For most of human history, survival depended on the cohesion of one’s own group and wariness toward outsiders. The distinction between us and them is not a modern pathology. It is an evolutionary inheritance. File Photo by Gian Ehrenzeller/EPA
A study published last November in Science confirmed what many observers of digital politics have long suspected: feed algorithms do not merely reflect human hostility. They can amplify it.
Researchers from the University of Washington, Stanford University and Northeastern University worked with 1,256 participants on X during the 2024 U.S. presidential campaign. They found that reducing users’ exposure to posts expressing antidemocratic attitudes and partisan animosity improved their feelings toward the opposing party by about two points on a 100-point scale. That may sound modest, but the researchers estimated that such a shift typically takes about three years to develop organically.
In other words, the algorithmic feed can compress years of political hardening into days.
That finding raises a question the study did not set out to answer: where does the hostility come from in the first place?
Algorithms did not invent tribal hatred. They amplified it. The fire was already burning.
For most of human history, survival depended on the cohesion of one’s own group and wariness toward outsiders. The distinction between us and them is not a modern pathology. It is an evolutionary inheritance.
What digital platforms have done, as the French philosopher Éric Sadin observed in L’Humanité augmentée, is take that ancient impulse and convert it into a technical system. Outrage generates engagement. Engagement generates revenue. The reptilian brain has become a source of profit in network society.
An old fire with new force
In Civilization and Its Discontents (1930), Sigmund Freud argued that civilization requires human beings to restrain their most primitive instincts in order to live together. He saw the human person as caught between two opposing drives: Eros, the force of love and construction, and Thanatos, the drive toward destruction.
When destructive energy is not channeled through work or honest debate, it can turn outward as hostility and blind hatred.
Carl Gustav Jung described a related dynamic through the archetype of the shadow, the darkness hidden within the psyche where people repress hatred, fear and selfishness. The shadow does not disappear when ignored. It waits for a trigger.
In digital environments, the trigger arrives constantly, at the speed of the feed.
As social psychologist Jonathan Haidt has observed, intuition moves before reason. When a group feels threatened, whether the danger is real or symbolic, rationality often shuts down. When disagreement enters the realm of sacred values, the opponent is no longer someone who thinks differently. He becomes a threat.
History confirms the pattern. Whenever practical debate hardens into dogma, reason loses ground. The Jacobin Terror, Stalinism and National Socialism all demonstrated that dehumanization begins before physical violence. It begins in language.
The adversary is first stripped of dignity. He is erased or condemned in speech. Only later is he treated as someone outside the circle of legitimate humanity.
Hannah Arendt located part of this danger in the failure to think. Dogmatism, whether ideological, religious or cultural, leaves no room for the reflection that coexistence requires.
Language, technology and the other
Digital platforms have not changed the sequence by which societies dehumanize the adversary. They have accelerated it. What once required pamphlets and public agitation now requires only a ranking system designed to hold attention.
Language, as later thinkers observed, does not merely describe reality. It helps generate it. The language of polarization, therefore, generates monsters.
The principle of polarity, present in philosophical and spiritual traditions since antiquity, holds that opposing forces are not separate realities called to eliminate one another, but two extremes of the same reality. Order and chaos, from this perspective, can be generative when held within a wider unity. That insight does not erase the tension. It reframes it. The question is not how to abolish destructive impulses from human nature, but how to prevent them from overwhelming public life.
What the data suggest
The Science study offers a partial and carefully bounded answer. The researchers built a browser-based tool that reranked posts in real time. It did not remove content and did not require the cooperation of X itself.
The experiment ran during a turbulent stretch of the 2024 U.S. election, a period that included President Biden’s withdrawal from the race. Within ten days, participants’ feelings toward the opposing party shifted measurably. The result held for both liberals and conservatives.
The finding matters because it clarifies part of the mechanism. If a particular ranking of content can alter political attitudes within days, then authority over that ranking is a civic matter, not simply a product decision.
Who controls these systems? Platforms? Regulators? Users? Independent researchers? Democratic societies have barely begun to answer, and in much of Latin America, where institutional trust is already fragile, the question is even more urgent.
Yet the study cannot solve the deeper problem. Algorithms exploit a vulnerability that predates them by millennia. Reranking content can reduce the symptom. It cannot dissolve the condition beneath it.
A civilizational question
The data point toward structural reform. But structure alone is not sufficient.
Radical polarization advances when societies lower their ethical guard, when people stop making the effort to recognize the adversary as a legitimate interlocutor. Digital platforms did not create that tendency, but they have given it an infrastructure of unprecedented scale and speed.
The architects of that infrastructure face a choice that is ultimately cultural. Whether they design systems that restrain the tribal impulse or monetize it will affect more than public discourse. It will influence the social conditions in which democratic life remains possible. That choice is also, in a different register, available to citizens, educators and those responsible for public institutions.
The beast that stalks the human soul is not technological. The response to it cannot be either. The first task is not to perfect the algorithm, but to recover the moral capacity to see an adversary as fully human.
Carlos Cantero is a Chilean academic at the International University of La Rioja in Spain and the author of Digital Society: Reason and Emotion. An international lecturer, adviser and consultant, he focuses on adaptability in the digital society, ethics, social innovation, and human development. The views and opinions expressed in this commentary are solely those of the author.
