// attention economy

All signals tagged with this topic

Netflix Faces Choice Between Finishing Content and Discovery

Netflix's recommendation engine faces a choice: prioritize completion rates (a metric that shows user stickiness) or prioritize discovery and catalog diversity (which can reduce churn and boost long-tail content). Optimizing for one actively suppresses the other. A completion-focused algorithm narrows suggestions toward similar content; a diversity-focused one risks surfacing content users abandon, which tanks completion metrics that Wall Street monitors. Most platforms default to completion because it's measurable and immediate. But studios, creators, and subscribers have incentives tied to the underutilized 90% of Netflix's catalog.

YouTube Overtakes Spotify as UK's Leading Podcast Platform

YouTube's ascent reflects a shift in how consumers discover audio content—they increasingly follow creators across formats rather than stay loyal to dedicated podcast apps. This collapses the distinction between video and audio consumption, giving YouTube's recommendation engine and ad stack a new revenue advantage while forcing Spotify to compete on something other than exclusive deals and podcaster payments. The winner will be whoever best monetizes passive listening and bundled entertainment consumption at scale.

Ivy League professor's in-person exam reveals widespread AI use in remote testing

A single data point—a 50% score drop when an Ivy League professor moved from remote to proctored exams—shows how thoroughly generative AI has infiltrated elite academic assessment. This isn't about student intelligence or character. It's about the collapse of an evaluation infrastructure built on the assumption that remote testing could maintain integrity without synchronous human oversight. Universities now face a choice: expensive real-time proctoring, redesigned assessments that can't be gamed by LLMs, or accepting that the traditional transcript has lost its signal value as a credential.

Brown Professor Documents Widespread AI Cheating in Economics Class

When an economics professor identifies suspicious patterns in student work with statistical rigor—flagging anomalies in writing style, error patterns, and problem-solving approaches—AI detection moves beyond subjective judgment into measurable fraud detection. Institutions lack standardized protocols for identifying at-scale cheating, leaving individual instructors to build detection frameworks alone while students operate under unclear rules about what constitutes acceptable AI use. This erodes confidence in grades and credentials at a moment when employers and graduate programs are already scrutinizing recent transcripts more closely.

A $185 Device Bets Consumers Will Pay to Avoid Their Phones

The modern "dumb device" category—epitomized by products like the Humane AI Pin and now a redesigned Palm Pilot—rests on a specific bet: that smartphone fatigue commands a premium price. It's a direct challenge to the smartphone's dominance in productivity, testing whether consumers will fragment their tools again if given friction-free alternatives. The economics require that distraction be costly enough to justify carrying, charging, and paying for a separate device. That assumes willpower scales as a consumer good, not a practice.

Parents increasingly anxious about children's AI dependency in schools

A survey on parental concerns about AI overuse in primary education exposes a generational divide over technology integration. Parents raised before widespread digital adoption are skeptical of tools their children already treat as cognitive utilities. This anxiety shapes purchasing decisions, school board politics, and family adoption of AI-enabled learning products—pitting ed-tech vendors pushing algorithmic tutoring against parents demanding evidence that AI isn't replacing their children's critical thinking.

Paywalls Erode Newsletter Writer Influence, Even for Top Creators

Noah Smith's admission that Substack's paid model has reduced his reach and impact cuts against the platform's core claim: that direct reader relationships generate sustainable income without sacrificing influence. The tension is real. Charging money shrinks your audience, which diminishes your ability to shape public conversation or attract high-profile collaborators. Writers face a choice between revenue and relevance. This exposes a miscalculation in Substack's pitch to ambitious writers—especially those whose influence depends on cultural cachet rather than niche expertise.

The End of AI's Novelty Discount

As frontier AI labs stop subsidizing cheap compute to drive adoption, users must now justify each query against real costs. AI shifts from experimental playground to utilitarian tool optimized for efficiency rather than exploration. This favors power users and established workflows over casual experimentation, likely concentrating AI value capture among those with high-stakes problems—knowledge workers, developers, analysts—while consumer AI adoption may plateau unless vendors can justify pricing through genuine productivity gains rather than discounted wonder.

How Medieval Guilds Solved the AI Trust Problem

Rather than wait for perfectly reliable AI systems, this piece proposes borrowing institutional scaffolding from medieval guilds—QA functions, review boards, appeals processes—to make unreliable agents accountable through structure rather than capability. Consumers don't need to trust the technology itself if they trust the organization operating it, which inverts how most AI companies frame the adoption problem. The near-term competitive advantage belongs to platforms that can layer traditional institutional practices around AI outputs, not those chasing alignment or interpretability breakthroughs.

Netflix's Audience Problem: Why Viewers Stop Watching

Netflix's retention is deteriorating. Viewers abandon shows mid-season across genres, and content volume alone no longer justifies subscription costs as competition fragments viewing habits. The company's pivot to advertising and paid sharing signals an internal acceptance that churn is structural, not cyclical—a shift from growth-at-all-costs to extracting more revenue per user. This reveals a gap in streaming's original promise: a single subscription replacing cable with everything audiences want. Instead, audiences now juggle four to six services to find something worth watching.