A new essay in The Walrus warns that Google Search is dying and taking the internet's collective memory with it. Author Vass Bednar opens with an AI answer that confidently got a basic fact wrong, then argues the world's dominant search engine has lost its edge even as billions still rely on it. The piece ignited Hacker News, drawing more than 850 points and 850 comments within a day.
The deeper problem is a retrieval crisis. AI systems now scrape and ingest Wikipedia directly, presenting summaries that eliminate the need to click through, so the encyclopedia's traffic and donations are drying up. Bednar calls Wikipedia the infrastructure of its own demise. Meanwhile, 404 Media reports companies are planting content on Reddit to steer AI search answers, contaminating the public record. Even the Wayback Machine is buckling under engineering strain, cyberattacks, and costly litigation, and at least one outlet deleted its entire archive over ad revenue.
For founders, the stakes are structural. If AI summaries replace search clicks, websites lose the traffic that funds them, and the corpus those models train on degrades in turn. The essay argues that what comes next, a web shaped by AI-generated answers and planted content, may be worse than the search era it replaces. The question is whether the open web can survive as its own memory fades.