What's this about?
A team found a 2009–2014 database backup from 0.mk, a link shortener used mostly in Macedonia, containing 657,958 links along with their click counts. They restored 657,607 records and crawled every destination in August 2026. Of 655,178 safe, crawlable records, 76.7% no longer return a loading page.
What exactly is "link rot"?
Link rot is the steady decay of the web's connective tissue: URLs stop working as domains lapse, companies shut down, sites restructure, and hosts vanish. It matters because links are how the web cites itself — a citation to a dead URL preserves the claim but destroys the evidence. Unlike a library, nobody is responsible for the web's continuity, so the default is loss. The authors are careful that this is one community's slice, not a census of the whole web.
What's the community saying?
- The 76.7% figure over roughly a dozen years is the number people carried away — worse than most expected, and a concrete argument for archiving anything you cite.
- Considerable discussion of the Internet Archive's role and the fact that a single nonprofit is carrying most of the world's link-preservation burden.
- The dataset's specific character — Macedonian local news, personal blogs, photo hosts, forums — made the loss feel less abstract: the casualties are small independent sites, while the large platforms of the era mostly survive.