Science Fiction Writing Tips Nobody Else Will Tell You

If you have ever read “ten tips for writing science fiction” anywhere on the internet, you have read approximately the same list, written approximately fifty thousand times. The list will tell you to “build a believable world,” to “ground your future technology in real science,” to “make your aliens more than humans in costumes,” and to “show, don’t tell, the futuristic elements.” None of these are wrong. All of them are useless, in the sense that any working SF writer knows them already and any beginner cannot apply them without a much more specific framework. The bad-advice ecosystem around SF writing is large. The good-advice ecosystem is small, scattered, and mostly hidden in the prefaces of books by writers who are too busy producing actual fiction to write a Medium post.

I’m going to share the SF-craft tips I’ve collected over fifteen years of reading working SF writers’ interviews, prefaces, and essays. These are the ones that are not in the standard listicle, that working writers actually use, and that will materially change what you produce if you take them seriously. They assume you already know the basics; if you want the foundational sci-fi writing primer, the existing CreativInfluence guide covers the ground. This is the post-foundation layer.

1. The technology has to cost somebody something

This is the rule that distinguishes serious SF from convention-mimicry, and it is the single most important craft principle in the genre. Every speculative element you introduce needs to be paid for, in the world, by somebody. The faster-than-light drive needs to require something nobody wants to give. The neural interface needs to do something nobody asked for. The colony on Mars needs to be funded by someone whose interests are not aligned with the colonists.

Octavia Butler’s Bloodchild (1984) is the cleanest example I can give you. The story is about humans on an alien planet who have a symbiotic relationship with the indigenous species; specifically, the indigenous Tlic lay their eggs inside human bodies, and human boys are raised to host them, with the surgical removal of the larvae handled by skilled adults. The technology (cross-species reproduction) costs everyone in the story something. The Tlic give up reproductive autonomy. The humans give up bodily autonomy. The “alliance” is uneasy precisely because the cost structure is real. Butler said in her preface that she wrote it to be “a love story between two very different beings,” and the love is legible only because both parties are paying continuously for the relationship.

If your speculative element doesn’t cost anybody anything, your worldbuilding is decoration, not fiction. Find the cost. Make somebody pay it. The story is in the payment.

2. The protagonist’s worldview should be unable to fully explain the world

Le Guin’s Steering the Craft makes this point in a slightly different framing, but the underlying principle is what separates literary SF from genre SF: in literary SF, the protagonist’s interpretive frame is not adequate to the world they inhabit, and the gap between what they understand and what is actually happening is part of the story’s engine.

In The Left Hand of Darkness, Genly Ai is a foreign envoy who keeps misreading the gendered politics of Gethen because his Earth-derived gender categories don’t apply. He is wrong throughout much of the novel about basic things. The novel is Le Guin teaching the reader, through Genly’s misreadings, what gender could be if it were not what we have. If Genly understood the world correctly from page one, there would be no novel.

This is harder to write than it sounds. The trick is to give the protagonist a coherent but insufficient interpretive frame. They must be intelligent. They must have reasons for their misreadings. Their misreadings must be reasonable given what they know. Then the reader, reading over their shoulder, accumulates the gap between the protagonist’s reading and what’s actually happening, and the gap is where the SF actually lives.

3. Use the second-best invention, not the best one

This is a Ted Chiang principle, articulated in his Paris Review interview (2018). Chiang said that when he develops a story idea, he often finds himself with multiple speculative premises that could power the same emotional arc. The temptation is to use the most spectacular one (the most futuristic, the most visually impressive, the most marketable). His rule is to use the second-best one; the slightly more mundane premise, because it forces the story to do more of the imaginative heavy lifting through character, language, and structural choices, rather than relying on the premise’s spectacle to carry the reader.

“Story of Your Life” (the basis for the film Arrival) is a perfect demonstration. The first-best version of that story would be alien linguists arriving on Earth to teach humanity a new physics. The second-best version (the one Chiang wrote) is a single American linguist working on translation in a tent, slowly realizing that the new language is rewriting her relationship to time, with the alien presence rendered minimally. The smaller premise produces a better story because the premise is forced to share narrative weight with the human element. The visual spectacle never has to do the work.

If you have a “high-concept” SF idea, write the version of it where the concept is one notch quieter than your first impulse. The story will be better.

4. POV is the most consequential SF craft choice

Most SF writing advice underweights POV (point of view). It is, in my opinion, the single decision that determines whether your SF can do what literary fiction can do or whether it stays in the convention-mimicry tier.

Third-person limited gives you the most versatility but requires constant discipline about what your character can and can’t perceive. First-person locks you inside one consciousness, which is enormously powerful for psychological SF (Chiang’s “Tower of Babylon,” Lispector-adjacent SF) but limits your ability to describe events the protagonist doesn’t witness. Second-person (rare, hard) can produce a specific kind of intimacy that genre SF almost never achieves; N.K. Jemisin uses it in The Fifth Season and the choice is load-bearing for that novel’s central revelation.

The foundational guide on first-person point of view covers the basic mechanics, and I’d recommend it before you commit to a POV choice for a long project. The thing it gets right is that POV is not just a perspective decision; it’s a contract with the reader about what the book is and isn’t going to do. Get the contract right and the book has a foundation. Get it wrong and the book will fight you forever.

5. The science doesn’t have to be correct, but it has to be consistent

Almost everything you’ve read about “scientific accuracy in SF” is wrong. Scientific accuracy is a moral preference of certain readers; it is not a craft requirement. What is a craft requirement is internal consistency. Your made-up physics has to behave the same way in chapter twelve as it did in chapter three. Your made-up alien biology has to follow the rules you established. If you change the rules, you have to do it in a way the reader can metabolize, usually by having the change be a discovery the characters make rather than an authorial cheat.

