Why the recent past is harder to verify than the distant one

This looks at the recent past from the practical end — what holds up once conditions stop being ideal.

Paper survives being ignored

A box of letters in an attic remains readable after a century of complete neglect. A hard drive, a floppy disk or a proprietary file format requires active migration to remain readable at all. Preservation of digital material is therefore a continuous cost rather than a one-off act, and almost nobody pays it for personal material.

This inverts the usual assumption that newer material is better preserved.

Link rot is measurable and severe

Studies of academic and legal citations repeatedly find that a large fraction of referenced URLs no longer resolve within a decade. This means the evidence base of published work is quietly degrading in a way footnotes to printed books do not. Web archiving projects capture a substantial amount and by construction cannot capture everything, particularly anything behind a login.

The workable defence is to archive a page at the moment you cite it and to reference that snapshot alongside the original, which several journals and courts now require precisely because the underlying problem is not going to be fixed.

Platforms are not archives

Services that hosted enormous volumes of personal material have closed, deleted content or removed access with limited notice. Where an archive was made it was usually by volunteers acting against the clock rather than by the platform.

What is easy to forget is that anything held only inside a commercial service should be understood as temporary, because it is. Deletion also arrives without a notice anybody reads, since a revised term, an inactivity policy or an account closed over a failed payment removes material through routine administration rather than through a decision that gets announced.

The formats themselves expire

Proprietary word processor formats, early video codecs and disc media from the 1990s are frequently unreadable without specialist effort. Open, well-documented formats are readable decades later, which is why archivists insist on them.

For a while, the practical advice is unglamorous: plain text, open formats, multiple copies, periodic migration. The hardest cases are not documents at all but things that were software — a site with a database behind it, a game that needs a server nobody runs, a work that only ever ran on one machine — none of which survives as a file you could copy.

What this does to history

Historians of the late twentieth century increasingly report gaps where personal, informal and ephemeral material should be. Institutional records survive better than personal ones, which biases the record toward organisations and away from ordinary life.

That is the opposite of the trend across most of the preceding century, and it is a genuine problem rather than a lament. What does survive online skews heavily towards the loud and the public, so an account of this period assembled from it would describe the arguments accurately and the ordinary silence around them not at all.

Dates in this period are approximate; the change was gradual and nobody logged it.

Abundance is its own kind of loss

The second failure is not scarcity but volume: an inbox holding two hundred thousand messages is perfectly preserved and effectively unusable without the context that made any of it mean anything. Archivists separate keeping material from describing it, and description is the expensive half that personal collections almost never receive. Dates are also less trustworthy than they appear, because file timestamps are silently rewritten by copying, syncing, backup software and format migration.

A photograph carrying a wrong date is harder to place than an undated print, since the number invites belief in a way that a blank back never does.

The takeaway

Paper survives being forgotten. Everything since has to be actively remembered.

Nobody decided to stop. It simply stopped being worth the effort.