Marketing is a perfect stress test for how signal should flow, precisely because it is neither purely signal nor purely noise. At its best it solves a real informational problem: something useful exists, and a participant who would benefit does not yet know it does — that is legitimate signal, and an ecology that suppressed it would become informationally closed, leaving people to encounter only what they already know to search for. At its worst it is engineered interference: capture attention, create or intensify desire, manipulate salience, and steer behavior toward an outcome that mainly benefits the sender. That is not signal flow at all; it is an attempt to alter the receiver. The same activity contains both, which is exactly why it tests the principle.
Discovery versus attention capture
The distinction the model needs is between discovery and attention capture. A useful ad says here is something relevant you may want to know about. A manipulative one says I will keep entering your perceptual field until your preferences bend. Those are different relationships, and the test is not whether persuasion occurs — persuasion is unavoidable in human life; a teacher persuades, a friend recommends a restaurant, a candidate argues, a company explains why its product is better. The real issue is jurisdiction over attention:
The right to signal does not include an unlimited right to occupy another participant’s attention.
The advertiser has standing to make a claim. The participant has standing to control how much access that claim gets to their attention. Both are real; neither erases the other.
Responsive signal, not attention extraction
So marketing survives in a mature ecology, but its default architecture would look very different from the feed-driven model. Instead of companies buying repeated involuntary access to attention, discovery becomes more intentional and relevance-based. Say I’m looking for a month-long place to stay, and commercial providers now have legitimate standing to present themselves; say I need a tool that does X, and product information becomes relevant signal. That is a world apart from interrupting someone reading about their grandmother’s illness with a cruise advertisement because an algorithm inferred they are a likely buyer. The first is responsive signal; the second is attention extraction — and the underlying line is between informing a preference and manufacturing one.
The boundary is genuinely not clean — you can truly discover something you never knew you wanted, and a new instrument, destination, book, technology, or food can enlarge your world. The model should preserve that unsolicited novelty, but under different constraints than today’s: low frequency, transparent sponsorship, easy dismissal, no disguised advertising, no punitive repetition, no dark-pattern coercion, stronger restrictions when targeting children or vulnerable participants, and a clear separation between informational relevance and paid ranking.
The ratio that changes the economics
The ecological question is not did this ad make someone buy? It is did this signal increase useful awareness without disproportionately consuming or manipulating the receiver’s attention? — and that can actually be measured. Weigh discovery value (how often people say I genuinely did not know this existed and it was useful) against attention cost (the time, interruption, repetition, cognitive load, and unwanted exposure required to produce it). Then a ratio like signal value per unit of attention consumed becomes a meaningful ecological measure: a genuinely useful notice seen once scores very well; ten thousand impressions needed to induce a single purchase looks terrible. That single inversion would radically change advertising economics, because it prices in the attention it currently treats as free — and attention, as renewability established, is a fragile resource whose extraction has ecological cost.
AI makes it much better — or much worse
AI can push marketing in both directions at once. Better: instead of a company blasting ten million people, an AI intermediary understands you are actively looking and surfaces three genuinely relevant options — a huge reduction in noise. Worse: AI models your psychology so well that persuasion becomes nearly frictionless, at which point marketing stops being communication and starts approaching behavioral control. That is where the model needs hard boundaries, and the governing rule scales with capability:
The stronger a system’s ability to model and influence a participant, the stronger that participant’s claim to protection.
Which yields a sharp line: personalization may improve relevance; it may not silently exploit vulnerabilities in order to override reflective choice. Personalizing here are walking shoes in your size is fine. Our model knows you buy impulsively when lonely at 11:30 p.m., so that is when we push luxury purchases is not personalization — it is ecological predation, and the model should name it as such.
The business model is the root incentive
Much of today’s information economy funds itself by selling attention, which creates an almost inevitable incentive to maximize engagement. If attention is treated as a renewable but vulnerable resource, an economy built on extracting the maximum of it is structurally misaligned — not because commercial communication is illegitimate, but because its funding model rewards depletion. That does not mean marketing disappears; it means the field shifts toward subscription, intentional search, marketplaces, recommendation intermediaries controlled by the participant, limited sponsored discovery, public or cooperative discovery systems, and direct opt-in relationships with brands. The important inversion is the question being asked. Today advertisers ask how can we get access to the person? The model asks instead:
Under what conditions does this signal legitimately deserve access to this person’s attention?
That connects straight back to standing: a company has standing to describe its product truthfully; it does not acquire standing over your attention merely because you are statistically likely to buy. Your attention remains yours — and yet you still benefit from an ecology in which information about genuinely useful new possibilities can reach you. The balance is to protect discovery without normalizing intrusion.
The test, and the clean line
So marketing does not disappear under the model; it becomes accountable to the same signal principles as everything else. Is it relevant? Is its source visible? Is the persuasion transparent? Can the receiver refuse it easily? Is its informational value commensurate with the attention it consumes? Does it increase the participant’s option space, or narrow it through manipulation? Which reduces to one clean distinction:
Good marketing expands informed choice. Bad marketing captures attention in order to manufacture behavior.
The model wants the first, and treats the second as a form of ecological depletion — attention consumed and reflective choice eroded, dressed as service.