an external situation — and whether the internal psychology is designed in advance or allowed to emerge. [ORIGINAL — structural observation comparing the two methods]

City vs. Travelogue Settings

Travelogue fantasy (the dominant mode): "You can walk away from the consequences of what you do because you're always going on to the next point in the map." The heroes leave Bree; they never see what happens after the ringwraiths' pursuit. They can murder-hobo their way across the continent and not account for any of it. [PARAPHRASED]

City settings: "You are a resident of that city. You've lived there all your life. You have an existing relationship with all of these people and factions and events. And when you do a thing, you have to stick around and live with the consequences." [DIRECT QUOTE]

The consequence structure is the narrative mechanism: small actions taken for petty reasons can cascade into city-wide upheaval, but only if the characters cannot escape. The city is a closed system; the travelogue is an open one. Closed systems produce more complex consequential narratives. They require more prep — the board with the red strings is not metaphorical — but are "commensurably more rewarding." [PARAPHRASED]

The same principle applies to any contained setting (a spaceship, a small town, an institution). Confinement is a narrative structure, not just a setting choice.

Fight Scenes as Narrative Vehicles

Fight scenes are not interruptions in the story — they are concentrated opportunities for character revelation and world demonstration.

Research discipline:

Acquire the knowledge: stage fighting, historical martial arts, LARP, technical consultation (the submarine designer for the water-filled octopus spaceship)
Then apply the iceberg principle: "Once you have learned all these clever things from whatever means you're using to learn about fights, you need to learn how to put as little of that as possible on the page. Because the temptation, as with anything you've done the research on, is I am just going to vomit all this stuff on the page to show how incredibly intensive my research has been. And at that point, pacing goes out of the window." [DIRECT QUOTE]
"Once you know that stuff, it will inform the details you do put in. And it will genuinely show through." The details that arise from genuine knowledge are the ones that make a scene real. [DIRECT QUOTE]

The iceberg principle generalizes: this is the operative principle for all research across all genres. You learn it for the authority; you hide most of it for the pacing. What makes it through is the handful of details that could only have come from someone who actually knew.

Three scales of fight:

Duel — technique is the narrative; the fight's choreography expresses the characters directly; what one fighter chooses not to do (the kill shot not taken) is character development; the two people's relationship is told through the physical exchange
Skirmish — mobile, complex; tracking multiple combatants requires spatial planning (literally toy insects on a desk); characters trading opponents; requires narrative management
Battle — two perspectives available: the general's-eye view (tactical overview, strategic picture) and the slot-in-visor view (The Heroes by Joe Abercrombie as the model: "this enormous panicky cacophony of people charging about and barging into each other") [PARAPHRASED]; each mode serves different narrative purposes; neither is superior

Emotional perspective over technical precision: Fight scenes should be told "as much through the emotions as through the footwork or the precise blade work." The insight came from LARP experience — "you feel it here" — the visceral physical reaction of standing in a battle line when a screaming charge comes down a hill at you. Even in a game, something in you reacts as if it's real. That embodied emotional knowledge is what goes on the page. The technique serves to prevent readers from throwing the book; the emotion is what makes them keep reading. [PARAPHRASED]

The Three-Sided Knowledge Structure

"You get this three-sided knowledge structure. There's what you as the author know, and there is what the characters in the book know, and there is what the reader knows." [DIRECT QUOTE]

Three primary configurations and their effects:

1. Character knows more than reader (unreliable narrator) The character is concealing information or has an ulterior motive; the reader is reading through their selective account. The reader gradually catches up to what was being withheld. City of Lost Chances: multiple POV characters, each concealing who has the missing item; none ever admit it; the reader eventually realizes the same. The tension comes from watching characters who know things act around each other. [PARAPHRASED]

2. Author and reader share knowledge the character lacks (dramatic irony) The reader is given a context the character cannot access. The reader watches the character make decisions that look reasonable from inside but catastrophic from outside. Dogs of War: Rex narrates war crimes he cannot recognize as war crimes — he lacks the ethical framework to understand what "small enemies" means. "You as the reader are potentially just kind of chewing your fingernails off saying this is this is this is really really terrible and Rex is such a this kind of amiable likable character who just thinks he's being a good dog." [DIRECT QUOTE — minor cleanup] This mode is particularly powerful in post-tech settings: the reader brings knowledge of the civilization that was lost; the characters are living in its wreck without knowing it. [PARAPHRASED]

