Tim Berners-Lee's 1998 W3C style note arguing that links should never break has made another trip to the Hacker News front page, collecting 112 points from a community whose median member was not yet born when it was written.
The core claim is that URIs are stable in theory and rotten only in practice: the domain owner controls the entire namespace beneath it, so a broken link is almost always an organizational failure, not a technical one. The only respectable excuse for a document vanishing is the owning company going out of business. Everything else — site redesigns, staff changes, 'we moved the files' — gets dismantled in turn. Web servers like Apache, he notes, let you map an abstract URI space onto whatever filesystem mess you actually have; exposing implementation details in the address is a choice.
He is particularly merciless about URIs that leak their machinery: 'cgi-bin' and '.pl' giveaways, status directories like 'draft' or 'old', author names, and subject classifications, all of which are guaranteed to change. His 1998 exhibit is the National Science Foundation, whose document URL http://www.nsf.gov/cgi-bin/getpub?nsf9814 he contrasts unfavorably with http://www.nsf.gov/pubs/1998/nsf9814/nsf9814.htm — the latter good enough, he imagines, to survive into 2098 without embarrassing whoever maintains the archive.
The design prescription is austere: start with the creation date (the one fact about a document that will never change), then leave out everything else — author, subject, status, access level, file extension, software mechanism, disk name. The same discipline applies to hostnames: 'cgi.pathfinder.com' and 'secure' subdomains are future link-rot at birth, and redirection and proxying can hide many servers behind one apparent host. He also dismisses the hope that URNs would fix persistence, arguing that any organization capable of minting durable URNs could simply use them as HTTP URIs today.
A practical footnote explains Apache content negotiation: keep the extension on the file (mydog.png) but link without it (mydog), letting the server pick the best available format now and in the future.
The piece closes with a 'hall of flame': a Boston TV station whose school-closings URL silently went blank while a new one appeared elsewhere on the site, and Microsoft's Netmeeting client shipping a built-in 'Free stuff' menu link that returned 404. The author's own W3C pages, he admits, fail his standard too — 'do what I say, not what I do.'
A historical note appended later explains for younger readers that at the end of the 20th century, 'cool' was a term of approval among the young — which may be the most poignant admission of bit rot in the entire document.
Comments
No comments yet — be the first.
Open the discussion
No account or password needed — just enter your e-mail and we’ll send you a one-time sign-in link. First time here? You’re set up automatically.
Your rating will be applied automatically after you sign in.
Check your inbox
We’ve sent a sign-in link to …. Open it on this device — this tab will sign you in automatically.
Nothing arrived? Check your spam folder — and mark the mail as "Not spam" so it lands in your inbox next time.