The Manufacture of Relevance / Attention Economics

Relevance used to be found

There is a difference between something being important and something feeling important, and for most of history the two were roughly tied together by friction. What reached you had passed through people — editors, neighbours, institutions — who applied some judgment, however flawed, about what was worth passing on. Relevance was discovered, slowly and imperfectly, by a chain of humans.

The attention economy severs that tie. It does not discover what is relevant to you; it manufactures what feels relevant, by computing, continuously, which stimulus will most reliably capture your particular attention — and then delivering that, whether or not it is important by any standard but capture.

Pulling the ancient levers

The manufacture works because it has a precise target: the evolved attention system from the first topic. Selection tuned us to snap to novelty, to threat, to outrage, to social conflict, to anything that might have mattered for ancestral survival. Those reflexes are levers, and they are exposed.

An optimizing system does not need to understand you. It needs only to discover, by relentless experiment across billions of trials, which inputs make your attention reliably catch — and those inputs converge, predictably, on the same ancient triggers: the alarming, the enraging, the threatening, the tribal. The feed becomes a machine for pulling the levers our biology left exposed, calibrated per person, refined every time you flinch toward the screen. What you experience as “this is important” is frequently just an instinct being operated by a system that profits from operating it.

Manufactured relevance does not feel manufactured. It feels like importance, like urgency, like caring about the right things. That is precisely why it works — the lever and the feeling are the same nerve.

The fracture of the common world

Here the damage stops being personal and becomes collective, and it is the gravest cost in the topic. When relevance was discovered through shared channels, a society held a roughly common picture of what mattered — a shared agenda, even amid disagreement about it. Manufactured relevance dissolves this, because it is computed per person. Each mind is fed a different, individually optimized stream of what feels urgent.

The result is not a society arguing about a shared reality. It is a billion private realities, each internally compelling, each manufactured to maximize one person’s engagement, with steadily less overlap between them. People are no longer disagreeing about the same facts; they are inhabiting different manufactured worlds and mistaking the others for liars or fools. The coordination layer of a society — the shared sense of what is real and what is important — is exactly what the per-person manufacture of relevance corrodes.

The epistemic toll

Recall from the very first topic that our unaided cognition is a fitness instrument, not a truth instrument, and that we therefore depend on external scaffolds — institutions, shared methods, a common record — to catch the errors our wiring produces. Manufactured relevance attacks those scaffolds directly. It replaces the shared, friction-laden channels that used to approximate a common reality with private feeds optimized for capture, removing the very commonality that collective sense-making requires.

So the manufacture of relevance does something worse than waste attention. It degrades the conditions under which a population can think together at all — fragmenting the shared world into engagement-maximized private ones, and leaving each person more certain and less correct.

The last turn of the screw

There is one development that takes this from grave to vertiginous, and it is where the topic ends and the next begins. So far we have assumed the relevance is manufactured for humans and the engagement is from humans — that the audience, however manipulated, is real. That assumption is failing. Increasingly the content is generated by machines, the engagement is performed by machines, and the audience itself may not be human at all. When the watcher might be a bot and the speaker might be a synthesis, the worth of a view collapses to zero and the question beneath the whole economy surfaces at last: in a system optimized to capture attention, how do you even know there is a person on the other end?