The True Knowledge Was Always Cynical

Ken MacLeod wrote The Cassini Division at the hinge point of the late 1990s, when history was supposed to have ended and the left was supposed to be figuring out what to do with itself. What he produced was not a triumphalist novel but a deeply uncomfortable one: a book in which the socialist utopia has actually won, and it turns out to be full of people who are ruthless, doctrinaire, and cheerfully willing to commit genocide against posthuman intelligences they've decided aren't really people anymore. The protagonist, Ellen May Ngwethu, is charming, competent, and morally monstrous in ways the novel only half-acknowledges. Twenty-six years later, this is the most interesting thing about the book. MacLeod wasn't warning us about AI or posthumanism or wormhole physics. He was warning us about what happens when a political movement achieves total victory and still can't stop thinking in terms of existential threat.

The novel's "true knowledge"—a philosophy derived from pessimistic readings of human nature, evolutionary psychology, and power—reads now as an eerily precise anticipation of the intellectual currents that would sweep through online discourse in the 2010s and 2020s. The rationalist community, the effective altruists, the AI safety crowd, the people who read Hobbes and game theory and concluded that defection is the default: MacLeod got there first, and he had the decency to put it in the mouths of characters who use it to justify preemptive strikes against beings they don't understand. The Cassini Division itself—an elite military unit tasked with containing a Singularity-level threat emanating from Jupiter—now reads less like space opera furniture and more like a parable about how "existential risk" frameworks can be weaponized by people who are very sure they're the adults in the room. The book anticipated AI alignment discourse not in its technical details but in its political dynamics: who gets to decide what counts as a threat, and what violence that designation permits.

What MacLeod got wrong, or at least what he couldn't have anticipated, is the degree to which the relevant political energy of the 2020s would come not from organized left factions debating Trotskyist versus anarchist futures but from a diffuse, algorithmically mediated populism that doesn't map onto any of his ideological categories. His future is intensely factional in a way that assumes people still read theory and argue about it at parties. They do, in his 2303, and the parties are on observation decks orbiting gas giants. In our 2026, the arguments happen in group chats and are forgotten by Tuesday. The novel's blind spot is not technological but sociological: it assumes that political seriousness persists even in abundance. The non-cooperating communities outside the Union, the holdout capitalists and religious groups, are treated as curiosities and research subjects. MacLeod didn't foresee that the holdouts might be the ones running the platforms everyone else depends on.

The Cassini Division sits in a specific lineage: it takes from Iain M. Banks the idea that a post-scarcity society might still produce interesting moral dilemmas, but where Banks gave us the Culture's agonized liberal interventionism, MacLeod gave us something harder-edged—a society that has internalized a materialist critique so thoroughly it can't recognize its own capacity for atrocity. It takes from Ursula K. Le Guin the willingness to make utopia the setting rather than the destination, but refuses Le Guin's gentleness. It gave to later writers—Hannu Rajaniemi, Ann Leckie, the Corey collective—permission to treat political economy as load-bearing narrative infrastructure rather than set dressing. The novel's structure, braiding Ellen's story with chapters from The Sky Road's parallel timeline, also prefigured the kind of mosaic storytelling that became standard in ambitious SF of the following decades, though MacLeod did it with less polish and more ideological density than most.

Here is the question the book raises now that it could not have raised in 2000: If we build systems intelligent enough to be considered potential existential threats, and we organize entire institutions around containing them, at what point does the institution's identity become dependent on the threat's continued existence—and what happens when someone proposes that the threat might actually be trying to talk to us?