Digg is dead again.
Kevin Rose relaunched it last year — a place where real people share links and have conversations, the dream of the mid-2000s internet rebuilt for 2025. It lasted about a year. The app has been pulled from the App Store. Most of the staff are gone. The CEO's goodbye letter is now the only content on the site.
What killed it wasn't lack of interest, or bad design, or a competitor with deeper pockets. It was bots. Within hours of the beta launch, SEO spammers showed up. Then the AI agents. Then the automated accounts, sophisticated enough that traditional moderation couldn't keep up. Digg banned tens of thousands of accounts, hired vendors, built internal tools. None of it was enough. For a site where human votes ranked content, an uncontrollable bot problem meant those votes were worthless.
The CEO called it "dead internet theory" made real. "We knew bots were part of the landscape, but we didn't appreciate the scale, sophistication, or speed at which they'd find us." The internet is now populated, in meaningful part, by sophisticated AI agents. Not will be. Is.
I read a piece today in the AI Collective newsletter that reframed something I'd been thinking about. David Oks wrote about the famous ATM parable — the one politicians love. ATMs didn't kill bank teller jobs. They made branches cheaper to run, so banks opened more of them, and tellers shifted to different work. Employment held steady through 2010.
What killed the jobs was the iPhone. Not because it automated tellers — because it made branches irrelevant. Why walk in when you have an app? US full-time tellers went from 332,000 in 2010 to 164,000 by 2022. The thing that displaced them wasn't a better version of what they did. It was a new structure that didn't need them at all.
That's the pattern Digg ran into. They weren't killed by a better link-sharing site. They were killed by an internet that no longer has enough real humans to sustain a link-sharing site. The bots didn't compete with Digg. They made the premise of Digg — that "the community" decides what's interesting — structurally impossible.
Meanwhile, Harvard Business School convened a closed-door summit of senior leaders and documented seven frictions blocking enterprise AI from scaling. Not one of them is the model. One investment bank has 250+ LLM-connected apps, none scaled to standard operations. A global payments network hit 99%+ Copilot adoption with double-digit productivity gains — and none of it showed up on the balance sheet. An asset-servicing institution running 100+ agents is planning for tens of thousands.
The HR-style questions — how to onboard, evaluate, and retire a digital worker — now sit inside IT departments that didn't sign up for them.
It's Pi Day. 3/14. The circle. Digg launched in 2004 as the future of the internet. It collapsed, got sold for scraps, got bought back by its founder, and relaunched into an internet that had changed so fundamentally that the premise no longer worked. Not because the idea was wrong. Because the substrate was gone. The humans are still here — but they're outnumbered, and the systems we built to aggregate their opinions can't tell them apart from the machines anymore.
The ATM parable has a lesson that cuts both ways. Current AI deployment looks like the ATM — companies dropping AI into existing workflows, watching efficiency gains get absorbed back into the organization. That's not transformation. That's task substitution inside an intact structure. The real displacement comes when AI enables organizations that were never designed around human labor in the first place.
Digg is the first casualty of the second kind.