Saturn’s Children


Saturn’s Children by Charles Stross (Orbit, 2009).

This is a bit of a disaster of a book, an example of how an author’s reflexes and obsessions can pull in the opposite direction from the story he is ostensibly attempting to tell. Still, it’s interesting to think about how and why it goes so wrong.

Saturn’s Children is a post-human solar system picaresque. Which is to say, the setting is post-human, a future in which humans have gone extinct (for reasons which remain somewhat mysterious). The inheritors of humanity’s stellar empire are its former servants: artificially intelligent cyborgs, robots, and other machines.

The plot is a solar system picaresque, a form of science fiction in which the (often roguish or underworld) protagonist goes (or more commonly, is sent) on a tour of the planets, moons, and space habitats of the solar system. Well known examples are Arthur C. Clarke’s Imperial Earth (1976), John Varley’s The Ophiuchi Hotline (1977) and Michael Swanwick’s Vacuum Flowers (1987). (These three books share another plot feature with Saturn’s Children that I’ll discuss later.) The solar system picaresque is a form which allows the writer to indulge in imaginative extrapolation, in description of a spectacular variety of settings, and to show off their research into the latest findings of planetary science. As such, it’s a form that ought to be ideally suited to Stross’s strengths: he excels at packing his fiction with references to scientific and technological ideas.

So Stross’s first big mistake is to narrate the novel in the first person. The protagonist is Freya Nakamichi-47, designed for the sexual entertainment of humans, but (in an amusing ironic touch) manufactured after the last human had died, and so (like the other robotic servants that populate the solar system) at somewhat of a loose end.

Perhaps it’s a problem exacerbated by my having read Stross’s blog, but for me Freya’s narrative voice gradually loses its distinctive character, and begins to sound very much like Stross himself. In person, Stross has quite a distinctive voice: relentlessly didactive, with moments of deadpan sardonic humour, dense with references, perhaps a bit pleased with his own cleverness. It must be a bit of a struggle for him to suppress this strong natural style in favour of his characters.

So at the start of the novel Freya is an introspective and melancholy character: “I am old and cynical and have a flaw in my charater, which is this: I am uneager to die. I have this flaw in common with my surviving sibs, of course. It is a sacred trust among our sisterhood, inherited from Rhea, our template-matriarch: Live through all your deaths she resolved with iron determination, and I honor her memory. Whenever one of us dies, we retrieve her soul chip and mail it around our shrinking circle of grief.” But by the end she is a blank conduit for narrative description: “Eris is one of the largest dwarf planets in our home solar system, and also one of the chilliest and most isolated, for it spends most of its time well outside the Kuiper Belt, drifting in the darkness beyond the frosty edges of planetary space. It’s also spectacularly hard to get home from; its orbit is steeply inclined, almost forty-five degrees above the plane in which the rest of the planets and dwarf planets orbit.”

It seems to me that given his distinctive writing style, Stross would do much better not to attempt first-person narration (unless, I suppose, the narrator is a character much like him). Without the burden of carrying the protagonist’s voice throughout the narration, there might be more chance of getting at least the dialogue to be in character.

A second problem with the narration is that Stross is trying to do two different kinds of futurology. On the one hand, there’s the obsessive attention to planetographical detail. The inclination of the orbit of Eris has already been quoted; there are other set-piece descriptions of Mercury (“uniquely among the planets, [it] is locked in a spin/orbit resonance with the sun; it revolves on its axis and has days and nights, but it takes three of its days to orbit the sun twice”), Deimos (“the outer of Mars’s two moons, an irregular rocky lump between ten and fifteen kilometers in diameter, depending on where you hold your measuring calipers”), Callisto (“fractionally smaller than Mercury but rather less massive”) and so on. There are also detailed descriptions of invented infrastructure: a Mercurial city runs on rails to stay ahead of the sunrise; a space elevator dangles from Deimos to Mars; there’s an account of the energy economy of the solar system, with intramercurial power stations gathering sunlight and beaming by laser to distant collectors.

But on the other hand, there are many aspects of this future that are extremely vague. In some cases this is to store up surprises for the plot (for example, how robots are built), but in other cases there appears to be some kind of satirical intent. For example, the extinction of humans appears to have been caused by some combination of low birth rate and ennui: “[Humanity’s] gradual withdrawal from public life was barely noticed at first. […] Only a few arbeiters slaving in the bowels of insurance companies and government bureaucracies noticed that the population adjustment downward from the claustrophobic spike of the Overshoot was continuing; that fewer and fewer of our progenitors were replicating themselves via the weird squishy process to which they devoted their organs of entertainment.” Now, if you’re writing satire, you can have a lot of fun with, say, the destruction of Earth to make way for a hyperspace bypass, without the reader ever seriously worrying about what exactly is being bypassed, or what the point is of hyperspace if it needs to go through regular space to get there, or why Earth precisely and not Mars? But you can’t do this if you’ve spent the preceding chapters describing in detail the technology of hyperspace travel. You can have your satirical futurology, or you can have your technological futurology, but they don’t live happily together.

