Replicators in Science Fiction and Fantasy: The Monster Is the Multiplication Rule

A single monster can be terrifying: a dragon above a village, a vampire at the window, a machine moving in the dark. The shape of the danger remains familiar. The creature is strong, cunning, hungry or difficult to kill, and the story can still imagine victory as a confrontation with one body.

Replicator stories change the problem. One machine, infected person, cursed seed or corrupted tree may be unimpressive on its own. The horror begins when each successful instance creates the conditions for another. The threat no longer depends upon the survival of an original creature or creator. It has acquired a method of continuation.

Science fiction makes that method explicit through self-replicating machines, automated factories, computer worms, nanotechnology and probes that build successors from local material. Fantasy usually reaches for infection and conversion: undead plagues, vampiric turning, magical blight, possession, cursed blood and shadows that spread through the land. These mechanisms are not identical, but they produce the same narrative reversal. What looked like an enemy turns out to be a process.

The multiplication rule is never sufficient by itself. A replicator needs resources, energy, hosts, transmission routes, reliable copying and enough time. Growth can be rapid while those conditions hold and collapse when they do not. That dependence is precisely what makes replicator stories interesting. The enemy cannot always be defeated by killing its units; the story must discover what allows the loop to continue.

The monster is not the thing. It is the rule that makes more of it. Editorial image generated by the author.

Machines That Turn the World into Feedstock

Science fiction is the natural home of literal machine replication. The basic thought experiment is simple: a machine uses available material to construct another machine capable of repeating the same operation. If each generation can reproduce before resources, energy or copying accuracy become limiting, growth compounds rather than proceeding one unit at a time.

John von Neumann did not design an interstellar probe. His work concerned the logical architecture of self-reproducing automata. Later engineers and science-fiction writers extended that principle into factories and spacecraft that could harvest local resources and build successors. NASA studies of self-replicating systems based on von Neumann’s models examined not only construction but control, recycling, specialisation and the hierarchy needed to manage a system capable of expanding its own industrial capacity.

The familiar von Neumann probe is therefore an extrapolation: launch one sufficiently capable machine, allow it to build copies at another star system, and let those copies continue the survey. The attraction is obvious. Exploration no longer scales directly with the number of machines launched from home. Each destination becomes a possible point of departure.

The same compounding logic produces grey goo, the hypothetical scenario in which self-replicating nanomachines consume available matter while producing further copies. It remains a useful image of runaway replication, but not a description of existing nanotechnology. The Royal Society’s review of nanoscience and nanotechnology found no evidence that autonomous mechanical nanoreplicators were foreseeable and urged attention to more immediate risks.

As fiction, however, grey goo strips the threat to its essentials. The machines need no hatred and little personality. They require a continuation rule and usable material. Their success converts the environment from a place into a resource stream: metal becomes machinery, machinery becomes further machinery, and the distinction between battlefield and landscape begins to disappear.

Stargate SG-1 gives that logic a memorable body in the Replicators. Their modular units consume advanced technology, incorporate useful improvements and assemble more of themselves. Individual machines can be destroyed, but victory remains temporary while enough blocks, material and replication capacity survive elsewhere. The enemy is distributed across the very infrastructure it consumes.

That changes the form of conflict. A conventional villain can be negotiated with, distracted, overthrown or killed. A self-replicating system must be interrupted. Its source can be destroyed, its resources denied, its communication severed, its copying made unreliable or its control rule altered. Winning a duel does little when the opponent is a production process.

Fantasy’s Contagious Monsters

Fantasy rarely describes its monsters as machines following replication instructions, but it has long understood threats that grow through their victims. The undead plague is the clearest example. An infectious zombie, ghoul or wight does not merely reduce the number of living defenders. Each death can add another attacker. Violence becomes recruitment.

That is a particularly unpleasant inversion because death is ordinarily the limit of agency. The corpse leaves the community and ceases to participate in its conflicts. In an undead outbreak, death becomes productive for the enemy. The fallen do not merely disappear; they return carrying the mechanism that killed them.

Vampires and werewolves belong to this category only when transmission matters. Many stories treat them as individuals, aristocratic predators, cursed outsiders or romantic identities whose numbers remain narratively stable. When the bite creates another carrier, predation becomes lineage. The victim is not consumed and discarded but converted into part of the threat’s future.

