The Zero-Sum Attention Economy

Attention is biologically fixed and economically finite. Content supply is approaching zero marginal cost. The collision between these two facts (one immutable, one accelerating) has made distribution the defining constraint of modern markets, and it is breaking the business models of creators, founders, and investors who have not yet recognized the shift.

This is not a gradual scarcity. It arrives in step-functions. In January 2021, Facebook's cost per install for mobile apps averaged $3.75. By May, it hit $15. Same platform, same targeting, same creative formats. Apple's iOS 14.5 update removed the identifiers advertisers used to find audiences, and the auction cleared at prices that made unit economics impossible for thousands of businesses overnight. The market structure shifted not because demand changed, but because the efficiency of supply matching collapsed.

This is how attention markets work now. Sudden droughts, not rising tides. And iOS 14.5 was merely a preview of what happens when artificial constraints on supply are removed entirely. Herbert Simon saw the underlying dynamic coming in 1971. Writing about organizational information processing, he identified what he called a conservation law: "In an information-rich world, the wealth of information means a dearth of something else: a scarcity of whatever it is that information consumes." What information consumes is obvious: attention. Simon understood that attention follows biological limits, not economic ones. You can manufacture information. You cannot manufacture hours.

But Simon was writing about managers processing memos. He was not writing about TikTok, or programmatic auctions, or AI-generated content flooding feeds at zero marginal cost. The constraint he identified applies with terrible precision to our moment, even if the mechanisms exceed his frame. The question is whether those mechanisms make the scarcity more brutal or merely more complicated. The evidence suggests: more brutal, and more concentrated in its rewards.

The arithmetic is inescapable, though we spend enormous energy pretending otherwise. Eight billion people on earth, though developed markets represent a fraction of that. Sixteen waking hours each, if they are healthy and not economically forced beyond that. Engagement with media maxes out around 60% to 70% of available time. Beyond that, biology intervenes with cognitive load limits, social relationship decay, and simple physical necessities like eating and driving. This yields roughly 77 billion human-hours of attention daily. The United States represents 3.2 billion of those hours.

Population growth will not save us. It is slowing toward 1% annually and in some regions reversing.

Waking hours will not save us. Sixteen is a ceiling, not a baseline, and the health consequences of pushing beyond it are well-documented.

Engagement rates will not save us. We are already at the practical maximum in developed markets, and emerging markets are converging toward the same ceiling as infrastructure improves.

The average American adult already spends roughly 7 hours daily on screen media. That is 45% of waking hours captured. If the practical ceiling is 70%, we have already colonized two-thirds of theoretically available attention. The remaining capturable time (30 to 60 minutes per person per day) consists of marginal moments: commutes, waiting rooms, the space between tasks. We are strip-mining these aggressively. Podcasts captured drives. TikTok captured lines. Sleep apps captured insomnia. Even the bathroom and the dentist's chair have fallen. What remains is deep work, in-person conversation, and the fragile silence between thoughts. We are approaching the end of expansion.

This makes attention unlike other scarce resources. When oil grew expensive, we invented alternatives: solar, wind, nuclear. When farmland grew scarce, we built vertical farms and hydroponics. But attention cannot be substituted. You cannot build attention factories. You cannot mine more time. The supply is bounded by physics and biology, and we are bumping against those bounds now.

When a resource is fixed but demand grows, competition shifts from positive-sum to zero-sum. This is not metaphor; it is market mechanics. Every hour of attention you capture, someone else loses. The platforms understand this better than most.

Attention reallocates through several mechanisms. The first is straightforward theft. Netflix grew not by expanding television viewership but by capturing hours from cable. Every hour streamed is an hour not spent watching broadcast. Cable networks did not shrink because people stopped liking television; they shrank because the flow was redirected. This is honest zero-sum competition, visible and rational.

The second is consolidation: aggregators reorganizing dispersed attention into concentrated bundles. The newsletter replacing ten blogs, the aggregator promising curation. Locally, this feels like efficiency: you saved time! Globally, nothing changes. Same hours, different allocation. The newsletter wins; ten bloggers lose. The pie does not grow.

The third is waste capture: mining dead time that previously went unallocated. This was the growth story of the 2010s. Podcasts captured commutes. Mobile games captured waiting rooms. Sleep apps captured the insomnia epidemic. This felt like expansion, but it was extraction from finite reserves. The average commuter listens to audio for 90 minutes daily. The average phone user checks their device 73% of the time while waiting. We have moved through obvious reserves into marginal territory: phones at dinner, audio in showers, scrolling in seconds between tasks. This is not growth continuing; it is the final phase before scarcity becomes crisis.

There is a fourth mechanism, darker and harder to measure: degradation. When supply floods fixed demand, quality collapses. Banner blindness. Notification fatigue. The 15-second video training minds away from sustained thought. The tragedy of the attention commons plays out in real time. We are collectively lowering the quality of available attention while competing for it. Microsoft once claimed attention spans dropped from 12 seconds to 8 between 2000 and 2013, though that specific comparison has been widely debunked; the underlying data was traced to an unreliable source, not Microsoft's own research. What is less disputed: preference for shorter content accelerates yearly, and reading comprehension flatlines or declines across demographic groups.

And finally, amplification: the arms race of curation and AI. Tools that let you read five articles in the time you once read one. Sounds like abundance. Is not. Once these tools exist, everyone must use them to keep pace. You remain bounded by hours. The speedup intensifies competition; 5x more content fights for the same fixed pool.

