Nearly 40% of 2013 Webpages Have Vanished, Shrinking AI Training Data
Reactions from ranked influencers
7 posts
We once believed something on the Internet would last forever. Instead we face a memory hole larger than any previous generation has known, with entire platforms and decades of conversation erased by corporate decisions and technical neglect. https://readmultiplex.com/2026/07/22/nearly-40-of-webpages-that-existed-in-2013-are-just-gone-ai-was-trained-mostly-on-the-leftover-parts-of-the-internet-that-were-easy-to-monetize/
WE ARE ACTIVELY RAISING AN AMNESIA GENERATION._ When primary sources from the recent digital past are missing, later interpretations rest only on whatever survived commercial filters. Younger generations are inheriting a form of structural amnesia. Early forum platforms, Yahoo Groups, and private bulletin boards contained the unfiltered history of human technological adaptation. When corporate winds shifted, they were deleted without warning. Read how this impacts your life below…
We once believed something on the Internet would last forever. Instead we face a memory hole larger than any previous generation has known, with entire platforms and decades of conversation erased by corporate decisions and technical neglect. https://readmultiplex.com/2026/07/22/nearly-40-of-webpages-that-existed-in-2013-are-just-gone-ai-was-trained-mostly-on-the-leftover-parts-of-the-internet-that-were-easy-to-monetize/
The loss of unscripted collaborative cognition. The erased data carried distinctive value. Early private forums and listservers contained unscripted exchanges on technical problem- solving, philosophy, and technological change. Participants reasoned together across months of threaded conversation without optimization pressures, threaded conversation without optimization pressures, Read more here:
We once believed something on the Internet would last forever. Instead we face a memory hole larger than any previous generation has known, with entire platforms and decades of conversation erased by corporate decisions and technical neglect. https://readmultiplex.com/2026/07/22/nearly-40-of-webpages-that-existed-in-2013-are-just-gone-ai-was-trained-mostly-on-the-leftover-parts-of-the-internet-that-were-easy-to-monetize/
“The future will know us by what we refused to let disappear.”—Brian Roemmele, 1979 It defined my life.
We once believed something on the Internet would last forever. Instead we face a memory hole larger than any previous generation has known, with entire platforms and decades of conversation erased by corporate decisions and technical neglect. https://readmultiplex.com/2026/07/22/nearly-40-of-webpages-that-existed-in-2013-are-just-gone-ai-was-trained-mostly-on-the-leftover-parts-of-the-internet-that-were-easy-to-monetize/
Frontier AI is inheriting a sanitized version of human history! Only <5% Modern Data Is Used For AI Training Data. AI models learn from the fragments that remain. They will carry forward the gaps, distortions, and amnesia. Read more below…
We once believed something on the Internet would last forever. Instead we face a memory hole larger than any previous generation has known, with entire platforms and decades of conversation erased by corporate decisions and technical neglect. https://readmultiplex.com/2026/07/22/nearly-40-of-webpages-that-existed-in-2013-are-just-gone-ai-was-trained-mostly-on-the-leftover-parts-of-the-internet-that-were-easy-to-monetize/
Preservation is the new counter-culture. Individual archivists, librarians, and those running garage labs are the modern monks of the digital age. Read more below…
We once believed something on the Internet would last forever. Instead we face a memory hole larger than any previous generation has known, with entire platforms and decades of conversation erased by corporate decisions and technical neglect. https://readmultiplex.com/2026/07/22/nearly-40-of-webpages-that-existed-in-2013-are-just-gone-ai-was-trained-mostly-on-the-leftover-parts-of-the-internet-that-were-easy-to-monetize/