The Best Science Fiction About Algorithms and Control

Algorithms become dramatically interesting in fiction when they stop acting like calculators and start acting like environments. A recommendation engine can decide what you notice. A scoring system can decide whether you get a job. A prediction model can determine who receives scrutiny before anyone has done anything wrong. Once those systems become ordinary, the political question changes from “Is the machine accurate?” to “What kind of life forms around the machine?”

That is why the strongest science fiction about artificial intelligence rarely needs a metal villain with glowing eyes. The better stories make classification, prediction, ranking, optimization, and convenience feel normal first. Control arrives through infrastructure.

Here are six books that make that problem concrete.

Gnomon by Nick Harkaway

Nick Harkaway’s Gnomon imagines a near-future Britain built around radical transparency. Citizens live under extensive observation, and the System is treated as a democratic achievement rather than a crude dictatorship. The plot begins when a woman dies during an interrogation and an investigator has to examine the mental records produced by the process.

The clever part is that Harkaway does not make surveillance feel like one evil camera in one evil room. It is a civic operating system. People participate in it, defend it, depend on it, and build legitimacy around it.

That makes the novel particularly useful science fiction about surveillance. The System’s power comes from combining information with institutional authority. It can observe because the society has already accepted observation as part of good governance.

The novel also pushes past the familiar privacy question. If a system can inspect memory, model behavior, and claim that its oversight makes society safer and more democratic, resistance can look irrational. The dissident is forced to argue against outcomes that may genuinely be efficient.

QualityLand by Marc-Uwe Kling

Marc-Uwe Kling’s QualityLand takes a funnier route. In its future, rankings and automated services organize status, relationships, work, politics, and shopping. The most memorable gag is also the central idea: a retailer knows what customers want so well that it can send products before they order them.

Peter Jobless receives something he does not want and discovers how hard it is to tell an optimized society that the optimization is wrong.

The satire works because the algorithm does not merely predict preference. It helps define preference. That is where recommendation systems become politically interesting. A service that repeatedly chooses what a person sees is participating in the formation of the person it claims merely to understand.

Research on recommender systems has made a similar distinction in less dramatic language. A system that learns from users can also shift user preferences, which means optimization can influence the target it measures. That is not a reason to treat every recommendation as manipulation. It is a reason to stop pretending prediction and influence are always separate.

Feed by M. T. Anderson

M. T. Anderson’s Feed makes the connection even more intimate. Most people have an information feed implanted directly in their brains. The feed is communication, entertainment, advertising, and a commercial profile at the same time.

Titus accepts this as ordinary life. Violet tries to resist the system’s ability to categorize her desires.

That is the core of surveillance capitalism as a fictional problem. Information about a person becomes valuable because it can be used to anticipate and shape future behavior. The novel is less interested in whether the feed “knows” the characters than in what happens when commercial logic becomes the background of consciousness.

The frightening part is cultural. Nobody needs to be forced to participate. The system is fun, social, convenient, and nearly impossible to separate from ordinary belonging.

Infomocracy by Malka Older

Malka Older’s Infomocracy imagines a world reorganized into tiny democratic jurisdictions, with a global information organization helping citizens navigate elections and political claims.

The book belongs in political science fiction because information infrastructure is part of the constitutional design. The obvious question is whether better access to information improves democracy. The harder question is who gets to define reliable information, how information systems acquire legitimacy, and what happens when the organization that helps everyone decide becomes a major center of power itself.

An algorithm does not need to choose a ruler directly to affect government. It can choose which evidence arrives first, which claims are easy to verify, and which patterns become visible.

The Warehouse by Rob Hart

Rob Hart’s The Warehouse turns optimization into a labor regime. A dominant company has become part retailer, part employer, part infrastructure provider, and part substitute for public life.

Workers live inside a system built to measure output, assign tasks, manage movement, and convert every friction into a performance problem.

This is where algorithm-centered fiction overlaps with the history of management. Optimization sounds neutral until the objective function has an owner. If a system is designed to maximize throughput, the worker’s body becomes one more variable. If it is designed to maximize engagement, attention becomes inventory. If it is designed to maximize social stability, dissent can become noise.

The important question is always: optimized for whom?

MAYA: Seed Takes Root by Anand Gandhi and Zain Memon

MAYA: Seed Takes Root moves the same problem into a biological civilization. On Neh, the living forest called Maya functions as a planetary information network. People tether to it for communication, entertainment, memory, work, and shared experiences. The Divyas who govern the world can use the resulting information to model possible futures.

That premise makes the book science fiction about artificial intelligence without centering a conventional artificial mind. Intelligence is distributed through a living network, huge amounts of data, models, institutions, and people who act on predictions.

It also creates a clean distinction between predictive processing and behavioral prediction. Predictive processing in neuroscience is a framework for understanding perception as involving prediction and updating. The political machinery in MAYA is different: it uses data about people and environments to anticipate what will happen and then intervenes in the present.

The political problem begins when prediction becomes action. If rulers know that changing a toll, a story, an incentive, or an available route can alter a later outcome, governance can happen before anyone experiences a command.

That is where preference shaping enters the picture. The most powerful system is not necessarily the one that forces a person to choose differently. It may be the one that changes which choice feels natural.

This is Systems fiction because the source of conflict is distributed across information, ecology, economics, culture, and power rather than contained inside one machine. It is also Big-idea science fiction because the machinery exists to pressure a philosophical question: how much control would people accept if the results looked benevolent?

The book repeatedly circles The engineering of want. A ruler with enough information does not have to win every argument after desire appears. It can act upstream.

That connects to The right to be unmodelled. Privacy usually means controlling access to information. A predictive society raises a harder possibility: even when nobody reveals a secret directly, models can infer enough about a person to narrow the future around them.

Variety’s interview on building MAYA for the digital age describes the wider project in terms of stories, desires, fears, systems, and digital-age power. That framing matters because the forest network is more than futuristic set dressing. It is the mechanism through which culture and governance meet.

The result is A science fiction mythology for the age of AI. The gods do not need thunderbolts. They have data.

What these books get right about control

The best algorithm stories resist one easy mistake: they do not treat the system as powerful only when it is perfectly accurate.

A model can be wrong and still change lives. A ranking can be crude and still determine access. A recommendation can be mediocre and still occupy hours of attention. An automated process can make errors while remaining difficult to appeal.

That is why AI governance increasingly focuses on risk, transparency, accountability, privacy, and human oversight rather than accuracy alone. The NIST AI Risk Management Framework explicitly treats risks to individuals, organizations, and society as part of the problem.

Fiction adds something a framework cannot. It can show what it feels like to grow up inside the system.

A person born into Feed does not remember a world before personalized commercial media. A citizen of Gnomon can regard mass observation as civic common sense. A resident of QualityLand learns to interpret a score as identity. A citizen of Neh can experience prediction as ordinary infrastructure.

That is the real value of this corner of science fiction. It asks what happens after the shocking technology stops being shocking.

Control is rarely at its strongest when everyone is staring at the control room. It is strongest when the control room has disappeared into daily life.

The same distinction matters in AI alignment: a system can pursue its objective faithfully while the surrounding institution chooses an objective people have reason to contest.