3. Character surprises the author (emergent discovery) In unplanned narratives, the character knows something the author doesn't — and the author only discovers it by writing the character long enough. In City of Lost Chances, Tchaikovsky didn't know who had the item when he started. He wrote each character from "an entirely agnostic position" of whether they might have it. By the end of the book, it became obvious which character must have had it, what they'd done with it, and when the reveal would come. [PARAPHRASED]

Not Planning Endings / The Ending as Most Important

These two principles sit in apparent tension; Tchaikovsky reconciles them:

Not planning endings: "What I've never done is actually work out the very last ending." The penultimate scene may be fixed. The final beat is not. "Letting the trajectory and momentum of the book to that point tell you how it should end has worked really well for me." [DIRECT QUOTE] The feeling: "You get to the point of feeling that you are chronicling." [PARAPHRASED]

The ending as most important: "Absolutely no bones about it, the ending is the most important part of the book. The most important part? Definitely because it's the bit that your readers are left with." [DIRECT QUOTE] "The ending needs to be the bit where everything that you intend to tie up in the book turns out to have been inextricably leading to that ending. Even though while it was doing that leading, that wasn't remotely obviously where it was going." A good ending is the logical result — but also a surprise. [DIRECT QUOTE — lightly edited]

The reconciliation: the reason not to plan the ending is precisely because the ending is the most important thing. Planning it in advance imposes it from outside the book's internal logic. Letting it emerge means the ending is the only thing it could have been — which is what makes it feel inevitable in retrospect. [ORIGINAL — synthesis of Tchaikovsky's position]

On Writing Maxims and Dogma

One of the sharpest claims in the source. "Any piece of advice is useful so long as it doesn't basically become 'this is the law.'" [PARAPHRASED]

The Hero's Journey: "It's just one structure. The idea of saying, 'Well, all this is the story that all stories must follow.'" When forced onto a book, the template requires interpretive gymnastics — "I guess this is kind of our descent into the underworld" — that expose how badly the model fits. "Once you are reducing something that far, then you're left with a model that is completely worthless." [DIRECT QUOTE] This is an explicit methodological claim: the models that claim universality are the most dangerous, not the most useful.

The show-don't-tell rule: Children of Time is "basically 60% exposition by volume" — and is his most successful book. "Sometimes that is the best way of doing something." [DIRECT QUOTE] The rule holds in many contexts; it fails when the subject matter (alien cognitive architectures; post-tech world history unknown to characters) requires exposition to be the primary vehicle.

Every writer I know does something I personally find unthinkable. "There's no one way of doing this, and at the end of it, you always converge on having a book." [PARAPHRASED] This is the most important craft meta-principle in the source: processes are personal; outcomes are what matter.

Rhetorical Worldbuilding (Mayya) — A Different Operation

Varun Mayya uses "worldbuilding" to describe a structurally different process from Tchaikovsky's ontological worldbuilding. Where Tchaikovsky builds a coherent secondary world with its own internal laws (the stone-in-pool method), Mayya's worldbuilding means: model the reader's existing mental world and construct your communication to fit inside it or modify it incrementally.

The core mechanisms:

Fit into an existing world: identify the reader's foundational beliefs (atomic units of agreement) and enter through them; build complex claims on top of shared premises
Build a new world: anchor novel elements in portions of the reader's existing reality before introducing departures (Dune's Arabic vocabulary and Islamic religious parallels as the known foundation for sand worms and Fremen)
Frame shift: validate the reader's worldview first; introduce modifications one step at a time; never attack a held belief directly

The key distinction: Tchaikovsky's worldbuilding generates a fictional secondary reality. Mayya's worldbuilding maps and leverages a reader's actual belief structure. Tchaikovsky is asking: what kind of world could logically exist? Mayya is asking: what does this specific reader already believe, and how do I enter through that? Both call it worldbuilding; they are different cognitive operations.

