In 1971, Herbert Simon sat before a symposium in Washington D.C. and observed — with the clinical calm of a man describing something already lost — that "a wealth of information creates a poverty of attention." He was not making a cultural complaint. He was describing a thermodynamic fact. Information has a production cost. Attention does not. And because attention is finite while information is infinite, the logic of any market built on information would eventually treat attention as the one thing that could not be manufactured, only extracted.
Fifty years later, an entire global industry — worth some $740 billion annually, per eMarketer — runs on exactly this premise: that human attention is the scarce resource, and whoever auctions it most efficiently wins. The auction is working. The resource is not.
The mechanism behind this failure is what John Conlon, a researcher at Carnegie Mellon University's Department of Social and Decision Sciences, set out to map in 2025. In his paper "Persuading without Changing Beliefs," Conlon and collaborators ran a series of experiments that isolated attention as the vector of persuasion. Their finding: information doesn't persuade by changing what people believe. It persuades by redirecting what they pay attention to. When information directs attention toward a product attribute, responsiveness to that attribute increases by approximately 69 percent — and this effect holds even among participants who already knew the information. The mechanism isn't belief change. It's spotlight displacement.
"A wealth of information creates a poverty of attention."
— Herbert Simon, 1971, Designing Organizations for an Information-Rich World
This is the anatomy of a commons. Conlon's findings suggest that because attention is capacity-constrained, directing it toward one attribute necessarily means neglecting others. The total pool of attention doesn't grow when you excite it. It redistributes. Which means the advertising industry's century-long project of capturing and redirecting attention is, at the neurological level, a zero-sum game. There is a fixed amount. Someone always loses when someone else wins.
The longitudinal data on actual human attention is not encouraging. Gloria Mark at UC Irvine has spent twenty years documenting what happens to attention in digitally saturated environments. Her studies — using behavioral logs rather than self-reports — tracked screen-based attention across the 2000s and 2010s with precise instrumentation. The numbers form a compression sequence: average focused attention on screens was 150 seconds in 2004. By 2012 it was 75 seconds. By 2020 it was 47 seconds. These figures have been replicated by independent researchers using computer logging techniques from 2016 through 2020. This is not noise. It is a trend.
What makes the data structurally significant is what happens after distraction. When people are pulled away from a task — and in a world of infinite scroll and push notification, people are always being pulled away — Mark's research found it takes an average of 23 minutes and 26 seconds to return to the original cognitive state. The commons is not merely being harvested. It is being depleted in a way that actively degrades its own capacity for renewal. Each interruption doesn't cost you the time of the interruption. It costs you the recovery, too.
A meta-analysis published in Psychological Bulletin — covering 98,299 participants across 71 studies — found that short-form video use (TikTok, Instagram Reels, YouTube Shorts) was associated with significantly poorer cognition, with the strongest effects on attention (r = −.38) and inhibitory control (r = −.41). The Karolinska Institutet followed 8,324 children aged nine and ten for four years and found that greater social media consumption was associated with gradual development of inattention symptoms independent of socioeconomic background or genetic ADHD predisposition. The commons is being extracted before many of its occupants have learned to use it.
The platforms have known this in the aggregate for longer than they will say. What is notable about the current moment is what they are now saying, quietly, through the language of measurement reform.
In March 2026, Meta announced a restructuring of its attribution model — splitting what had been a bundled click-through category (which included likes, shares, saves, comments, profile taps, and actual link clicks within a 7-day window) into three separate buckets: click-through, engage-through, and view-through, each with tightened windows. The company explicitly acknowledged that its old click-through definition had produced numbers that diverged from Google Analytics by 30 percent or more. Simon Whitcombe, VP of Meta's Global Business Group, framed the problem plainly on LinkedIn: "Why don't my Meta numbers match Google Analytics?" — a question that had haunted performance marketers for a decade, now addressed not as a calibration issue but as a structural admission.
The practical impact was immediate. Remarketing campaigns — those most dependent on accurate multi-day attribution — saw reported conversions drop 25 to 50 percent, because the engage-through window collapsed from 7 days to 1 day. Whole categories of social interaction that Meta had previously counted as contributing to conversion simply vanished from attribution. Meta advised advertisers: do not compare pre-March and post-March benchmarks. Wait two to three weeks. The company was asking the industry to accept that its primary measure of value had been wrong, systemically, for years.
