On the Origins of Genre




first published in Extrapolation Vol.44 No.4, Winter 2003; reprinted in Speculations on Speculation: Theories of Science Fiction edited by James Gunn and Matthew Candelaria, Lanham, Maryland, The Scarecrow Press, 2005


There is no starting point for science fiction. There is no one novel that marks the beginning of the genre. We have all had a go at identifying the ur-text, the source from which Heinlein and Ellison and Gibson and Ballard and Priest and Le Guin and a host of others flow. Brian Aldiss famously named Frankenstein by Mary Shelley, and his suggestion has been taken up by a number of later commentators. Other strong contenders include H.G. Wells, or Edgar Allen Poe, or Jules Verne. Gary Westfahl has nominated Hugo Gernsback as the true father of science fiction. Still others (including myself) have gone back to Thomas More’s Utopia.

We are all wrong.

We have to be wrong, because there is no ancestral text that could possibly contain, even in nascent form, all that we have come to identify as science fiction.

What part of Frankenstein, for instance, as diluted as homeopathic medicine, is to be found in Philip K. Dick’s The Martian Timeslip, or Gene Wolfe’s The Book of the New Sun, or Octavia Butler’s Parable of the Talents, or Isaac Asimov’s Foundation? What, come to that, could possibly link these disparate texts, other than the fact that we have come to apply the name ‘science fiction’ to all of them?

This inability to define science fiction is a problem we have long recognised. In his 1986 work, Critical Terms for Science Fiction: A Glossary and Guide to Scholarship, Gary K. Wolfe included 33 different definitions of science fiction, many of which overlapped to some degree or other, but all of which included contradictions. The critical test for any definition is that it includes everything we believe should be included within the term, and it excludes everything we believe should be omitted. Strictly applied, every single one of those definitions would admit to the genre works that we would prefer to exclude, or would omit works we feel belong in the genre. Even Darko Suvin’s cognitive estrangement, expressed thus: ‘SF in general – through its long history in different contexts – can be defined as “a literary genre whose necessary and sufficient conditions are the presence and interaction of estrangement and cognition”’ (p.4), which seems to have become the default definition of choice of most academic critics, is a prescriptive definition which works fine as long as we are comfortable with what it prescribes, but can lead to extraordinary convolutions as we try to show  that certain favoured texts really do conform to the idea of cognitive estrangement, and even more extraordinary convolutions to reveal that familiar non-sf texts don’t.

Since Wolfe’s tour d’horizon, science fiction scholarship has expanded exponentially, and most commentators have felt the need to add a new way of defining what it is they are talking about. These vary from formulations that are so imprecise as to be virtually useless – Kim Stanley Robinson, in his Guest of Honour Speech at Readercon in 1997 said, ‘Science fiction is the history that we cannot know’ – to those that attempt to touch every base, and end up tying themselves in knots in the process. The Sixth Edition of The Oxford Companion to English Literature (2000), says:

    The label ‘science fiction’ suggests a hybrid form, not quite ordinary fiction, not quite science, yet partaking of both. Beneath this label, we find a variety of wares, some of which trail off from a hypothetical central point into utopianism or dystopianism, heroic fantasy, horror, and books on UFOs and the paranormal. Yet its startlements are normally based either on a possible scientific advance, or on a natural or social change, or on a suspicion that the world is not as it is commonly represented. It follows that one of the unacknowledged pleasures of reading science fiction (or SF) is that it challenges readers to decide whether what they are reading is within the bounds of the possible. (p.906)

This is an interesting definition, clearly written by someone sympathetic to the genre, yet as a means of identifying what science fiction is, it is useless. ‘[N]ot quite ordinary fiction, not quite science, yet partaking of both’: what does this mean? In what way is it not quite ‘ordinary fiction’ (and what might that be)? To what degree is it ‘science’? By what proportions does it ‘partake of both’? The qualification ‘normally based’ implies that a goodly portion of science fiction might not be based on any of the categories that follow. Yet these categories are hardly restricted to science fiction: ‘a suspicion that the world is not as it is commonly represented’ has been the hallmark of conspiracy theories throughout history, without any suggestion that such theories are necessarily science fiction. And literature based on some form or notion of ‘social change’ includes such key science-fictional texts as Sense and Sensibility, The Road to Wigan Pier and at least half the oeuvre of P.G. Wodehouse. I am being unduly harsh on what is, after all, an honourable attempt to apostrophize, briefly, the indefinable. But much the same criticisms could be levelled against any extant definition of science fiction, and it is only by looking at why these definitions fail that we can start to consider what it is that makes science fiction indefinable.

