We like to believe that evidence changes minds. When new information arrives, we weigh it, adjust our views, and move closer to reality. This is the story we tell ourselves about how reasonable people behave.
It is increasingly untrue.
A growing number of us now treat facts the way earlier generations treated flags: as things that belong to one side or another. A fact is no longer primarily something that happened. It is something our side has decided to hold. The other side’s facts are treated as propaganda by definition. Our side’s claims are treated as established by definition. Once this shift occurs, evidence stops functioning as evidence. It becomes ammunition or an obstacle, depending on whose side it appears to help.
This is not ordinary partisanship or the familiar human tendency to favor our own. It is something more systematic. I call it relocated deference: the transfer of one’s own moral and evidentiary judgment away from the self and onto a group whose authority is felt but cannot be located or held accountable. The judgment still feels like one’s own. The deference does not feel like deference. It feels like clarity, loyalty, and moral seriousness.
The mechanism is not new. Psychologists have documented versions of it for decades. Stanley Milgram showed that ordinary people would administer what they believed were dangerous electric shocks when a credible authority told them the experiment required it. Later researchers demonstrated that the most obedient participants were not the most passive. They were the ones who had come to identify with the project itself. They were not being forced. They were volunteering.
What has changed is the direction and the scale. In the laboratory, responsibility was handed upward to a person in a white coat who could, in principle, be questioned or blamed. In the digital age, responsibility is handed sideways and downward — to the tribe, the feed, the consensus of people like us. That consensus has no office, no spokesperson with standing to bind it, and no address at which it can be held responsible. It is a fog with the authority of a sovereign and the accountability of weather.
This relocation becomes especially powerful when it is wired to identity — above all the kinds of identity that are visible, hard to exit, and already loaded with grievance and solidarity. Once such a frame is placed over an event, it tends to determine the response before the evidence arrives. Contradictory facts are not weighed and rejected. They are reinterpreted as further proof of the narrative. A verdict that goes against the group proves the system is rigged. A verdict that goes for the group proves the system occasionally fails under pressure. Every outcome feeds the loop. The belief grows stronger in proportion to the evidence against it.
Digital platforms have turned this ancient tendency into an industrial process. Recommender systems optimized for engagement do not primarily show us what we would choose on reflection. They show us what reliably captures attention. Content that triggers moral outrage toward an out-group performs exceptionally well. The machine learns this and serves more of it. At the same time, it rewards users for expressing that outrage, training them through likes, shares, and approving replies to escalate. And because it amplifies the most extreme voices, it creates a distorted picture of what the other side actually believes. We come to see our opponents as more hostile and more numerous in their hostility than they are — and we adjust our own behavior accordingly.
The result is visible in real time whenever a contested verdict lands. When a court decides against the figure one tribe has adopted as its own, that tribe does not engage the evidence; it attacks the legitimacy of the process — the makeup of the jury, the motives of the prosecution — and rallies harder around its champion. When a court decides for that same figure, the opposing tribe declares the system rigged and the outcome proof of rot. Each side treats the verdict less as a finding about what happened than as confirmation of a story it held in advance. The evidence is secondary. Both responses are the same move in opposite jerseys.
This pattern is not limited to high-profile trials. It appears whenever a claim, however weakly supported, spreads because it flatters a group’s self-image while contradictory information is ignored or attacked. The common feature is the relocation of judgment. People are not primarily asking what is true. They are asking which side they are on, and allowing the answer to settle the question of what happened.
The costs are concrete. Individuals are turned into symbols and stripped of their particularity. Bystanders absorb harm from crowds that have diffused responsibility into no one in particular. Institutions whose purpose is to settle contested questions — courts, expert processes, parts of the press — lose authority once they are successfully tribalized. Most seriously, the shared world itself frays. When facts become tribal property, disagreement ceases to be a disagreement about the same reality. It becomes a collision between two sealed realities that cannot be reconciled because they were never built from the same materials.
There is no simple fix. Greater intelligence does not solve the problem; on identity-charged questions, analytical ability often makes people better at constructing justifications for whatever their group already believes. Simply becoming aware of the mechanism does not confer immunity either. The surrender of judgment does not feel like surrender. It feels like conviction.
What remains possible is a narrower but real form of resistance. It begins with learning to recognize the moment when certainty arrives too quickly and too comfortably on a charged question. It continues with the deliberate introduction of friction: refusing to form a firm position until one has actually examined the evidence, holding oneself answerable for the accuracy of claims one repeats, seeking out the strongest arguments from people who disagree, and being willing to affirm a fact that damages one’s own side. None of this is easy. All of it is uncomfortable. It offers no guarantee of being right and no protection from social cost.
But it is the only practice that keeps the authorship of one’s own judgments from being quietly reassigned to the group. In an environment engineered to make that reassignment feel natural and virtuous, that practice may be the most radical thing an individual can still do.