Science fiction: ideas vs. furniture
· The StoryProp team
Most science fiction drafts that stall have the same problem, and it is not a lack of imagination. It is too much imagination spread evenly. The writer has invented flying transit, synthetic protein, an orbital court system, a slang word for cigarettes, and a corporation with a hard-consonant name, and none of it is doing more work than any other part. Everything is equally new, which is the same as nothing being new: the reader has no idea where to look. The fix starts with a distinction worth keeping on an index card near the desk, between an idea and a piece of furniture.
The one new thing
Critics have a useful term for the idea at the center of a science fiction story: the novum. The new thing. It is the single change to the world that the story exists to examine — the premise that, if you removed it, would collapse the book rather than merely redecorate it. A story can carry more than one, but not many, and the second is always in service of the first.
Say your novum is that grief can be professionally outsourced. Someone dies, and the family hires a person to feel the worst of it on their behalf, through a procedure the story never needs to explain in detail. That is a new thing, and it generates story on contact: who takes that job, what it costs them, what happens to a family that skipped a stage it needed, whether the practice is legal, whether it is respectable, whether the rich do it differently than everyone else.
Now consider a different item. In this world, elevators announce floors in a synthetic voice with a regional accent. Charming, specific, completely inert — delete it and the story is unchanged. That is furniture, and furniture is not a sin. A world made only of its central idea reads like a philosophy seminar with names attached. You need the accent in the elevator, the vending machine that takes two tries, the ugly municipal carpet. What matters is knowing which is which, because the two demand entirely different amounts of your attention, your page count, and your rigor.
Consequences are the actual work
Here is the most common failure in a first science fiction draft: the writer invents something genuinely new and then uses it exclusively for chases. The technology exists to be stolen, hacked, escaped from, or shot at. It never touches anything else. But real changes to the world do not confine themselves to action sequences. They leak. Ask what your novum does to four things it has no obvious business touching.
Work. Who does this for a living now, and who used to do something else? Outsourced grief means there is a trade, and trades have training, hierarchies, burnout, and either a union or a conspicuous absence of one. Someone is very good at it and is not proud of that. Love. What does the new thing do to intimacy? If your spouse hires a professional to feel your mother's death, is that a kindness or an insult? Couples now have a fight available to them that no couple has ever had before. Write that fight.
Law. What is the statute? Someone has been prosecuted by now. There is a precedent case everyone in the industry knows by a two-word name, and a rule that came out of it that everyone finds slightly humiliating to follow. Manners. This is the one writers skip and readers love most. What is rude now? Perhaps asking a professional mourner who they worked for last week lands the way asking someone their salary lands. That single social rule tells a reader more about the world than a page of infrastructure.
Run those four questions and you will have more story than you can use. The premise stops being a gadget and becomes a society, which is the only thing that makes a novum worth a whole book.
Incluing instead of explaining
The second-order problem is delivery. You now know a great deal, and the temptation is to hand the reader most of it early so they can admire your rigor. Resist. The technique that works is sometimes called incluing: letting the reader assemble the world from what characters do and take for granted, rather than from what the narrator stops to explain. A character glances at the door before saying a name. Someone apologizes for a smell that no longer needs apologizing for, and someone younger looks confused. A form has a checkbox the point-of-view character has ticked before and dreads ticking again.
The practical rule: characters should not explain to each other things they both already know. If exposition must be spoken aloud, give it a reason — a newcomer, a dispute about the facts, an official reciting a policy they resent. Better still, let the reader stay slightly behind and trust them to catch up. Confusion that resolves within a page or two is pleasure; confusion that persists for thirty pages is a different animal, and you can tell them apart by asking whether the reader knows what the character wants. A reader will tolerate not understanding the technology indefinitely as long as they understand the desire.
One mechanical note. Terminology is cheap to introduce and expensive to maintain. If you name a thing, name it once, use it consistently, and do not rename it in chapter nineteen because you thought of something better. Every invented word is a small debt the reader agrees to carry, and a draft that keeps issuing them without paying anything back is exhausting to read.
Plausibility budgets
You have a finite amount of rigor. Spend it where the reader is looking. This is the principle that saves science fiction writers the most time: not every element of your world needs to survive expert scrutiny, only the elements the story leans on. If the plot turns on the legal status of outsourced grief, the legal reasoning must be tight enough that a lawyer reading it nods. If someone merely takes a train in chapter four, the train does not need a propulsion system. It needs to be late.
Work out where attention lands. It lands where the story spends time, where a character argues, where the plot pivots. Those are your load-bearing walls, and they get the research, the second-guessing, the friend who actually knows something about the field. Everything else gets one confident sentence and no further inspection. A brisk unexamined detail reads as competence; a long explanation of something irrelevant reads as anxiety. Which points at a related trap, the belief that more explanation makes a thing more plausible. Usually the reverse. The more you justify, the more surface area you hand the reader to doubt. State it, move on, and behave as though it is ordinary — because to your characters, it is.
This differs from how invented rules tend to work in fantasy, where a magic system often earns its keep by being legible as a set of costs and limits the reader can reason with. Science fiction can do that too, but it is not obligated to; a novum can stay a black box as long as its effects stay consistent. If the rules-and-costs approach is what you are after, the fantasy magic systems discussion takes it apart properly.
Keeping the world without drowning in it
The material accumulates fast: the precedent case, the checkbox on the form, the two-word slang for the procedure, the rule about what you may not ask a colleague. Six months in, you will not remember whether you established that the practice is licensed by the state or by a professional body, and you will have written scenes that assume each. The failure mode there is not forgetting. It is not knowing which of your notes are decisions and which are still options, because a file of undifferentiated worldbuilding is a file you stop trusting.
So mark the difference between what is settled and what is furniture you might still rearrange. The novum and its four consequences are settled. The elevator accent is not, and does not need to be. A notebook will hold the notes. What it will not do is keep that distinction visible six months later, when every line in the file looks equally decided. That is the job AI novel writing software exists to do: what you have settled is written down as settled, and what you have not stays a suggestion. The fuller method, including how to avoid writing an encyclopedia instead of a book, is handled in worldbuilding without the wiki trap.
A test before you start drafting
Write two sentences. The first states your novum in plain language, no invented vocabulary allowed: one new thing about this world. The second states what it costs someone. If you cannot write the first without proper nouns you made up, you have furniture where your idea should be. If you can write the first but not the second, you have a premise rather than a story — an interesting condition that nobody has to live through yet.
The Quiet House, the demo manuscript this site keeps returning to, has no speculative element at all: a woman named Mara, her brother Daniel, a chipped cup on a windowsill that means something because of who last drank from it. Science fiction carries the same requirement and one more on top. The cup still has to matter. The new thing only has to change what it would mean to break it. Get the one idea clear, follow it into work and love and law and manners, spend your rigor where the story looks, and let the reader assemble the rest. The furniture arrives on its own once you know what room you are standing in.
When the notes outgrow your memory — the precedent case, the slang, the rule about what nobody asks — StoryProp keeps the settled decisions as records the agent reads before it drafts, so the world holds still while you keep deciding what it means.