In brief: the more comprehensively a definition seeks to encompass science fiction, the more unsatisfactory it seems to those of us who know the genre. To which one response is that we simply ignore the question altogether. The Clute and Nicholls Encyclopedia of Science Fiction (1993) contains reference to just about every form of science fiction, but though there is an entry on ‘Definitions of SF’, it doesn’t actually include a definition of sf. The article, which covers much the same ground as the entry in Gary Wolfe’s book, is a conspectus of the different and often incompatible definitions that have been proposed for the genre. But it does not arrive at a single, comprehensive overview of what science fiction is. Either there is no such single, comprehensive definition or, as when The Oxford Companion to English Literature concludes that science fiction ‘challenges readers to decide’, we finally admit that science fiction is defined not by something intrinsic to the genre, but rather it is in the eye of the beholder. In other words, many of us end up echoing Damon Knight: science fiction is ‘what we point to when we say it’? (Though it is worth remembering that what Damon Knight actually wrote in In Search of Wonder (1956) was: ‘The term “science fiction” is a misnomer, … it will do us no particular harm if we remember that, like “The Saturday Evening Post”, it means what we point to when we say it.’ (p.1)) Knight’s ostensive definition of science fiction has been so frequently misquoted and adapted – one brutalist variant is Norman Spinrad’s (in Modern Science Fiction): ‘Science fiction is anything published as science fiction’ (quoted in Wolfe, p.110), which excludes Frankenstein, most of the works of H.G. Wells, and practically every contemporary science fiction novel, at least in Britain where publishers have mostly abandoned putting the descriptor ‘science fiction’ on their books – that its subtlety and value is often missed. I will be trying to exploit that subtlety in this essay.

The two questions are, of course, intimately connected. Where we find the starting point for science fiction inevitably affects how we define the genre, and vice versa. Thus Brian Aldiss’s famous definition from Billion Year Spree (1973) – ‘Science fiction is the search for a definition of man and his status in the universe which will stand in our advanced but confused state of knowledge (science), and is characteristically cast in the Gothic or post-Gothic mould’ (p.8) – cannily excludes anything that might precede his chosen ur-text, Frankenstein. Alexei and Cory Panshin, in The World Beyond the Hill (1989), want to present science fiction as being centrally concerned with transcendence, so their definition, ‘Science fiction is a literature of the mythic imagination’ (p.1), throws the net wide enough to include any of early man’s attempts at myth-making, while excluding more recent science fictions that shade into social realism. Gary Westfahl wants to prove, in The Mechanics of Wonder (1998), that Hugo Gernsback was the true father of science fiction and so constructs a ‘description’ of science fiction in which the community engendered by Gernsback through his magazines is a vital part of what makes the genre: ‘If we define a genre as consisting of a body of texts related by a shared understanding of that genre as recorded in contemporary commentary, then a true history of science fiction as a genre must begin in 1926, at the time when Gernsback defined science fiction’ (p.8).

In the beginning is the definition. And what we conceive science fiction to be inevitably dictates how we identify its origin. What’s more, where we place that starting point inevitably affects what we see as the history (and the prehistory) of the genre, which in turn changes our perception of what science fiction is. It is a mobius loop: the definition affects the perception of the historical starting point, which in turn affects the definition.

Except that just as we have no commonly agreed definition of the genre; we have no primum mobile that everyone can accept. Everyone is on a different mobius loop.

This is because, as I indicated in my opening statement, there is not one definition of science fiction but many, there is not one ur-text but many. We choose whichever best suit our conceptions of science fiction, and change those choices (or devise new ones) as our conceptions change. In other words, whenever we talk about science fiction, we are effectively using a private language.

The amazing thing is that there is still enough overlap in our understanding of the terms for us to know, more or less and in most cases, what other people are talking about. The reason for this, I suggest, is best understood by taking a different approach to the linked questions of the definition and the starting point for the genre.

It is a neat and rational idea to imagine that we can look at science fiction, identify within it the various necessary elements that define the genre, and trace these back to find the earliest single instance in which these necessary lineaments are unequivocally combined. But reality is rarely so neat or so rational as this model might suggest. How do we identify which elements are necessary and which are sufficient to define the genre when, as I have noted, we cannot agree on a definition in the first place? Which combination of these elements has to be in place for that earliest instance to be unequivocal?

