The reason a shutdown erases a decade of work

Eleven years of posts, revisions and comment threads, referenced in three hundred other places. The service announces a wind-down with ninety days’ notice and an export button that returns a JSON file nobody can render. The archive did not degrade. It was switched off.

What everyone sees

Digital work is assumed to be durable because copying is free, so anything valuable will have been copied. Copying being possible is not the same as it having been done. The material lived in a system whose continued operation was the only thing making it readable, and that operation was always a business decision.

What is actually happening

Zittrain, Albert and Lessig measured reference rot in legal and scholarly citations and found that a substantial majority of links in Supreme Court opinions and law review articles no longer pointed to the cited material — loss occurring quietly, across sources everyone assumed were permanent. Rosenthal and colleagues set out what preservation actually requires: independent replication across administrative domains, format documentation, and periodic verification, none of which a single operator provides by default. The gap between those two findings is where the decade goes. Material held in one operator’s system has a lifespan equal to that operator’s interest in hosting it, and there is no notice period in which three hundred inbound references can be repointed.

Why it stays hidden

The exposure hides because everything works. Availability yesterday is read as evidence of availability tomorrow, and there is no signal that distinguishes a service which will exist in five years from one which will not. Export features make it worse, since their presence implies a preserved copy exists somewhere.