Skip to main content
The Ezra Klein Show

Everything Wrong With the Internet and How to Fix It

87 min episode · 3 min read
·

Episode

87 min

Read time

3 min

AI-Generated Summary

Key Takeaways

  • Enshittification Pattern: Platforms follow a three-stage decay: first attract users with good service while creating lock-in through network effects, then degrade user experience to attract business customers, finally extract from both groups simultaneously. Facebook moved from promising no surveillance in 2006 to algorithmic feeds where under 20% of content comes from accounts users follow, with the remainder algorithmically selected for engagement maximization.
  • Amazon Marketplace Economics: Amazon's take from third-party sellers increased from below 20% around 2010 to over 50% today. The platform generated $56 billion in 2023 just from sellers bidding against each other for higher search rankings, making this more profitable than Amazon Web Services. The first search result averages 17% more expensive than the best match for user queries, demonstrating extraction over service quality.
  • Algorithmic Wage Discrimination: Nursing staffing apps purchase workers' credit histories from data brokers and offer lower wages to nurses carrying more debt or overdue payments. This desperation premium creates a vicious cycle where economically vulnerable workers receive progressively lower pay for identical work. The practice relies on lack of federal privacy law—the last consumer privacy legislation passed in 1988 covered only VHS rental records.
  • Anti-Circumvention Barrier: Section 1201 of the Digital Millennium Copyright Act makes it a felony to modify apps or devices without manufacturer permission, even for legal purposes. When two teenagers created OG App in 2022 to show chronological Instagram feeds without ads, Meta had Apple and Google remove it from app stores. This prevents competitive alternatives that serve user interests over platform extraction.
  • Utility Regulation Framework: Essential platforms should face non-discrimination duties preventing self-preferencing and potentially margin caps around 30%. The electric utility model succeeded not through perfect regulation but by preventing the network from extracting profits from innovations built on top of it. Platforms function as infrastructure where discrimination between users or products enables extraction rather than value creation for the broader economy.

What It Covers

Ezra Klein interviews tech critics Cory Doctorow and Tim Wu about platform decay through "enshittification" and "extraction." They trace how Facebook, Amazon, and other platforms shifted from serving users to maximizing profits, now taking 50%+ margins from sellers while degrading search quality. Solutions include interoperability mandates, privacy laws with private enforcement, and banning anti-circumvention rules that prevent users from modifying software they own.

Key Questions Answered

  • Enshittification Pattern: Platforms follow a three-stage decay: first attract users with good service while creating lock-in through network effects, then degrade user experience to attract business customers, finally extract from both groups simultaneously. Facebook moved from promising no surveillance in 2006 to algorithmic feeds where under 20% of content comes from accounts users follow, with the remainder algorithmically selected for engagement maximization.
  • Amazon Marketplace Economics: Amazon's take from third-party sellers increased from below 20% around 2010 to over 50% today. The platform generated $56 billion in 2023 just from sellers bidding against each other for higher search rankings, making this more profitable than Amazon Web Services. The first search result averages 17% more expensive than the best match for user queries, demonstrating extraction over service quality.
  • Algorithmic Wage Discrimination: Nursing staffing apps purchase workers' credit histories from data brokers and offer lower wages to nurses carrying more debt or overdue payments. This desperation premium creates a vicious cycle where economically vulnerable workers receive progressively lower pay for identical work. The practice relies on lack of federal privacy law—the last consumer privacy legislation passed in 1988 covered only VHS rental records.
  • Anti-Circumvention Barrier: Section 1201 of the Digital Millennium Copyright Act makes it a felony to modify apps or devices without manufacturer permission, even for legal purposes. When two teenagers created OG App in 2022 to show chronological Instagram feeds without ads, Meta had Apple and Google remove it from app stores. This prevents competitive alternatives that serve user interests over platform extraction.
  • Utility Regulation Framework: Essential platforms should face non-discrimination duties preventing self-preferencing and potentially margin caps around 30%. The electric utility model succeeded not through perfect regulation but by preventing the network from extracting profits from innovations built on top of it. Platforms function as infrastructure where discrimination between users or products enables extraction rather than value creation for the broader economy.
  • Surveillance-Driven Business Models: Over 96% of Apple users opted out of Facebook tracking when given a simple choice, disproving revealed preference arguments that users accept surveillance. Microsoft Office now offers bosses the ability to stack-rank divisions by mouse movements and typing patterns, then compare this data against competitors. Amazon drivers face facial recognition monitoring that previously docked them for driving with mouths open, representing technology adoption curves starting with prisoners and refugees before reaching mainstream workers.

Notable Moment

Doctorow reveals that Facebook internally calculates 10% of its advertising revenue comes from ads for actual scams, and a Reuters investigation found the company charged advertisers premium rates for fraudulent video view counts during the pivot to video era. This led news organizations to restructure their entire businesses around fabricated metrics, demonstrating how extraction harms not just users but entire industries dependent on platform data.

Know someone who'd find this useful?

You just read a 3-minute summary of a 84-minute episode.

Get The Ezra Klein Show summarized like this every Monday — plus up to 2 more podcasts, free.

Pick Your Podcasts — Free

Keep Reading

More from The Ezra Klein Show

We summarize every new episode. Want them in your inbox?

Similar Episodes

Related episodes from other podcasts

This podcast is featured in Best Politics Podcasts (2026) — ranked and reviewed with AI summaries.

You're clearly into The Ezra Klein Show.

Every Monday, we deliver AI summaries of the latest episodes from The Ezra Klein Show and 192+ other podcasts. Free for up to 3 shows.

Start My Monday Digest

No credit card · Unsubscribe anytime