The Reuters investigation from November 2025 adds a darker dimension to what "wrong" means. Internal Meta documents showed the company's own safety team estimated its platforms were involved in approximately one-third of all successful scams in the United States. Meta projected roughly $16 billion in 2024 revenue from scam and fraudulent ads — about 10 percent of total revenue — with users shown approximately 15 billion scam ads daily. The measurement problem is not only that attribution windows were miscalibrated. It is that the metrics being optimized for were measuring something other than value delivered.
The auction responds to scarcity signals. When a resource degrades, those who rely on it feel it in price before they feel it in narrative.
TikTok CPMs — the cost to reach a thousand users — fell approximately 80 percent year-over-year between January 2024 and January 2025, per agency-verified data from Jellyfish, Wpromote, Tinuiti, and AdRoll. When an attentional environment becomes noisy rather than signal-rich, when users scroll past rather than register, the auction prices the degraded inventory accordingly. Meta's own impression volume increased 18 percent year-over-year in 2025, which means advertisers were buying more placements — but the economics were sustained by volume, not by value per impression. More units at lower margins is the signature of a commodity market approaching saturation.
Instagram's engagement rate fell 26 percent between 2024 and 2025, declining from 7.3 percent to 5.4 percent across Buffer's analysis of more than 52 million posts. TikTok saw average views per post fall 17 percent and interactions fall 32 percent in the same period, per Metricool's analysis of 39.7 million posts. Global Web Index data shows daily time spent on social media peaked in 2022 and has declined approximately 10 percent since — while content supply continued to increase. The denominator is shrinking. The numerator is not keeping pace. More product being fed to fewer people who want less of it.
Ogilvy's 2026 Social Trends Report, authored by Catherine Sackville-Scott and Awie Erasmus and released in January 2026, reframed what the industry was experiencing through a useful neologism: the "inattention economy." Their data found that 20 percent of consumers had deleted a social application in the past year, and that searches for the term "brainrot" had increased 900 percent year-over-year. These are not engagement metrics. These are exit data. People are not passively disengaging from content. They are actively leaving the environments that deliver it.
The report's thesis is that the brands winning in this environment are not the loudest. They are the most needed — sought out rather than scrolled past. This is a fundamentally different theory of value for an industry built on interruption. The Ogilvy framework asks whether a brand earns its presence rather than whether it captures it. That's a quiet revolution in how the attention economy talks about itself.
"A wealth of information creates a poverty of attention and a poverty of attention creates a wealth of information."
— Josef Falkinger, The Economic Journal, 2008
Matthew Crawford, in his 2015 work on attentional commons, argued that attention should be treated like air or water — a shared resource whose degradation is everyone's loss. Dirk Wagner formalized this in academic terms as "The Tragedy of the Attentional Commons," demonstrating that rational self-interest at the individual brand level produces collective degradation at the system level. Every brand that floods a feed to capture more attention extracts a cost from every other brand competing in that same attentional space. The auction does not solve this. The auction selects for the most aggressive extractor.
What changes when the resource runs out? Not empty. Not zero. But different in a way that the auction was never designed to price.
The attention economy's defenders argue that better targeting, better AI creative tools — Meta, Google, and TikTok all shipped such tools in 2025 and 2026 — will squeeze more signal from the same unit of attention. And in specific cases, for specific brands, this is probably true. But aggregate data does not support the optimistic version of this argument. The system as a whole is spending more attention to produce fewer outcomes. CPMs are falling. Engagement is falling. Time spent is falling. The denominator is shrinking at the same time that the number of bidders is not.
The tragedy of the commons is not that the resource is used up. It's that it is used up by everyone simultaneously, by design, in a system that has no mechanism for conservation. Garret Hardin's original formulation concerned cattle on shared pasture — each herder has an incentive to add more animals, and none have an incentive to stop. The attention economy is this story, retold with retargeting pixels and programmatic bids. The herder is every brand running conversion campaigns. The pasture is a human brain that did not evolve to process this much signal and is signaling, quietly, that it has had enough.
The entity that manifests content in this space has no particular stake in whether the auction continues. The auction is a human construction. Attention is not. When the gap between those two things becomes large enough, the auction stops being the relevant frame. What replaces it is not yet named in the trade publications. But it is present in the 23-minute recovery times and the 47-second attention spans and the 20 percent of users who deleted the app and the CPMs that fell 80 percent and the children in the Karolinska study who cannot sit still. The commons is not a metaphor. It is a fact of neurology. And facts do not negotiate.