Every ad ever served was subsidized by an unpaid worker: the person watching it. The entire attention economy is built on acquiring a scarce human resource at a price of zero and selling it at a price of not zero. That spread funded the modern internet, and for three decades nobody with power had an incentive to question it. The consumer was the product, the product had no invoice, and everyone upstream got rich on the margin.
I think that arrangement is ending, and not because of ethics. It is ending because of tooling. For the first time, ordinary people are getting agents, wallets, and interfaces that can meter, price, and gate their attention automatically. When the cost of charging for something drops below the value of the thing, markets form. Attention is the last major asset class where that has not happened yet. This paper is about what advertising looks like when it does.
To be clear about my position: I do not believe most people will sit at a dashboard selling their browsing history for pocket change. That vision failed repeatedly and deserved to. What I believe is subtler and bigger. Payment for attention will be embedded, automated, and mostly invisible, negotiated by software on the consumer's behalf, and it will reprice the entire funnel.
Key Findings
- Direct consumer compensation for attention has failed historically because the transaction costs exceeded the payout. Agent mediated negotiation removes that constraint, which changes the outcome, not just the efficiency.
- I introduce the Attention Wage: the effective hourly rate a person earns from the commercial use of their attention and data. Today it is near zero for consumers and enormous for platforms. That spread is the single largest arbitrage left in media.
- There is a Dividend Threshold, the payout level at which a meaningful share of consumers will actively gate access to themselves. Below it, nothing changes. Above it, reach becomes a purchased permission, not a harvested default.
- Paid attention destroys low quality advertising first. When impressions carry a direct consumer price, spray and pray budgets collapse fastest.
- Brands with genuine relationships get a discount in this market. Strangers pay rack rate.
- Compute your brand's exposure now. Estimate what share of your current reach depends on unpaid, unfiltered access to strangers. That share is your repricing risk, and for most brands I audit, imagine it at well over half.
- Shift budget toward earned admission: relationships, subscriptions, communities, and owned channels where people have already said yes. In a paid attention market, a real opt in is a standing discount.
- Start treating consumer value exchange as a line item, not a gimmick. Whatever you would grudgingly pay an agent gatekeeper tomorrow, invest in direct value today and buy loyalty instead of tolls.
- Rehearse the negotiation. Decide today what a verified minute of your ideal customer's attention is worth to you in dollars. If you cannot answer, you are not ready for a market that will ask.
- Watch the messaging layer for the first real Dividend Threshold crossing. It will look small and gimmicky. It will not stay small.
The Barter Nobody Agreed To
The classic defense of ad funded media is that it was always a trade: free content in exchange for attention. Fine, but examine the trade. One side set the terms, measured the goods, priced the inventory, and kept the books. The other side clicked "accept" on a document written to be unread. That is not a market. That is a barter arranged by one party on behalf of a counterparty who never saw the exchange rate.
What makes 2026 different is that the counterparty is getting representation. On device assistants now read, filter, summarize, and block on our behalf. Inbox agents triage what reaches us. Browsers negotiate consent programmatically. Each of these is, structurally, an agent standing between commercial messages and human attention, and every agent standing in that position will eventually ask the obvious question: what is my principal being paid for this?
I have written about what happens when machines become the audience you market to, and the data dividend is the economic half of that shift. Once an agent can perfectly filter, attention stops leaking for free. Anything that stops leaking gets priced.
The Attention Wage
Run the arithmetic that platforms would prefer nobody run. Take global digital ad spend, divide it by the human hours of attention that generated it, and you get a crude global hourly rate for watching ads. Back of the napkin, the number lands in the range of loose change per hour. Now look at where that value actually settles: overwhelmingly with platforms and intermediaries, marginally with publishers, and with the attention worker, almost nowhere.
That gap is the Attention Wage spread, and I use the word wage deliberately. Attention is time plus cognition, the same inputs as labor. We have simply never had payroll infrastructure for it. The historical objection was practical: you cannot cut a two cent check. Correct, a human cannot. An agent can. Micropayment rails, programmable wallets, and automated negotiation make sub cent settlement boring plumbing rather than science fiction.
I am not predicting that consumers get rich. The Attention Wage will stay modest per person. The prediction is directional: the wage goes from zero to nonzero, and in economics the move from zero to nonzero is never small. Free inputs get wasted. Priced inputs get allocated. Advertising has been wasting human attention for decades precisely because it was free to waste.
The Dividend Threshold
Here is the mechanism that decides the timeline. Every consumer has a price below which managing the sale of their attention is not worth the bother. Call the point where the payout clears that bother the Dividend Threshold. Below the threshold, data dividend schemes stay novelty apps with terrible retention, which is an accurate description of the last fifteen years of attempts.
Agents move the threshold, because they remove the bother. When gating your attention requires zero effort, even a trivial dividend clears the bar. Suppose an agent quietly earns its owner a few dollars a month by filtering commercial contact and admitting only paid, relevant offers. Nobody organizes their life around that money. But nobody turns it off, either. And a filter nobody turns off is, from an advertiser's perspective, a paywall around the human race.
The threshold will not be crossed evenly. Expect it first in the inbox and messaging layer, where agent adoption is furthest along and where the spam analogy makes payment feel like justice. Then commerce intent data, which is the most valuable and easiest to price. Broad media attention comes last, dragged there by competition. My timeline: meaningful paid access programs in messaging and commerce by 2028, and by 2030 a mainstream consumer expectation that cold commercial contact from a stranger brand carries compensation, the way we now expect a cookie banner whether we like it or not.
What Gets Repriced
When attention carries a direct price, the worst advertising dies first. The economics of high volume, low relevance targeting only work when the marginal impression costs fractions of a cent and the audience cannot bill you. Add a consumer side price and the spreadsheet collapses. This is the same extinction event that already hit organic distribution, where I watched brands lose free reach they had mistaken for a birthright. I documented that collapse in the death of organic reach, and the data dividend is its sequel: first the platforms charged you for reach, now the people will.
The corollary is the interesting part. Relationship becomes a discount. A consumer's agent will admit a brand its principal knows, likes, and buys from at a lower price, or free, because the principal genuinely wants the contact. Strangers pay rack rate. That inverts decades of practice where acquisition budgets dwarfed retention budgets. In a paid attention market, being unknown is the most expensive thing a brand can be.
It also means volunteered data becomes the premium substrate, because data given willingly comes with permission attached rather than a bill. The economics of that exchange are different enough that I give them their own treatment in Paper No.36.
The Objections Worth Taking Seriously
Two objections deserve honest answers. First: platforms will simply ban paid gating and keep the spread for themselves. Some will try. But agents sit closer to the user than platforms do, at the OS, browser, and device layer, and history says the layer closest to the human eventually wins the fight over the human's defaults. Second: a paid attention market prices poor consumers' attention cheap and rich consumers' attention dear, which is regressive. True, and worth regulating. But notice the current system prices everyone's attention at zero and keeps the money. A low wage beats no wage, and transparency beats extraction.
What I Would Do About It
The attention economy was never free. The costs were just booked to people who had no way to invoice us. They are getting one. Price accordingly.