Your Attention, Priced

FOR nearly a century, mainstream economics operated on a simple, foundational premise: human beings are rational actors attempting to maximise utility within a system of scarce resources. Money was the primary medium of exchange, and price was the ultimate mechanism for balancing supply and demand.

Today, that model is fundamentally breaking down – not because human nature changed, but because the primary asset being traded is no longer physical capital or currency. It is attention.

The idea is not new. As early as 1971, the economist and cognitive scientist Herbert Simon observed that in a world rich in information, information consumes something scarce: the attention of its recipients. What has changed is that an industry now exists to harvest that scarcity at planetary scale.

We have fully entered the algorithmic economy, a financial ecosystem built on the monetisation of human focus, cognitive friction, and behavioural prediction. While consumers experience these systems as free digital services – social feeds, search engines, recommendation engines – the underlying macroeconomic reality is far more complex. We are participating in an asymmetric market where user attention is extracted as a raw material, refined through predictive machine learning models, and sold to the highest bidder in real-time ad auctions.

The Economics of Cognitive Friction

In classical market theory, efficiency means minimising transaction costs. A frictionless market allows buyers and sellers to meet, exchange value, and leave satisfied. The algorithmic economy operates on the exact opposite incentive structure.

For modern digital platforms, friction generates revenue. Every extra second a user spends navigating an interface, scrolling past a target output, or reacting to high-arousal content represents an increase in inventory – specifically, ad impressions and data points. This creates a systemic divergence between user utility and platform profit: the classical model treats platform value as user efficiency multiplied by task completion rate, while the algorithmic model treats platform value as user engagement time multiplied by ad load efficiency.

When platform survival depends on maximising time spent within an ecosystem, algorithms are optimised not for truth, utility or well-being, but for cognitive retention. The most efficient retention mechanisms are psychological triggers: outrage, novelty, tribal identification, and variable reward schedules.

Novelty in particular is not a neutral quality. In the largest study of its kind, Soroush Vosoughi, Deb Roy and Sinan Aral tracked every verified true and false news story distributed on Twitter between 2006 and 2017 and found that falsehood travelled significantly farther, faster and deeper than the truth – not because bots pushed it, but because human beings found it more novel and shared it. A system tuned for engagement is therefore tuned, structurally, for whatever is most surprising rather than whatever is most accurate.

Structural Spillover: From Feeds to Markets

The macro effects of this incentive structure extend far beyond smartphone screen time. As attention extraction becomes the primary revenue driver for the world’s largest technology firms, it induces structural distortions across the broader economy.

Knowledge work relies heavily on deep, uninterrupted focus. As attention becomes fragmented by hyper-optimised notification loops and algorithmic media, individual cognitive productivity faces a measurable tax. What economics traditionally classified as a personal productivity issue is increasingly an environmental, systemic externality.

The tax is measurable in a fairly literal sense. Gloria Mark and her co-authors, observing office workers under experimental interruption, found that people finish interrupted work no more slowly than uninterrupted work – they simply compensate by working faster, and pay for it in stress, frustration, time pressure and effort. The cost does not appear in output figures. It appears in the worker.

In political economics, stable institutions require low political volatility to encourage long-term capital investment. Algorithmic distribution, however, disproportionately scales extreme or inflammatory content because it generates the highest engagement metrics. Political consensus degrades, polarisation increases, and institutional trust drops – introducing real economic risk factors into national markets.

The effect is measurable rather than merely anecdotal. When Smitha Milli and colleagues compared Twitter’s engagement-based ranking with a simple reverse-chronological feed, the algorithmic feed served users noticeably more content expressing hostility towards political opponents – and, asked afterwards whether they wanted to see more of it, users rated that content lower than what the chronological feed had shown them. What maximises engagement and what people actually value had come apart.

With continuous streams of behavioural data being collected to fuel recommendation models, firms increasingly possess the data required for perfect price discrimination. Rather than setting a uniform market price, algorithms can predict an individual’s exact willingness to pay based on their history, urgency and emotional state – effectively capturing consumer surplus and transferring it directly to platform margins.

This is no longer hypothetical. The United States Federal Trade Commission’s study of “surveillance pricing”, published in January 2025, found intermediary firms using location, browsing history, mouse movements and abandoned shopping carts to help retailers set individualised prices; a shopper profiled as a new parent, in one of the study’s examples, could be shown the more expensive baby thermometer first.

Reclaiming the Commodity

If attention is the fundamental currency of the modern global economy, we must begin treating it with the same regulatory and analytical rigour applied to labour, capital and natural resources.

Currently, our economic accounting frameworks, such as GDP, treat digital platforms as zero-cost consumer surpluses – after all, using a search engine or social app costs nothing out of pocket. But this ignores the massive opportunity cost of depleted human focus, and the social externalities of algorithmic polarisation.

Economists have begun to build the alternative: Erik Brynjolfsson and colleagues have proposed GDP-B, a measure that prices free digital goods by what users would have to be paid to give them up. The results cut both ways. When Hunt Allcott, Luca Braghieri, Sarah Eichmeyer and Matthew Gentzkow paid a randomised group to deactivate Facebook for the four weeks before the 2018 US midterms, subjective well-being rose, polarisation fell, factual news knowledge fell too – and, tellingly, participants valued the platform less afterwards than they had before, which suggests conventional measures overstate the surplus these services provide.

Addressing this requires a shift in how we structure digital governance. It means taxing data extraction, treating massive data harvesting not as a passive side effect of service provision but as a resource extraction model subject to regulatory oversight – a case strengthened by the finding of Daron Acemoglu and his co-authors that data markets systematically underprice data, because what one user reveals also reveals the people who resemble them, so individuals sell too cheaply and too much is collected.

It means algorithmic auditing, requiring transparency around the engagement-maximising algorithms that drive public discourse and market behaviour; the European Union’s Digital Services Act has already made independent audits mandatory for the largest platforms. And it means designing for friction, valuing platforms that prioritise user autonomy over retention metrics.

The central challenge of twenty-first-century economics will not merely be managing the distribution of goods or the stability of fiat currencies. It will be determining who owns, controls and profits from the finite landscape of human consciousness.