The Critical Journal of the British Science Fiction Association

The New X - Issue 249

Storying Genres, by Graham Sleight

Storying lives, eh? Okay, here's a story: I saw Steven Spielberg's AI twice in the cinema. The first time was July 2001, in a sweltering cinema in Cambridge, Massachusetts. As far as I remember, the audience's reaction was quiet, attentive, a little bemused by the shifting tone of the film. The next time was four months later, in a cold and rainy London. The reaction was much the same, except for one shot, which provoked gasps of surprise. When we approach the film's frozen Manhattan, we see the towers of the World Trade Center trapped in the ice. Their windows are missing, parts of the structure are gone, but both are clearly still intact. Even then, a month or two on from the terrorist attacks on Manhattan, it was clear that whatever replaced the WTC towers, it would not be an exact replica. So, apart from the shock of having the attacks referenced by accident, as it were, I think the audience was responding to the fact that the film was now wrong, that it was telling a story about the future which (in this small respect) could not be true.

I found myself thinking, afterwards, about whether that wouldn't make a useful definition of science fiction: sf stories are those that advance hypotheses about the world that might become true but which are falsifiable. (Alternate-world stories are those which might once have been sf by this definition, but which are now falsified: we know now that Hitler didn't win WWII, but he could have.) I was thinking of Karl Popper's definition of science: that scientific experiments are those which can be falsified. We construct a hypothesis that gravity works in a certain way, and drop a ball off a table to see if the hypothesis will be disproved. This rattled around in my head for a while before I came to the conclusion that, while it might be a workable definition of sf, it wasn't very useful. It didn't particularly explain why, for instance, far-future space opera is sf: many of the technological advances put forward in such stories are so far removed from our present state of knowledge that one can't begin to say whether or not they might be possible.

I have to say, in general, that debates about the definition of sf (or fantasy, or horror) don't exercise me very much though of course that may reflect a lack of rigour on my part. I am quite taken by Samuel Delany's view that we should not try to define genres because, for instance, definition inevitably means concentrating on boundary cases at the expense of the core of the genre, because it sets up a target which critics and writers can game, and so on. But there are plenty of people who do try to define sf in radically differing ways, and I thought it might be useful to try and sort some of those ways out:

1. Trope-based. "Science fiction is that which contains robots/spaceships/ray-guns." Probably the commonest, but also the most flawed way of creating definitions. It leads you into just looking at the surface of stories, and into errors as a result. For instance, it means you wind up thinking of Gene Wolfe's The Book of the New Sun as fantasy (because that's where its props are from) rather than as sf (because all its fantasy tropes ultimately have underlying sf justifications, as do the large-scale moves of its story.)

2. Effect-based. "Science fiction is that which induces sense of wonder", "Horror is that which makes you experience horror", and so on. Darko Suvin's famous definition of sf as a literature of cognitive estrangement within a 'novum' probably belongs here. The problem is that such definitions tend to land you with a single aesthetic criterion by which to judge whether a work of sf is any good.

3. Architecture-based. I recently read Farah Mendlesohn's excellent book on Diana Wynne Jones, in which she attempts to define and classify fantasy stories based on what might be called the macro-structure of each story. Similarly, one could come up with an obvious architecture-based definition of the detective story a crime happens, and the large-scale plot of the book is the action of solving it and bringing about some kind of restitution.

4. Move-based. I'm thinking here of what John Clute, in particular, has wound up creating through his encyclopedias and other critical work: a sense of what's most characteristic about sf or fantasy through small or mid-scale moves of story. A Clute term like, say, "thinning" doesn't define the whole story, just describes a part of it.

5. Market-based. Science fiction is that which is marketed as science fiction the pure Marxist definition, if you like. The problem is that plenty of sf has been written using the tropes and techniques of the genre but published outside it: and even if Margaret Atwood wants to exclude The Handmaid's Tale from the canon, I don't want to.

6. Tradition-based. Damon Knight's famous statement that " will do us no harm if we remember that [science fiction] means what we point to when we say it" (In Search of Wonder, 3rd edn, 11) is basically an assertion that sf is a tradition or genealogy, and that those sufficiently acquainted with the tradition can say what falls into it.

I suppose the Knight definition is the one I feel closest to, if pushed. Sf in particular is a peculiarly time-bound conversation, texts responding to each other in a way that also embraces, for instance, our critical discourse, the conversations we have at conventions, and increasingly the online world as well. But it seems to me that the most interesting development of the last decade or so is the extent to which genre boundaries are being gamed and subverted: which I'll talk about next issue.

Graham Sleight lives in London and writes for The New York Review of Science Fiction, Strange Horizons and Interzone.

This article first appeared in Vector 249. Back issues of Vector are available from