The truth is that this model could only work if science fiction was one thing, if every instance of the genre shared 99% of the same DNA. But if we take as the DNA of a genre the various story elements that we tend to hold up as identifiers of science fiction – for the sake of argument, these might include such disparate and often imprecise things as hardware, setting, theme, authorial intent, feel – it is possible for two stories to share none of the same DNA and yet have readers willing to identify them as science fiction: The Time Machine by H.G. Wells and Pavane by Keith Roberts, for instance. In Damon Knight’s terms, we can confidently point to both these works, along with, for example, Report on Probability A by Brian Aldiss, Counter-Clock World by Philip K. Dick, The Affirmation by Christopher Priest, The Female Man by Joanna Russ, and proclaim: yes, that is science fiction. But we could point to nothing that makes every one of these science fiction at the same time and in the same way. And it is worth remembering that no story is wholly science fiction. Even stories like ‘Day Million’ by Frederik Pohl or ‘Love is the Plan, the Plan is Death’ by James Tiptree Jr – works which seem comprehensively to occupy the science-fictionality of their worlds – will share something, be it use of language, characterisation, satirical intent, or whatever, which still links it with non-science fiction works. So we can not extract a unique, common thread which we could trace back to a unique, common origin.

Science fiction in particular, and probably all genres in general, do not work that way. A genre does not emerge, entire and fully armed, from the body of literature. A better analogy might be evolution by means of natural selection. There is an inchoate mass of story, each individual writer struggling with each individual story to produce something that will succeed, that will sell, or will please an editor, or will please a reader, or will make a particular point, or will work in a formal experimental sense, or, more likely, that will do several or all of these and perhaps more besides. In order to do this they might use ideas or themes or settings picked up from other writers, or which are a reaction against those of other writers; they might distort something old and familiar or invent something entirely new; they might take bits and pieces from a dozen different sources and recombine them in a novel way or regard them from a novel perspective. The exact details of this evolutionary process need not concern us, but eventually enough writers will be producing work that is sufficiently similar for us to start recognising patterns.

Once we have this identifiable pattern, and we have given a name to it (let us, for the sake of argument, call it ‘science fiction’), some people will work strictly within the pattern, others will deliberately avoid the pattern, still others will occupy a vague hinterland part in and part out of the pattern, while there will be yet more who cross the borders working within or outwith the pattern as the inspiration takes them. Yet none of them, even those working strictly within its boundaries, will replicate the pattern precisely in every instance. Were they to do so, they would be writing the same book; as long as writers are writing different books they will be in a constant process of taking different things from and adding different things to the pattern. The pattern, the genre, is hence in a state of constant flux.

Having already appropriated the phrase ‘private language’ from Ludwig Wittgenstein’s Philosophical Investigations, I now want to borrow another of his terms: ‘family resemblances’. When he was looking at the way language is used, Wittgenstein had a problem with the word ‘sport’. We use the word to identify a clearly understood set of activities, but what it is that identifies those activities as sport is not so clear. Some sports use a ball, but not all do. Some use some form of racket or bat, but many don’t. Many involve strenuous physical exercise, but some (diving, snooker, target shooting) don’t. Some require acute hand-eye coordination, many don’t. Some demand brute strength but not grace, others demand grace but not brute strength, still others don’t demand either. And so on. For every identifying characteristic, there are activities which we call sports which do not possess them (and there are activities which we do not call sports, certain games for instance, which may possess them). In other words, the more closely we look at what the word ‘sport’ means, the fuzzier it becomes; but we are all quite clear in how we use the word, and we are confident when we say ‘X is a sport, but Y isn’t’ (we may disagree on particular instances – is synchronised swimming a sport? – but that doesn’t affect our confidence in using the word). Wittgenstein’s conclusion was that sports bear ‘family resemblances’ to other sports. To simplify the process, we might recognise that a sport with characteristics A, B, C and D bears a family resemblance to a sport with characteristics B, C, D and E, and this in turn has a family resemblance to a third sport with characteristics C, D, E and F, and so on. It is because we recognise this network of resemblances that we are able to use words like ‘sport’ which stand for a wide variety of very different instances. It is also easy to see that we can trace this network of resemblances to discover a relationship between two sports which actually have no individual characteristics in common.

The analogy with science fiction is, I hope, obvious. Again, we have a term, ‘science fiction’, that we use to apply to a wide variety of individual works and groups of works, some of which have no obvious characteristics in common with others. Again, we are confident, as Damon Knight implied, in using the term to say: ‘X is science fiction, but Y isn’t’ (and again, we may disagree over individual instances – is the film Apollo 13 science fiction? Is the novel Perdido Street Station by China Miéville? – but that does not affect our confidence in using the term). Thus ‘science fiction’ is as amenable to the idea of family resemblances as ‘sport’ is, and I would suggest that it is a more productive way of looking at our use of the term than is any attempt at definition.