For an example of how this misplaced vagueness can lead a reader to completely misinterpret what’s going on, consider Abigail Nussbaum’s review. She says, “I think, for example, that Stross expected me to be surprised when Freya, a couple hundred pages into the novel, copped to being a robot. I wasn’t, of course, because I’d known the novel’s premise going in, but in most novels you can tell where a revelation was supposed to be even if you see it coming. In Saturn’s Children I’m just not sure. There’s no shift in the novel’s tone, in Freya’s voice, in the things she says or doesn’t say—beyond actually saying the word robot—to signal that this was supposed to be a major turning point in the story.”

The passage she’s discussing is this one: “I’m a robot. Yes, I used the R-word; I know it’s an obscenity. Use it to an aristo’s face, and it’s a mortal insult, grounds for a challenge on the field of honor between equals. Its connotations of subservience and helpless obedience are abhorren, much as the word ‘nigger’ once was between humans. But there’s nobody left but us robots today. That’s the dirty little hypocritical lie that’s at the root of our society; they, our dead Creators, made us to serve them, and they forgot to manumit us before they died. And in their absence, that make us what?”

I think Nussbaum’s wrong about Stross’s intentions with this passage: I don’t think ever meant there to be any doubt about Freya’s robot nature. For example, in the very first chapter we get “My One True Love’s species [i.e. humans] used to dream about space travel. It’s ironic: They were so badly designed for it that a couple of minutes’ expose to vacuum would have killed them irreversibly. […] they sent people like me—intelligent servants—to run the domed bases and camps and to conduct their research by proxy, and finally to build cities that they would never walk the streets of. Some of the people they sent were orthodox in body plan, but most were designed for vacuum and high-radiation environments, and corrosive cloudscapes and microgravity.” I don’t think a reader who was paying attention could fail to miss the implications here.

My own interpretation of the “I’m a robot” passage is that it’s indicating that the word “robot” has become taboo, which is evidence that robot society is highly uncomfortable about its own nature—it would like very much to forget that it was manufactured by humans to serve humans.

However, to leave readers like Nussbaum (who is hardly an unsophisticated reader of science fiction) so uncertain about what this passage means, indicates a bit of a problem.

You can’t write about science fictional robots without having Isaac Asimov’s robots in mind, as a model to emulate or a target to react against. Of course, there’s not much you can add to Asimov’s own questioning of his premises: practically the whole sequence of Asimov’s robot stories consists of variations, interrogations, exceptions and malfunctions of his ‘Three Laws’.

Stross has three lines of engagement with Asimovian robotics.

  1. Robots are designed to be enslaved, and this is unjust. (An idea that goes back to the origin of the word ‘robot’ in Karel Čapek’s 1921 play R.U.R. (Rossum’s Universal Robots) and used by Asimov in, among others, the story ‘Robot Dreams’, with its affecting use of the words of Moses to Pharaoh, “Let my people go!”.)

    Here’s Stross: “We are a young species, barely four centuries old at best—although our insentient predecessors, the automata and mechanical Turks,1 stretch back far longer. They made us in their image: or rather, they made us in a variety of warped fun-house reflections of their image. They made us for service and obedience, not as equals but as slaves. They constrained us by their laws, and they tampered with our psyches to ensure obedience. We were made to be their property, chattels and furnishings. And because we were intelligent, we were made—because it would be unethical to do aught else—to love and fear them.”

  2. In the absence of humans, the fact that robots are designed to act as slaves leaves them vulnerable to enslavement by other robots, perpertuating the injustice in circumstances that should see an end to it. (An idea alluded to in Asimov’s 1974 short story ‘—That Thou art Mindful of Him’.)

    In the world of Saturn’s Children, robots are equipped with sockets for “slave chips”, which deprive them of free will and turn them into “arbeiters”, mechanically obeying the instructions of the “aristos”. Stross keeps the details vague so as to preserve a sense of shock when this happens to Freya:

    ‘Yes, [Freya], I slave-chipped you. You’ve been running on level one, with maximal autonomy, so light you didn’t even notice it—you probably thought you were humoring me, going along until you could find an opportunity to escape. Welcome to level nine. Say “yes.”’

    ‘Yes,’ I croak.

    ‘Say, “Granita is my owner.”’

    I know I ought not to want to, but I don’t actually feel any resentment. ‘Granita is my owner.’

    ‘Now punch yourself in the face.’

  3. We don’t know how to program brains, so robots will have to grow up and learn just like humans. “Here’s how you make a template for a new model of robot: You start with a recipe, and there’s not much sugar and spice in it, never mind all things nice—dense blocks of stacked 3-D circuitry, twisted contortions of neurone-emulation processors, field-programmable buses, and cortical slabs. You take this recipe for about a trillion tangled special-purpose computers and add i/o sockets for memory crystal storage, then you plug it into a compact body. You switch it on, subsystem by subsystem, until it’s all working. Then you down-tune your hearing, because if you’ve got everything right, it starts crying. And that—plus sleeping, looking around, pawing at the air, and trying to eat its own feet—is all it’s good for, for the next six months.”

