YAHEW!
A FIELD GUIDE TO THE OLD WEB

Sources: Link rot V · Extinction

← Back to the Link rot plate

Source. Every percentage on this page is verbatim from a named study and none of it is interpolated. The 38 of 100, the quarter of all pages between 2013 and 2023, the one in five over two years, the 11% of Wikipedia references, the 21% of government pages and the 23% of news pages are all from Pew Research Center, When Online Content Disappears, 17 May 2024, sampling roughly a million pages from Common Crawl. The 50% for US Supreme Court opinions and the figure above 70% for the Harvard Law Review, the Harvard Journal of Law and Technology and the Harvard Human Rights Journal are read off page 178 of Zittrain, Albert and Lessig, 127 Harv. L. Rev. F. 176, 178 (2014). The one in five and seven in ten for scholarly articles are from Klein et al., Scholarly Context Not Found, PLOS ONE, 26 December 2014, which is CC BY 4.0. The trillion is the Archive's own self-reported figure read off its markup on 2 August 2026, which is why this page says the Archive says it rather than stating it.

The mechanism is observed, not described from memory. The cs_, js_, im_ and id_ suffixes and the bundle-playback.js, wombat.js, ruffle.js load order are read directly out of an archived copy of geocities.com from 25 October 2009, saved in reference/5-extinction/link-rot/. The Google redirect was observed on 2 August 2026 and is reproducible. The banner above is a recreation: the real one carries no stated license, so its CSS was re-implemented rather than pasted.

Deliberately absent. Any figure for the average lifespan of a web page. The numbers that circulate, 44 days, 75 days, 100 days, two and a half years, could not be traced to a primary study that still stands up, so none of them appear here. Also absent: any claim about what share of the whole web the Internet Archive holds, because that is not knowable from these sources.

← Back to the Link rot plate · The directory