Critical divergence: Tchaikovsky's "one big lie" permits exactly one departure from reality, supported by maximum honesty everywhere else. Mayya's framework has no equivalent — his entire approach operates within the reader's existing reality, introducing only what the evidence can support. The constraint is not "one lie permitted"; the constraint is "no departures from what can be documented and accepted incrementally."

See Writing as Applied Psychology for the full Mayya framework.

Prompting as Worldbuilding

Mayya's pivot claim: effective AI prompting is not merely analogous to worldbuilding — it is worldbuilding, technically.

The argument: an LLM without a specific prompt exists in a broad, undefined region of its probability space — capable of generating nearly anything from its training. A prompt constrains the model to a specific region. The more detailed and internally consistent the prompt, the more precisely the model's output is constrained. Every prompt is constructing a temporary universe for the model to operate inside.

Technical grounding (simplified): token embeddings place concepts in a high-dimensional mathematical space; related concepts are geometrically close. A vague prompt ("tell me a story") places the model in a generic, undifferentiated region. A rich prompt ("write a detective story set in 2049 cyberpunk India where the protagonist communicates only in gang signs") teleports the model to a highly specific, constrained region. The attention mechanism — which determines what the model "pays attention to" when generating each token — is directed by the specificity of the prompt context. Detailed prompts are scripts for the attention mechanism.

Advanced prompting techniques as worldbuilding equivalents:

Few-shot examples = giving the model lore and precedent for the world
Style constraints = setting the physical laws of the world
Negation rules ("don't use the phrase 'not just X, it's Y'") = banning specific behaviors from the world
Persona instructions = establishing the character the model inhabits inside this world

"You are actually spawning universes every time you prompt." [DIRECT QUOTE — Mayya, ~35:35]

Caveat: Mayya's claim that this is "not just an analogy" but "an exact technically accurate description" is overstated. The structural parallel is real and useful. But a human reader processes language through identity, prior beliefs, and social meaning; an LLM processes through mathematical operations on token probability distributions. The leverage points are structurally similar (context constrains output); the mechanisms are not identical. The prompting-as-worldbuilding framework is a powerful practical heuristic, not a mechanistic equivalence.

Vault implication: this opens AI/prompting as a domain in the vault for the first time. The newsletter's focus on mid-career creatives navigating AI-augmented practice makes this a high-relevance cross-domain bridge.

Intellectual Nomadism as Research Practice

The writer's research discipline is not just accumulating knowledge in the subject of the novel — it is deliberate border-crossing into adjacent unfamiliar fields.

Shafak: "I've always believed in being intellectual nomads. We shouldn't have comfort zones... when a novelist becomes interested in neuroscience or when a scientist is drawn to poetry... those moments I think are the best, when our mind is open to learning." [PARAPHRASED — Shafak]

This is a stronger version of Tchaikovsky's iceberg principle: the knowledge you carry into a novel includes material from fields that have nothing obvious to do with the subject. What makes it through to the prose carries the authority of genuine border-crossing — the unexpected accuracy of a novelist who has read water science, or a fantasy writer who has studied historical martial arts. The iceberg's hidden mass is partly built from intellectual nomadism.

The related principle: oral culture as a distinct epistemological domain. Shafak spent time with Yazidi grandmothers during research for There Are Rivers in the Sky — a community with no written holy book whose entire cosmology lives in oral transmission. Written archives cannot access this layer. The research practice that produces the most unusual material is one that listens for what cannot be found in text. [PARAPHRASED — Shafak]

Evidence and Sources

How to Write Absurdly Well — Adrian Tchaikovsky — ontological worldbuilding; stone-in-pool method, left wall, one big lie, world-generating character, city vs. travelogue, fight scenes, three-sided knowledge structure; practitioner interview; 60+ books; clean transcript
Best Skill to Learn in 2026 — Varun Mayya — rhetorical worldbuilding; reader-first orientation; atomic units of agreement, frame shift, reach from ground truth, prompting as worldbuilding; video transcript; inductive reasoning from entrepreneurial/content creation practice
Shafak, The Key to Truly Beautiful Writing — intellectual nomadism as research practice; oral culture as distinct epistemological source; the Yazidi fieldwork model