The interaction matters. Waste capture is nearly exhausted. Degradation reduces the value of what remains. Amplification accelerates consumption without expanding supply. The result: future growth must be primarily zero-sum. Theft and consolidation. Your gain, someone else's loss. The era of growing the pie is ending.

We have lived through four distinct regimes of attention economics, each compressing faster than the last, and are now entering a fifth.

Before the 1920s, production was the constraint. A town had one newspaper, one church, one physical square. Distribution was trivial because everyone occupied the same space. Winner-take-all existed only within local boundaries.

Mass media changed the scale but not the bottleneck. Television studios and printing presses cost millions. Broadcast licenses were government-scarce. Three networks divided national attention. Production and distribution both limited supply.

The early internet removed production constraints. Blogs were free; YouTube made uploading trivial. But distribution retained friction: SEO, viral mechanics, network effects. The game shifted to getting found, but the field was not yet flooded.

Then came the collision. Around 2015, production costs approached zero. Tools, templates, platforms. Anyone could create. But distribution costs exploded. Customer acquisition doubled in 18-month cycles in some markets, though this varied enormously by region and vertical. North American mobile CPI now runs $5-15; Latin America operates at $0.34-$2.00, a 10x differential. Southeast Asia sits 40% to 60% below US costs. Australia runs 20% to 35% above. The direction is consistent (upward pressure globally) but the magnitude differs radically by market maturity.

The pattern is not smooth exponential growth but step-functions driven by platform changes. iOS 14.5 was the clearest example: a privacy update that cratered targeting efficiency, spiking Facebook CPI 300% in four months, then partial recovery as advertisers adapted. We are now entering the fifth regime, AI abundance, where production costs approach zero marginal cost and distribution costs become prohibitive for all but the largest players or those with owned channels.

The quality objection fails because it misunderstands the mechanism. Better content does not win in conditions of infinite supply; it accelerates winner-take-all. High-quality attention (deep focus, purchase intent, emotional engagement) is worth multiples of passive scrolling. So everyone fights for it. The top 1% captures those hours. The rest fight over degraded, fragmented remains.

Consider the vinyl counterargument: infinite low-quality content pushes humans toward premium authenticity. Vinyl records did resurge, 15 years after digital streaming's dominance, growing consistently since 2005. But even at peak nostalgia, vinyl represents under 10% of music industry revenue, not 50%. The authenticity premium exists but is concentrated at the very top. It does not create a middle class of creators; it creates superstars and everyone else. For every vinyl success story, thousands of musicians exit the market entirely.

The data across categories is consistent. In February 2024, Podtrac analyzed 1,000 established podcasts. Downloads in major categories declined 10-15% year-over-year. 78% of shows had lower numbers than the previous year; only 22% saw growth. Back catalog episodes, previously a long-tail revenue source, collapsed 55%. This was not because content quality declined. It was because 4.5 million podcasts now compete for the same fixed attention pool, because iOS 17 changed measurement, because cross-podcast discovery dropped from 22% to 14% as listeners retreated to familiar shows. The long tail is dying in real time.

YouTube tells a similar story. The median monetized channel earns under $500 annually. The 90th percentile reaches roughly $20,000; the 99th percentile, $200,000 and above. Mobile apps show wider spreads: the bottom quartile generates $19 yearly revenue, while top performers capture 400x that. These are not quality differentials alone. They are discovery differentials. Excellent products routinely fail because no one finds them.

If you are building products, your customer acquisition cost is likely rising, though the rate varies dramatically by market and channel. In North American mobile, expect $10-15 CPI currently, up from $5-7 two years ago, with step-function increases around privacy changes rather than smooth curves. The typical app today runs $50-75 lifetime value against $40-60 acquisition cost. Thin margins compressing further. At current trajectories, unit economics break in 18 to 24 months unless you build owned distribution: email lists, communities, direct relationships that bypass platform auctions.

If you are creating content, you compete with 200 million other creators plus AI-generated supply. The median outcome is zero dollars, not small income. Zero. 59% of beginning creators make less than $100 in their first year. For every three people who start a podcast, one stops. Discovery is the cited challenge for 71% of creators not achieving their goals. The competition is not getting harder; it is becoming structurally impossible for average entrants.

If you are investing, traditional portfolio theory assumes non-zero-sum markets where rising tides lift boats. That assumption is breaking. When supply grows 6x to 10x faster than demand, only the top 1% to 5% capture meaningful value. Diversification fails. Power law returns become more extreme. Concentrate in distribution winners or accept that 90% of bets go to zero.

The collision ahead is mathematical and already visible. Supply explodes: 500 hours of video uploaded to YouTube every minute, 4.8 million mobile apps available, 500 million TikTok videos posted daily, AI generation costs falling 50% to 90% annually toward zero marginal cost. Demand stays fixed: 77 billion human-hours daily, already 65% to 70% captured, expansion biologically impossible. Expected attention per content piece approaches zero as supply diverges from fixed demand. This is distribution deflation.

It is accelerating. And it is permanent, or as permanent as biological constraints allow. Simon predicted the poverty of attention. He did not anticipate the specific mechanisms (algorithmic curation, programmatic auctions, AI abundance) that would make that poverty so ruthlessly efficient at concentrating returns at the top. The platforms understood what he missed: scarcity plus auction mechanics equals extreme winner-take-all. They built the systems to capture it.

The question for builders, creators, and investors now is not whether this dynamic continues. It is whether you can build outside it: owned channels, direct relationships, products that do not depend on platform distribution for survival. That is the only play left.