A definition attempts to fix the pattern that applies to science fiction, but the pattern, as I have shown, is in constant flux, and no definition has successfully managed to encompass all that it is, all that it has been, and all that it might be. Family resemblances are more flexible, since they allow us to keep pace with every change in the genre. A radical new work that takes science fiction in an unexpected direction would not require a redefinition; all that is required is that it bear a family resemblance to another work that we commonly agree is science fiction. (If the new work is so radical that we cannot agree that it bears any resemblance to any other work of science fiction, then perhaps we must concede that it is not science fiction; but such an instance would take us beyond the purview of this essay.) Thus, family resemblances recognise science fiction as a restless, dynamic form that might head out in multiple different directions from multiple different origins, and yet still be something that we can talk about sensibly under the one heading: science fiction.

If all that is required for us to call a work science fiction is that we recognise it is a member of the same family as another work which we have already identified as science fiction, it might seem that we are simply pushing the question of definition back one stage. We cannot say that A is science fiction because it resembles B, which in turn resembles C, which resembles D, which in turn … etc. This is an endless regress that tells us nothing, and if all that family resemblances are doing is embarking on such a regress it is as useless as any attempt to define the genre. But, in fact, all we need as a starting point is common agreement that something is science fiction. What makes it science fiction may, and probably will, vary from instance to instance, and is in fact irrelevant for our purposes here. It is not important why we agree that X is science fiction, it is only important that we agree. And there are many hundreds, indeed thousands, of works whose identity as science fiction is not problematic. We do not engage in heated arguments about whether The War of the Worlds or The Moon is a Harsh Mistress or I, Robot is science fiction, because such arguments would be fruitless. It is not in the heartland of science fiction that definitions, or family resemblances, are an issue, but on the borders, where science fiction is changing into something else, or something else is changing into science fiction.

Why certain works are unequivocally part of the heartland of science fiction may be historical accident; it may be as simple as certain editors (Hugo Gernsback, John W. Campbell) buying stories of a familiar type which were published in magazines labelled ‘Science Fiction’. Whatever the precise details of the case, what has undeniably happened is that certain works with certain characteristics in common have been published so consistently under the label science fiction that they have come to be seen as representative of that label. Consequently, there are any number of works – ‘The Cold Equations’ by Tom Godwin, ‘The Rose’ by Charles Harness, Fahrenheit 451 by Ray Bradbury – that, for most purposes and in most circumstances, we would not dream of labelling anything other than science fiction. We have no problem in identifying these and countless similar stories as science fiction, so much so that the question of what actually makes them science fiction never arises. As we move away from this heartland, however, the science-fictionality of individual works becomes less clear-cut, less unquestioned. Is the Canopus in Argos sequence science fiction despite the fact that it was written by an author, Doris Lessing, not normally associated with the genre, and who has deliberately eschewed the term in favour of her own coinage, ‘space fiction’? Is Perdido Street Station science fiction despite the fact that it makes significant use of devices more commonly associated with genre fantasy or supernatural horror? Is Frankenstein science fiction despite the fact that it predates our commonly accepted use of the term by a century and its tropes have been most successfully taken up by the horror film?

It is precisely here, where questions of our use of the term science fiction are most pertinent, that definitions of science fiction are generally least valuable. By establishing rigid formulations designed, of necessity, to encompass the heartland, they become questionable and often counter-intuitive just where they are most needed, on the borderland. By tracing family resemblances, however, we have no problem with whether or not the name ‘science fiction’ can be applied, even to works which actually cross the imprecise and ever-changing borders of the genre.

Moreover, it is rare for a work, even in the heartland of genre, to be all of one thing or all of another. A novel such as The Caves of Steel by Isaac Asimov, for instance, is clearly and unequivocally science fiction, but it also clearly and deliberately partakes of the detective story. It is a recognisable member of two different genre families. But there is no problem in saying that in respect of characteristics A, B and C it resembles science fiction, and hence discussing it as a work of science fiction; that in respect of characteristics X, Y and Z it resembles detective fiction, and hence discussing it as a work of detective fiction. Whether the science fiction, or any other genre aspects, are done well or ill is beside the point; that they are there at all is what we recognise in a work. By the same token, in terms of the beauty or otherwise of the language employed or the characterisation, or the scene setting, or any of countless other qualities, we might equally recognise a novel as a work of science fiction and at the same time more broadly as a work of fiction. Again, a novel like Perdido Street Station, which partakes of elements of, among others, science fiction, fantasy and horror, can be included in the discussion of science fiction in relation to those elements which resemble science fiction, just as it can be included in the discussions of fantasy and horror in relation to those elements which resemble fantasy and horror.

