Infoporn: 100 Years of Sci-Fi, Explored

Using data scraping, network analysis, and machine learning, the Science Fiction Concept Corpus includes more than 2,600 books written since 1900. Here's what we found.
illustration of a UFO over an open book
Emily Waite

AI Researcher Bethanie Maples has been reading science fiction since she was given a copy of Dune at 10 years old. Still, two decades and nearly 1,000 books later, the self-described sci-fi fanatic struggles to find books that delve into her most niche interests, like the link between AI and transhumanism. So last year, while working at Stanford’s Human Computer Interaction lab, she teamed up with data scientists Eric Berlow and Srini Kadamati to create a book recommendation tool based on more than 100 salient sci-fi themes, from hyperspace to magical feminism. Using data scraping, network analysis, and machine learning, the resulting Science Fiction Concept Corpus includes more than 2,600 books written since 1900. We made our own voyage into Maples’ sci-fi universe.

Alternate Histories

The Science Fiction Concept Corpus is built on plot descriptions, reviews, and user-generated tags scraped from Goodreads, sci-fi forums, and other sources. “It was interesting to see how sci-fi authors foreshadowed developments in history, like AI winters,” says data scientist Eric Berlow, who helped create the Corpus.

Expand Your Horizons

The Sci-Fi Corpus reveals “first-degree neighbors,” books that share some—but not all—common themes. The tool helps readers discover a broader range of relevant books from the past and present.

Book Recommendation Generator

The Corpus suggests titles based on 108 topics of interest, enabling intelligent browsing rather than algorithm-­driven results, Maples says.

Sci-Fi Concepts Over Time

The researchers analyzed the prevalence of more than a dozen high-level concepts in science fiction, from human control to augmentation. “Powerful books can fuel our imagination or instill fear,” Maples says. “You can often draw a slender thread between technology trends and social movements.”

Source: The Corpus equates popularity with the number of public reviews on Goodreads
Genre Benders

By linking books that share relevant keywords, the Corpus exposes hidden correlations between various sci-fi themes.

Source: The Corpus equates popularity with the number of public reviews on Goodreads
By the Numbers
The most popular sci-fi books, by decade:
  • 1900–09: The Wonderful Wizard of Oz (Oz #1), L. Frank Baum
  • 1910–19: The Metamorphosis, Franz Kafka
  • 1920–29: Steppenwolf, Hermann Hesse
  • 1930–39: Brave New World, Aldous Huxley
  • 1940–49: 1984, George Orwell 1950–59: Fahrenheit 451, Ray Bradbury
  • 1960–69: A Wrinkle in Time, Madeleine L’Engle
  • 1970–79: The Hitchhiker’s Guide to the Galaxy, Douglas Adams
  • 1980–89: The Alchemist, Paulo Coelho
  • 1990–99: The Giver, Lois Lowry
  • 2000–2009: Catching Fire, Suzanne Collins
  • 2010–present: Divergent, Veronica Roth

View the complete Sci-Fi Corpus here.


Lauren Murrow (@­laurenmurrow) wrote about the tech gender gap in issue 26.10.

This article appears in the March issue. Subscribe now.


More Great WIRED Stories