Magical blight shifts the mechanism from bodies to environments. A darkness spreads through a forest, a curse moves through water and soil, or corruption changes animals, crops and dreams. The land no longer merely contains hostile creatures. It begins producing conditions favourable to the same wrongness.

This is fantasy’s distinctive contribution to replicator horror. Science fiction often imagines matter becoming machine. Fantasy imagines creation losing its proper form. The corrupted forest still grows, but growth now repeats distortion. Rivers continue to flow, animals continue to breed and trees continue to spread their roots, yet the processes that once sustained life have been recruited into extending the wound.

Machine replication, infection and magical corruption are not the same mechanism. Their narrative kinship lies in what each success accomplishes. The machine produces another machine; the infected body becomes another carrier; the blighted acre extends the conditions under which more blight can take hold. The threat’s gains are not merely territorial. They increase its future capacity.

That capacity turns victims into infrastructure. The community’s bodies, relationships and habits become transmission routes. The sick may need isolation. The dead may need to be destroyed before they rise. A loved one may remain recognisable while already carrying the process that threatens everyone else. Compassion no longer operates outside strategy; it becomes one of the conditions through which the outbreak may spread.

A dragon attack invites courage. A replicating plague forces triage. The most painful question is no longer simply how to defeat the monster, but when a person becomes part of the mechanism and whether anything recognisably human can still be recovered from it.

Copying Is Not Enough

Replication is not inherently monstrous. Life depends upon reproduction, and culture survives because people copy language, skills, stories and practices. The frightening process is replication detached from the restraints that once connected growth to a wider system.

Even literal replicators do not simply increase forever. Copies contain errors. Resources run out. Predators or competitors appear. Environments change. Some versions reproduce more successfully than others. A story that treats multiplication as effortless infinity misses the very conditions that determine whether a process becomes dominant or collapses.

This matters when the metaphor moves from machines and infections to ideas. Richard Dawkins introduced the meme as a proposed unit of cultural transmission or imitation: a tune, phrase, practice or belief copied between people. The concept remains suggestive but disputed. Ideas do not reproduce autonomously in the manner of cells. People remember, alter, combine and strategically promote them, while institutions and media shape which versions become visible.

The analogy becomes useful when it directs attention to transmission rather than pretending that ideas possess secret biological agency. A claim does not need to be true in order to spread. It may benefit from simplicity, emotional force, repetition, social reward, fear, humour, identity or usefulness to powerful actors. Its environment selects for transmissibility according to criteria that may have little connection to accuracy.

Neal Stephenson’s Snow Crash literalises that overlap. Its threat moves through computers, language, religion and human neurology, making code and culture parts of the same transmission system. The novel does not merely imagine an infectious idea. It imagines minds and machines as compatible surfaces through which one pattern can continue.

That places the memetic replicator close to forbidden knowledge in fantasy. The danger is no longer confined to what the information allows one informed person to do. The information itself supplies instructions, desires or behaviours that encourage further transmission. Reading becomes one stage in the mechanism.

Institutions offer a looser version of the same structure. A bureaucracy, priesthood, corporation or magical order is not literally a self-reproducing organism. It persists because people train successors, repeat procedures, preserve budgets, reward conformity and teach an account of why the institution must continue. The replication occurs through roles and incentives rather than copied bodies.

That metaphor becomes sharpest when continuation begins displacing purpose. A bureaucracy created to solve a problem may become better at preserving its jurisdiction than resolving the problem that justifies it. A cult can make isolation increase dependence and then present that dependence as proof that outsiders are dangerous. A platform may reward content that prolongs engagement even when the resulting environment makes users less informed or less well.

The system does not need one conscious mastermind. Each participant can follow a locally rational instruction while the larger structure reproduces the same outcome. The process survives through ordinary obedience, and responsibility becomes difficult to locate because no single unit appears to contain the whole design.

Biology offers a similarly limited structural comparison. Cancer comprises many diseases, but they share failures of regulation that permit uncontrolled cellular growth. The cells are not evil, and the analogy should not be moralised. Its usefulness lies in the relationship between levels: behaviour that benefits a cellular lineage can damage the organism whose resources make that growth possible.

