The End of the Internet: From Open Blogs to AI-Driven Vulnerabilities

From Open Blogging to Subscription Niches

Thompson traces the rise of the blogosphere in the early 2000s, noting pioneers such as Dave Winer (1997), Justin Hall (1994) and the 1998 launch of kottke.org. By 2003, political blogs like Talking Points Memo, Andrew Sullivan’s platform and Boing Boing proliferated, turning the web from a technologist-centric space into a venue for mass-reach commentary.

Google’s search engine became the decisive aggregator, offering free distribution but monetising discovery through AdSense. Thompson observes that as ad inventory swelled, payouts shrank and intrusiveness grew, illustrating “the Internet fails” when openness creates infinite competition.

A pivot to subscription models emerged as a sustainable alternative. Thompson cites Sullivan’s near-million-dollar annual revenue from reader subscriptions, emphasizing that niche content with low elasticity of substitution can command premium prices. He acknowledges the trade-off: scarcity limits readership, yet for specialized audiences, higher per-user revenue outweighs mass-scale ad dollars.

Social Media’s Contraction into Private Groups

The next wave of discovery arrived with social platforms. Early Twitter functioned as a public community where “interesting conversations happened in public,” according to Thompson. However, the ease of entry and algorithmic curation amplified content volume, forcing platforms to prioritise engagement over genuine discourse.

In contrast, “group chats” now dominate, offering high-trust, small-scale interaction. He frames this as a “redemption” of social media: instead of scaling audiences, the value lies in proliferating “infinite finite groups.”

AI-Accelerated Security Crises Signal an Endgame

The article lists recent security incidents: a critical PHP remote-code-execution bug, compromised intermediate CA keys, new vulnerabilities in Node.js, GitLab, Java and Cisco products, plus supply-chain breaches at LiteLLM and Railway. Thompson stresses that these are only a “partial list” and predicts worsening conditions.

He attributes the escalation to AI, which provides attackers with “limitless time and the ability to substitute iteration for cleverness.” While acknowledging that defenders also gain AI tools, Thompson foresees a period where software is rewritten and defended by AI, ultimately improving security but only after a turbulent transition.

The Cumulative Shift

Combining these trends, Thompson concludes that the Internet’s original promise of open, universal distribution is giving way to fragmented, subscription-driven content and private social structures, all under the shadow of accelerating security threats. The “end of the Internet” he describes is less a sudden collapse and more a gradual reconfiguration of how value is created, discovered, and protected online.