Overview
This is a research-driven interactive web experience that traces the story of women sci-fi writers across decades and languages. It consists of personal essays and archive materials that unfold the system behind a piece of women-produced sci-fi: how it gets translated, classified, published and awarded. More importantly, what this system leaves out. Sci-fi by women has long been underrepresented and miscategorized in this male-dominated field; and once you factor intersectionality, the voice from women of color rarely reaches the Anglophone readers.
It is an ongoing project. So far: a 76-author dataset, 10 source-language reconnaissance memos, a cross-recon synthesis, a narrative spec, and a scroll-driven storybook site. They live in the open at Shellwecode/feminist-scifi.
Why I started
I made it for my own curiosity. I wanted to read more of the feminist science fiction I’d been missing and to understand why I’d been missing it. I was eight when I first read 张晓风 (Zhang Xiaofeng), at the time filed as an essayist but whose early work is now recognized as the first piece of Chinese-language science fiction. Then, in college, I saw translation decide, in real time, which work got to travel beyond its source language and which didn’t.
I am curious about
1. How much of the women’s sci-fi has gone absent, untranslated, or filed under a different label?
2. Who draws the genre line, and who gets filed by it?
3. What role does translation play in communicating and preserving the story?
4. What does the map of absence look like?
Bias in the dataset
Phase 1: Individual authors
Phase 1 was per-author research files across 7 regions. I researched 76 authors total, weighted toward Anglophone (29) but spanning East Asian, African + diaspora, Latin American, European non-English, South Asian, and MENA writers. I capped the Anglophone canon, and studied how the works influence each other. In , I also explicitly documented the exclusion in the literary world: pre-2000 Brazilian women’s SF, the ninety-year South Asian gap, Korean SF before 1992 — and the “women writers” framing’s exclusion of non-binary writers, flagged as a judgment call still open for revisiting, not settled.
Phase 2: Source-language reconnaissance
Since phase 1 was built on English-language sources, I ran a second research pass with an inverted methodology from phase 1: source-language venues first, English only as supported confirmation. I was surprised that around 30% of the English-language results from phase 1 were inaccurate. For instance, Nora Naji is actually an Egyptian novelist but not Maghrebi; Sonia Sant’Anna writes historical novels but not sci-fi. The claim that Chinese women’s novel-length SF is “sparse” is empirically wrong in Chinese.
The gap here is in the translation pipeline.
More importantly, the translation gap is not equivalent to writing gap. Six regions have rich source-language traditions that simply never reached English. Treating “scarce in English” as “scarce” had been distorting the whole dataset.
The recurring form of absence
I thought the source-language approach would have cleared up most of the inaccuracies, but it only corrected the surface. More structural bias appears:
1. The foundational women are filed elsewhere.
In 8 of 10 regions, the woman closest to a tradition’s founding was catalogued as “literary,” “essayist,” or “poet”, and missed by genre history entirely. This includes Karin Boye, Dinah Silveira de Queiroz, 張曉風, Magdalena Mouján Otaño, Buchi Emecheta.
2. Gendered genre-policing.
On 晋江, women’s speculative writing at massive scale is filed as 言情/仙侠/玄幻, but almost never 科幻. They are excluded from the SF canon by definition alone. Prestige publishing does it too: Bazterrica’s Cadáver Exquisito is ciencia ficción in Argentina, “horror” in the US.
3. Prize asymmetry.
Germany’s Kurd Laßwitz best-novel award: one woman laureate in 45 years. Recognition skews harder against women than publication does and a prize rarely comes with a translation budget.
A more honest narrative
The data naturally revealed a more honest narrative, and I connected them together.
Feminist sci-fi is not yet a global conversation but a scattered set of solitary efforts. The translation lag is closing, but it largely depends on the named translators’ decisions. Behind the scenes, the story of feminist sci-fi is also a story about the field builders, editors, prize-founders, translators.
Building the site
Three build iterations, each a full working site: a “narrative museum” that
proved the structure, storybook vignettes where vertical scroll drives
horizontal page-turns, and a hand-drawn illustrated study as the opening —
all vanilla HTML/CSS/JS, fully keyboard-navigable, with
prefers-reduced-motion fallbacks throughout.
What I learned
- The bias is in the sources, not just the canon. Building from English-language references reproduces the exact exclusions the project set out to document. The 30–40% error rate on inherited names is the strongest argument I have for source-language-first methodology — and it only became visible because Phase 1’s assumptions were written down explicitly enough to falsify.
- Record what you reject. The “considered and rejected” lists and
gaps.mdturned out to be the most intellectually valuable files in the repo. The dataset’s edges — who’s excluded, and by what rule — carry as much of the argument as its contents. - Let the data revise the thesis. The romantic version didn’t survive contact with the evidence. The honest version — scattered solitaries, an Anglo-American crystallization, a conditional and fragile translation surge — is a better story because it’s true.
- Absence needs typology. “Missing” is four different phenomena with four different implications. Designing empty rooms with labels was more honest — and more affecting — than filling them with weak evidence.
- In design, the quiet solution usually wins. Nearly every visual iteration moved from more (washes, outlines, glows, big titles) to less (existing tones deepened, a hairline border, a smaller title in the corner).
What’s next
- Phase-2 dataset expansion: ~115 source-language-verified Tier-1 candidates across ten regions — a 2–3× expansion, with the new data fields the recons demanded.
- Revisit the framing exclusion: extend non-binary inclusion to the seven flagged writers whose work is central to the feminist-SF conversation.
- Finish the rooms: replace placeholder illustrations in the vignettes and closing cases; build out the field-builders room.
- Verify the load-bearing edge: 钱莉芳’s 《天意》 (2004) → Liu Cixin’s 《三体》 — if the verbatim source confirms, a woman’s novel is the acknowledged precedent for the most famous work in Chinese SF.
