Isaac Asimov’s psychohistory remains one of science fiction’s cleanest fantasies of social prediction. Gather enough information, use the right mathematics, and the behavior of huge populations becomes forecastable even when individuals remain unpredictable.
That idea sits at the center of *Foundation*. Hari Seldon predicts the fall of the Galactic Empire and designs a project intended to shorten the dark age that follows.
The technology is fictional. The desire is familiar.
Governments forecast inflation, migration, disease, crime, and elections. Companies forecast demand and churn. Platforms forecast what people will click, buy, watch, or believe. Campaigns model voters. Insurers model risk.
That is why books like Foundation feel newly relevant in the algorithm age, and why new books like Foundation increasingly focus on deployed models rather than abstract prophecy. The question has shifted from “Can society be predicted?” to “What happens when institutions act as though it can?”
MAYA: prediction as government
The MAYA Narrative Universe pushes that logic into biology. Its central network is not a server cloud. The Maya is a living network grown through trees, connecting seven sentient species across Neh and embedding information inside ecology.
That network does what information infrastructure does in our world: communication, entertainment, memory, connection, and data. The crucial difference is scale and integration. The rulers who control it can use that information to predict events and intervene before undesirable outcomes happen.
It’s a science fiction about artificial intelligence into a political problem without requiring the usual humanoid AI. The predictive system is embedded in the world.
The creators have described MAYA in explicitly systemic terms. Its political systems grow from access to information and prediction rather than standing apart from the technology. A Variety report discusses the project as a universe built around systems-level conflicts across books, games, and other media. The wider project also includes the tabletop game Trials of MAYA, but the novel is where the predictive political system is explored as lived society.
The interesting question is what happens after prediction becomes ordinary governance. If a system can see harmful outcomes early enough, intervention can look like care. Prevent a murder. Redirect a riot. Change an incentive. Put the right story in front of the right person.
At that point, prediction does not need police power to influence behavior. It can shape the environment in which choices are made.
MAYA reads like a science fiction epic built for the biggest possible screen.
Foundation: prediction as historical engineering
Psychohistory is not interesting because Seldon can make a graph. It is interesting because knowledge of the forecast changes political action.
The Foundation is created because a prediction exists. Institutions are built around an expected future. People later interpret crises through the framework Seldon left behind.
Prediction becomes governance.
That move has shaped a lot of later political science fiction. The model does not need to be perfect. It only needs to become powerful enough that people organize decisions around it.
Infomocracy: prediction inside data infrastructure
Malka Older’s *Infomocracy* brings the problem closer to the present. Its world combines a global microdemocracy with an information infrastructure that has enormous influence over what people know about elections and political claims.
The novel understands that prediction is never isolated from information quality. Before an institution can model people, it needs data. Before people can respond rationally, they need some way to decide which information is trustworthy.
That makes the information layer political even before anyone writes an algorithm.
The Three-Body Problem: forecasting under civilizational threat
*The Three-Body Problem* stretches strategic thinking across generations. Humanity learns that an alien civilization may eventually arrive, and knowledge of that future reorganizes present-day loyalties.
The forecast is not statistical in Asimov’s sense, but it creates a similar political problem. People act now because of something expected later.
A prediction about civilization changes civilization before the predicted event occurs.
The Quantum Thief: prediction when identity is data
Hannu Rajaniemi’s *The Quantum Thief* imagines a far future where memory, identity, encryption, and information are deeply entangled.
The details are much stranger than ordinary recommender systems, but the underlying question is recognizable: what happens when personal information becomes infrastructure?
Prediction gets easier when identity is legible. Resistance gets harder when systems can observe more of the self.
Why prediction creates power before it creates certainty
Modern institutions rarely need perfect predictions to change behavior.
A credit score can be wrong and still affect a loan. A risk model can be incomplete and still change how a neighborhood is policed. A recommendation system can misunderstand a user and still decide what that user sees. A hiring model can be noisy and still filter candidates.
The power comes from deployment.
That is the lesson science fiction can make obvious. A model becomes political when somebody uses it to allocate attention, money, freedom, protection, or opportunity.
Recommendation is a quiet form of forecasting
The word “prediction” can make people imagine a dramatic forecast about war, elections, or catastrophe. A lot of modern prediction is quieter. A system estimates which song you will play, which product you will buy, which route you will take, or which post will keep you on a screen.
Those forecasts matter because they are tied to ranking. The predicted option is shown first. The less likely option becomes harder to encounter. A model can therefore influence the behavior it is measuring simply by controlling the menu.
That makes today’s algorithmic power feel closer to the MAYA premise than a literal crystal ball. The system does not have to force an outcome. It can change what is easy, visible, timely, or socially rewarded.
Accuracy is only half the governance problem
Public arguments about algorithms often focus on whether the model is correct. Accuracy matters, but governance adds a separate set of questions. Who can challenge the output? Who sees the inputs? Can somebody opt out? Does the institution keep using the model after errors become visible?
A prediction system can be statistically impressive and politically unacceptable. It can also be mediocre and politically powerful if institutions trust it anyway.
That is why the best algorithmic science fiction spends time on authority rather than treating prediction as a neutral superpower.
Prediction can create its own evidence
Forecasts also produce feedback loops.
If a system predicts that someone is likely to leave a job, the employer may stop investing in that person. If a platform predicts that a user prefers a certain kind of content, it may show more of that content until the preference becomes stronger. If a state predicts unrest, the measures it takes can either prevent unrest or help cause it.
This makes the old dream of objective prediction much messier.
The model is inside the system it is modeling.
Free will becomes a design question
Classic debates about free will often sound metaphysical. In algorithmic societies, they can become practical.
How many options do you need before a choice counts as meaningful? Does a person choose freely if every option was ranked, priced, framed, and timed by a system that knows their behavioral history? Can you consent to influence you cannot see?
Those questions explain why the algorithm age keeps producing new variations on Asimov’s old idea. They also explain why prediction stories increasingly care about interfaces, incentives, and institutions rather than only the mathematical model itself.
Prediction changes responsibility
A prediction also changes who can plausibly claim ignorance. If an institution can foresee a harm with reasonable confidence, does it acquire a duty to act? If acting prevents the harm by constraining somebody’s choices, how much intervention is justified?
That is where forecasting turns into ethics. The better a system becomes at anticipating outcomes, the harder it is to separate knowledge from responsibility. Science fiction can push that tension to an extreme and make the tradeoff visible.
Why Foundation still matters
People continue reading Isaac Asimov books because psychohistory captures something institutions still want: a society that can be understood well enough to manage.
The newer fiction is often less confident about what happens next.
Asimov asked whether mathematics could reduce historical chaos. Contemporary writers are increasingly asking who gets to own the model, who gets modeled, and whether prediction changes people before they can surprise it.
That is the real bridge from Foundation to the algorithm age.
It also explains why the topic keeps widening beyond conventional AI stories. Prediction touches finance, health, education, policing, media, work, and relationships because all of those domains produce data and decisions. The science-fiction version simply makes the hidden model visible enough that characters can fight about it.
The future may never become perfectly predictable. Power can still grow from the attempt, especially when everybody else has to live with the model’s decisions and has no equal model of their own to challenge the result or expose its blind spots.