Tensions

Ontological worldbuilding (Tchaikovsky) vs. rhetorical worldbuilding (Mayya): Tchaikovsky builds a coherent secondary reality with its own internal laws. Mayya models and leverages the reader's existing mental world. Same word, different operations. The one big lie rule (one permitted departure from reality, honesty everywhere else) has no parallel in Mayya's framework (which requires staying within the reader's acceptable reality at every step). These are not competing accounts of the same thing — they are tools for different problems: one for constructing fiction, one for constructing persuasive communication. But both rely on a prior insight — you must understand the world you are building for. [ORIGINAL — structural comparison]
World-first vs. character-first: Tchaikovsky's method produces characters embedded in a relational web before they have a psychological interior. The Weiland/Herne framework (Ghost/Lie/Want/Need/Truth — see Character Arc Architecture) constructs the psychological interior first. These are different generative orders. A character produced by world-first has a given situation; one produced by psychology-first has a given wound. Whether these converge on the same endpoint (a character with both a situational context and an internal arc) is an open question. The two methods may be complementary rather than mutually exclusive if used in sequence. [ORIGINAL]
Not planning endings vs. ending as most important: reconciled in Tchaikovsky's account (see section above) but the paradox is real and worth maintaining as a teaching tension.
Exposition as best tool (sometimes) vs. prose-level compression: The Herne taxonomy at Level 6 (writing around feeling; subtext; elevated scope) implies that the best prose avoids stating what it means. Tchaikovsky's 60% exposition claim implies that sometimes direct exposition is the best tool. These are not contradictory if applied at different levels: exposition can be the right structural choice even when each individual passage is still being written with precision and economy. But the tension between "show the feeling through the prose" and "tell the reader directly what is happening" needs a more developed account of when each applies. [ORIGINAL]
Stone-in-pool worldbuilding and the left wall: The stone-in-pool method produces logical internal consistency. The left wall imports external constraints. In hard science fiction, these two forces align. In fantasy, the writer builds their own left wall — but the stone-in-pool logic now operates within a self-constructed constraint space, which means the writer must be scrupulous about not cheating their own rules. The discipline required is identical, but the external accountability is gone. [ORIGINAL]

Connected Concepts

→ Character Arc Architecture — world-first methodology as productive alternative to psychology-first; characters from world vs. characters from Ghost/Lie/Truth
→ Theme as Moral Argument — "monsters that tell us about us"; the world-building choice is already a thematic argument
→ Prose as Transmission — the three-sided knowledge structure; the iceberg principle for research; the exposition-as-best-tool counter-case to show-don't-tell
→ Arc Types: Positive, Negative, Flat — Tchaikovsky's city-based, consequence-trapped settings produce the conditions for character arcs to run fully; travelogue settings allow characters to avoid arc completion
→ Writing as Applied Psychology — Mayya's rhetorical worldbuilding as a distinct operation; the reader-first orientation; prompting as worldbuilding

Open Questions

Is the stone-in-pool / world-first method suitable for writers with smaller output targets, or does it require the kind of prolific practice (60+ books) that trains the subconscious to fill the gaps reliably?
The numinous is described as an unachieved goal — a quality Tchaikovsky recognizes in Mythago Wood and Piranesi but cannot yet produce. Is this a worldbuilding property (how the world is constructed), a prose property (how it is written), or a metaphysical property (what the writer actually believes about the boundary between the seen and unseen)? This connects to the Prose as Transmission Level 7 (sublime prose) — both are pointing at something that cannot be directly achieved, only approached.
What is the relationship between the three-sided knowledge structure and established narratology? Does this appear in academic narrative theory (Genette, Booth, etc.) under different terminology?
Tchaikovsky reads peers rather than the greats for development. Is this a general principle for mature practitioners — you have already internalized the greats; what you need now is to see what's possible at the frontier? And does this generalize to other domains (eastern spirituality, philosophy, psychology)?

Last updated: 2026-04-18 (Shafak ingest: intellectual nomadism as research practice section added; oral culture as epistemological source; Shafak source added)