What I am proposing is a development of Damon Knight’s ostensive definition of science fiction. That is science fiction which we point to and say: it has family resemblances with what we agree is science fiction.

One thing to recognise, therefore, in this web of resemblances, is that one work might bear different resemblances to many other works. And any number of those resemblances might constitute what we would call science fiction. By thinking of science fiction as a network of such family resemblances, it is easier to see that science fiction is not one thing. Rather, it is any number of things – a future setting, a marvellous device, an ideal society, an alien creature, a twist in time, an interstellar journey, a satirical perspective, a particular approach to the matter of story, whatever we may be looking for when we look for science fiction, here more overt, here more subtle – which are braided together in an endless variety of combinations. A newcomer to the genre might find these combinations unsettling, hard to unravel, formulaic; but the more familiar we are with the genre, the more readily we can accept their variety, the more subtly we might interpret their combinations. So much so that at times we might identify a work as science fiction for no other reason than that ‘it feels like science fiction’; the furniture of a story, the obvious plot devices have become subsumed under our interpretation of authorial intent. What constitutes the warp and weft of science fiction, therefore, is endlessly subtle and intricate, made up at times of more things than we can readily identify. Which is why what makes science fiction is so hard to pin down, but what is science fiction is so easy to recognise.

What is more, the web of resemblances extends backwards in time as well as forwards. If we recognise a new work as science fiction because it resembles an earlier work of science fiction, so we can go back and recognise an historical work as science fiction because it resembles works we would later come to call science fiction. Supposing we accept Gary Westfahl’s contention that science fiction only really began in 1926 when the term (or at least Gernsback’s coinage, ‘scientifiction’) was first applied to what we now recognise as genre sf; we can still identify certain works of the 1890s – The Time Machine, The War of the Worlds, The Invisible Man – as science fiction because they resemble so closely works that are clearly science fiction. Tracing the resemblances ever backwards will not lead to that original ur-text, of course – there is no such a thing – but it does lead, rather, to a series of ur-texts.

Science fiction, as I have described it, consists of a series of threads (themes, devices, approaches, ideas) that are braided together. This is cognate with David Seed’s argument, in Anticipations, that science fiction, rather than being one genre, is actually a series of sub-genres that have come together over time. But rather than thinking in terms of distinct sub-genres, I want to suggest a series of strands, none of which would stand as a genre or even a sub-genre in their own right but which, braided together in any of a possibly infinite number of combinations, make what we have come to recognise as science fiction. Any one of the threads might be removed from the braid and it would still be science fiction. The threads that make up the braid might be separated and then re-wound to make two separate braids, both of which are science fiction. But there is not one single thread that can be removed and which in itself is science fiction. Similarly, there is probably no critical point at which we can say: this is the minimum number of threads required for us to call a work science fiction; or: these particular threads are the ones that must be included if the whole is to be called science fiction. Tracing family resemblances backwards through the literature, therefore, is not going to lead us to the origins of science fiction; but it might lead us to the origins of some of the threads that constitute science fiction.

Thus, there are strands extant in contemporary science fiction which may have their origins in classical, or pre-classical, mythology. There are threads which may go back to identifiable individual works, to the satirical celestial journeys of Lucian of Samosata, for instance, to Thomas More’s Utopia, to Mary Shelley’s Frankenstein, to countless other works by Copernicus, by Cyrano de Bergerac, by Jules Verne or Edgar Allan Poe or H.G. Wells. Whether this necessarily makes any of these works the ur-text of science fiction, or whether it makes them science fiction at all is open to debate, but they certainly resemble aspects of what we would come to call science fiction, they might be considered the starting point for threads that would come to constitute science fiction. Where the braid of these threads reaches the critical point at which it unequivocally becomes science fiction is impossible to say. All we can say is that science fiction starts here … and here … and here … and here … Which of these particular starting points we choose, therefore, comes down not to a question of definition but of inclination.

Science fiction is what we point to when we say ‘science fiction’, and where the genre begins historically and what constitutes that genre will vary as the direction in which we point varies. But because we can see the resemblances between works of science fiction, because we can identify the various threads that combine to form the whole, so we can talk sensibly about the genre and understand others when they do the same, and so we can draw an historical model for the genre in which the details may vary but the overall narrative tells a story we all understand.