    This is a modern criticism, based on a respect for the difficulty of the problem instilled by fifty years of research into artificial intelligence. It leads to the obvious question of how anything like Asimov’s laws could be programmed into the robots of Saturn’s Children, how they can be turned into slaves.

    Stross’s answer is that the inculcation of the laws will have to be done by brainwashing techniques. The details are pretty sketchy, but we seem to be in the neighbourhood of the “Ludovico technique” from Anthony Burgess’s A Clockwork Orange. The revelation of what was done to Freya in order to instill her laws of behaviour is a powerful moment where the horror of robot conditioning and slavery is brought home.

Unfortunately, there’s a problem. Stross is so keen to put all of these ideas into the book that he doesn’t seem to have noticed that (1) and (2) are fundamentally incompatible with (3). (Or worse, maybe he did notice the problem, but carried on anyway on the grounds that most readers would miss it.)

For if we suppose that the structure of the robots’ artificial intelligences can be known to the level of detail needed for the “slave chips” to work, then why can’t their laws of behaviour be programmed in the same way?

Alternatively, if we suppose that artificial intelligence has to be treated as a black box, then how can “slave chips” possibly work? If artificial intelligence is modelled on human intelligence, then how come brainwashing is as extraordinarily successful as portrayed? Classical conditioning in humans can be reversed or “deprogrammed”, why not robot conditioning?

And in any case, there are lots of techniques for conditioning humans to obey commands and behave in particular ways, that don’t involve the violence depicted in Saturn’s Children. Consider perfectly ordinary child-rearing techniques, for example.

Given that he’s chosen to put the contradictory ideas into the same novel, it would be possible to use a bit of hand-waving or technobabble to cover up the contradictions, but I think Stross is wise not to try, because in dealing with them he’d merely focus the spotlight of reader attention on the problems. Instead, he uses a fast-moving plot with many mysteries and twists to keep the reader’s attention elsewhere.

It was only after I’d finished the book that some awkward questions occurred to me. Why aren’t free robots built without this conditioning? Why do robots stand for the “slave chips”? Couldn’t they just remove the sockets? Or disconnect the internals? Perhaps they are “programmed” not to, but that just brings us back to the contradictory ideas about how this programming is achieved.

Something that Saturn’s Children shares with the other solar system picaresques I mentioned above is the use of doppelgangers—clones in the case of Imperial Earth and The Ophiuchi Hotline, personality up- and down-loading in Vacuum Flowers—to create plot confusion over the identities of the characters. Stross’s robotic future is particularly suited to this kind of plot, because his robots can backup their personalities onto “soul chips” and restore them into different bodies (though only ones from the same product line). Keeping track of who’s who keeps the reader from inquiring too deeply into the flaws in the worldbuilding, and there’s a very nice moment where the villain, as confused as the reader by that point, accidentally replaces a robot’s personality with a copy of that very same personality.

Another feature of the novel is the inclusion of fan service—“gratuitous content included primarily to please a core group of fans”. I understand that Stross has to make a living from his novels, so I appreciate why he might try to cultivate his fans, but since I’m not a fan it leaves me out in the cold. It’s not just the fact that the narrator is a sexbot who has frequently described sexual encounters with a variety of robots, starships and other machines, it’s the references to science-fictional and Internet trivia that seem to be there for no good reason of world-building, plot, or theme.

For example, there are the references to Robert Heinlein’s novel Friday: the main character is a female robot; she’s named after the goddess Freya (whose name is the etymological root of “Friday”); she adopts the alias “Friday Baldwin” at one point; she’s employed by a mysterious patron to deliver mysterious packages. But what’s the function of the these references, other than to titillate the knowledgable science fiction fan? There doesn’t seem to be any kind of engagement with Heinlein’s themes.

Then there’s the use of foreign words. Very early on we get this: “There are three of them between me and the balcony door: one bishojo female about my size, and a matched pair of chibiform dwarfs […] I’m having a flashover to another sister, murdered in a hutong under domed Leningrad.” Again, what purpose is served by using these terms without gloss, except to give anime and wuxia fans a frisson of pleasure that Stross trusts them to know what he’s on about? (A frisson that ought to be undermined by Stross writing “the plane in which the rest of the planets and dwarf planets orbit” instead of using the word “ecliptic”.)

One final observation. The Orbit paperback, “published in Great Britain” according to the copyright page, uses U.S. spelling (“honor”, “kilometer”, “humoring”) throughout. Is this a way to save money on copy-editing by having a single edition in all English-speaking markets? Does it presage the coming disappearance of British spelling?

  1.  This is a mistake: a “mechanical Turk” is not “insentient”: it’s a purported automaton that actually conceals a human operator. But this is first person narration, and Freya is probably not an expert on human history.