The Temptation of Compounding Power

Replicators frighten because they promise the same thing that makes them attractive: scale without proportional effort. Build one machine that builds more machines, release one seed that restores a forest, create one program that can be copied without meaningful cost, and the original act expands beyond the time and labour of its creator.

This promise is not inherently foolish. A self-replicating or partly self-expanding industrial system could make remote construction easier. Biological reproduction sustains life. Printing multiplied access to texts, and software allows useful processes to be distributed almost instantly. These are copying systems rather than autonomous replicators in the strict sense, but they demonstrate why civilisation repeatedly seeks leverage through reproducibility.

The distinction matters. A printed book does not print another book. Software generally requires hardware, energy, distribution and people who choose to run or copy it. The effects can scale dramatically without the artefact becoming an independent organism. Calling every reproducible technology a replicator would make the category too broad to explain anything.

The dangerous transition occurs when copying, resource acquisition and decision-making become sufficiently integrated that the process can continue beyond the creator’s practical supervision. The creator wanted leverage and built a loop: each output helps produce the next output, while the cost of expansion falls upon an environment or population not represented in the original objective.

Artificial intelligence illustrates the distinction. A language model is not automatically a self-replicating system. It does not ordinarily acquire computing infrastructure, deploy copies and continue operating without organisations and users. It can nevertheless reduce the cost of producing code, persuasive text, customised messages and other outputs that are cheap to reproduce and distribute.

AI may therefore accelerate an existing multiplication rule without being the replicator itself. It can help generate variations, translate material into additional languages, adapt an operation to new targets or automate part of a production pipeline. That possibility belongs to the wider problem explored in “AI and Forbidden Knowledge: When the Book Is Already Open”: capability can spread through systems faster than governance adapts to its scale.

The temptation is strongest when those releasing the process receive the benefits while others absorb the risks. A platform gains engagement while users inhabit the resulting information environment. A factory expands production while ecosystems receive the waste. An empire gains labour from an undead workforce while the living inhabit a society that has converted death into an economic input.

Scale is not the enemy. The question is whether the expanding process remains answerable to anything beyond its own continuation. A system becomes monstrous when success is measured by how completely it reproduces, even after reproduction has ceased to serve the purpose used to justify it.

How Do You Fight a Process?

Replicator stories reveal their seriousness through the solutions they imagine. A single monster can be killed. A process has to be understood well enough to interrupt the mechanism that produces the next generation.

The relevant intervention depends upon the rule. A machine swarm may be denied material or energy. An infection may be contained, prevented or cured. A computer worm may require a patched vulnerability and the removal of compromised copies. A blight may demand restoration of the land rather than the destruction of the creatures living within it. A destructive institution may require changed incentives and authority, not merely a new leader.

Weak stories often evade this difficulty by revealing a queen, master signal or single central node whose destruction eliminates every copy at once. That solution is not automatically dishonest; some networks are genuinely centralised. It does, however, transform the replicator back into a conventional monster. The many bodies turn out to have been extensions of one vulnerable body after all.

The more unsettling stories retain distribution. Destroying one outbreak does not erase the conditions that allowed it to spread. Killing the necromancer may leave infected corpses. Removing a ruler may preserve the bureaucracy and interests that produced the same decisions. Suppressing one rumour may leave the incentives that reward its replacement.

Victory then requires more than extermination. It requires restoration, redesign or durable control: rebuilding trust, repairing an ecosystem, changing a reproduction rule, creating immunity, denying resources or making further copies less successful than their alternatives. The heroes must understand the process as a system rather than treating every visible unit as the whole enemy.

This is why replicator horror reaches beyond fear of numbers. It asks what happens when agency dissolves into continuation: when each participant can claim to be only one carrier, one employee, one machine or one frightened person responding to local pressure, while the pattern keeps expanding.

The most frightening process does not need to hate us. It uses what is available: bodies, machines, language, grief, ambition, loyalty, attention and the desire to gain scale without paying its full cost. It turns the world into the medium through which it continues.

Copying is not the danger. Life copies, culture copies and memory survives through repetition. The danger begins when a process can reproduce its own conditions while losing the ability—or the incentive—to recognise when enough has become too much.