Cixin Liu’s Three-Body Problem trilogy is interesting because the science is audaciously wrong in places (the eponymous three-body problem is solvable in many real-physics cases, the dark forest scenarios depend on game-theoretic premises that are contestable, etc.) but the trilogy’s internal consistency is brutal. Liu commits, hard, to the rules he’s invented, and the consistency makes the wrongness of the underlying science irrelevant to the experience of reading. The world holds together. That is what readers are actually responding to, regardless of whether the underlying physics would hold up to a peer review.

6. Names matter more than worldbuilding

Samuel Delany has a long essay (in his About Writing collection) about the specific power of proper nouns in SF. His argument: a planet called “Velmaran-7” lands differently than a planet called “the second Lascaux.” The first sounds like 1960s pulp; the second sounds like literary SF. The difference is not the noun’s content; it’s the noun’s register, which signals to the reader what kind of book they’re holding.

The corollary: spend disproportionate time on your names. Names of planets, characters, technologies, ranks, organizations, languages. They will do more imaginative work than thirty pages of worldbuilding exposition because the reader processes register before content, and your registers are set by your nouns. Le Guin was famous for this. Earthsea works partly because the names work; “Ged,” “Sparrowhawk,” “Roke,” “Ogion,” each register-correct for what they’re describing.

If your names sound like your favorite genre’s clichés, you are writing inside that genre’s clichés regardless of what your prose does. If your names sound like nothing else, you are establishing a register-space your prose can then occupy.

7. The first chapter is the contract; the last paragraph is the receipt

Most working SF writers I’ve read on craft (Jemisin, Le Guin, Chiang, Delany) agree that the opening of an SF novel is unusually load-bearing compared to other genres. Because SF has to do worldbuilding, voice-establishing, and reader-orientation simultaneously, the first chapter is teaching the reader how to read this specific book. Get it right and the rest can do almost anything; get it wrong and even brilliant later chapters won’t be received correctly.

The corresponding rule on the other end: the last paragraph of an SF novel is a retrospective contract. It tells the reader what the book was actually about, which is sometimes different from what the reader thought the book was about while reading it. Le Guin’s last paragraph of The Left Hand of Darkness (Genly seeing the Karhidish snow) does not summarize the political plot; it locates the reader in a specific emotional fact about gender, journey, and friendship. The reader closes the book understanding that the journey was the point all along, even though the political plot had felt central for 300 pages. That last-paragraph reveal is the receipt.

Write the first chapter knowing what your last paragraph is going to do. If you don’t know what your last paragraph is doing, you don’t yet know what the book is about, and you should keep developing it before you finalize the opening.

8. The “explain or imply” axis is your most-used dial

Every speculative element in your SF can be either explained (the reader is told how it works) or implied (the reader is shown its effects without explanation). Both are valid. Most beginners over-explain, because they want the reader to know they’ve worked out the system. Most experienced SF writers under-explain, because they have learned that the reader’s imagination of the explanation is more vivid than any explanation the writer could provide.

Ted Chiang’s “Story of Your Life” never fully explains the alien physics. Becky Chambers’s Wayfarers novels under-explain almost all their tech. N.K. Jemisin in The Fifth Season withholds the central worldbuilding fact for hundreds of pages and the reader, asked to construct meaning from the gap, builds something more powerful than any explanation could have been.

The dial is contextual. Some readers want explanation. Some don’t. Find your audience by deciding which dial position feels right for the specific story you’re telling, and then commit to it; readers can metabolize a lot of explanation or a lot of implication, but they cannot metabolize inconsistency about which mode the book is operating in.

9. Read outside SF

This is the most disrespected piece of SF craft advice and it is, in my opinion, the most consequential one. Almost all the working SF writers I admire read voraciously outside the genre. Le Guin read Tolstoy. Delany reads Proust. Chiang reads philosophy. Jemisin has talked about her debt to Toni Morrison. The SF that works (the kind that survives multiple decades) draws its craft DNA from sources that are not other SF, because SF as a genre tends to recycle its own conventions, and a writer who reads only SF will produce SF that is increasingly downstream of itself.

If you want to write SF that doesn’t read like genre fiction, your reading list should be 60% non-SF, minimum. Read the great realist novels. Read poetry. Read essays. Read history. Read the writers who have nothing to do with your speculative impulses, because they will give you the structural moves and the prose-level vocabulary that your SF needs to not be merely SF.

10. The world has changed and your SF should know it

This last one is the meta-tip. The genre conventions of SF were largely set in the 1950s-1980s, in a Cold War context, by writers responding to a specific set of cultural anxieties (nuclear war, computer rationality, expansion of empire into space). The contemporary cultural anxieties (climate, AI, surveillance, late capitalism, the death of monoculture) are different, and SF that doesn’t know they’re different is going to feel decades out of date. The most exciting recent SF (Jemisin, Solomon, Chambers, Liu, Tchaikovsky, Ling Ma’s Severance) is responding to now, not to the genre’s conventions. The genre rewards writers who write into their actual present, even if their plot is nominally set on a generation ship in 2347.

If your SF could have been written in 1985, it probably should have been written in 1985, and the version you’re writing now is going to read as nostalgic. Write the SF that 2026 needs, which nobody else has written yet because they’re too busy writing the 1985 version one more time.

These are the tips. None of them are tips you’ll see in a “ten tips for writing SF” listicle, which is exactly why they’re the ones you’ll need. The genre has more room to grow than its conventions suggest. The growing happens when working writers refuse to mimic the conventions and write the next thing instead. That’s the whole job. The rest is paperwork.