Nobody Can Correct Me About 1991
Go looking for proof of a specific afternoon in 1991 and the failure has a texture you don't expect. Gaps, fine, you were braced for gaps. The surprise is that the gap looks identical to the thing never having happened. No partial record, no degraded copy, nothing that survived badly enough to argue with. The query returns nothing, and nothing is exactly what the present returns for events that are simply fictional. Absence of evidence and evidence of absence collapse into the same blank page, silently.
The web's memory has a start date, and it's later than people assume. The Wayback Machine's captures go back to 1996, the year the Internet Archive was founded, and almost everything before that floor is online because somebody later decided to put it there. Which makes it back-fill rather than a record. A 1993 photograph is searchable today because a particular person, at some point after 2004, owned a scanner and had a reason. What survives of that world was therefore selected twice: once by whatever chance preserved the physical object, and again by whoever later felt strongly enough to digitise it. Bands get back-filled. Football clubs get back-filled. Ordinary Tuesdays don't.
None of which means the period went unrecorded. It was documented at the wrong resolution. Newspapers, electoral rolls, planning applications, the local paper's account of a factory closing in March 1992. You can establish the closure to the week and find nothing whatsoever of the people who worked there, which is usually who you were looking for. Individuals appear in that record as a line in a register, present but not described. I've written before about the way information had mass in that period, how knowing something required physical movement. The same physicality governed being known.
The asymmetry that does the damage isn't about volume. The earlier half of a life can't be audited. I can be corrected about 2016 by a timestamp: someone produces a message showing I've misremembered the order of events, and I concede, because the external record outranks me. Very little performs that function for 1991. A payslip can settle a date and an electoral roll can settle an address, so the auditing isn't zero, it's just confined to the handful of facts an institution had a reason to write down. Everything the memory is actually made of, the sequence, the texture, who said what and how it landed, runs unchecked. It drifts the way memory always drifts, and no mechanism anywhere will ever catch the drift. The comparative baseline isn't only unsaved. Most of it was never falsifiable in the first place.
Being fair to the other side of the boundary complicates this rather than dissolving it. The continuous archaeological layer we've been depositing since about 2000 erodes while it forms. Pew Research found that 38% of web pages that existed in 2013 were unreachable a decade later, and that a quarter of all pages from the 2013 to 2023 span have gone. Jason Scott of the Internet Archive put the physics of it well in a piece Adrienne LaFrance wrote for The Atlantic: a piece of paper can burn and you can still get something from it, whereas with a hard drive or a URL, when it's gone there's zero recourse.
The two regimes still fail differently. Modern loss leaves a shape behind. Pew could count the missing 2013 pages because a list of them existed to check against, and a dead link is itself a durable record that something was once there, even when the contents are unrecoverable. That's what makes web archaeology possible at all: Peter Webster reconstructed a late-1990s web sphere of conservative British Christian sites by working outward through hyperlink data in the UK Web Archive, inferring the shape of what existed from traces in the pages that survived. I found that work through a British Library blog post about it, which is now a 404. I checked twice, because it seemed too neat.
For 1991 there isn't even a dead link to fail to follow. Nobody can tell you what proportion of that year is missing at the level of an individual life, because constructing the denominator would require exactly the archive we're saying didn't exist. The people who were there are the index now, and a biological index is unversioned, unaudited, and shrinking by attrition. When you interrogate the present for evidence of that world, the present isn't withholding anything. It was never asked to keep the file.
Sources:
-
When Online Content Disappears — Pew Research Center, May 2024
-
Raiders of the Lost Web — Adrienne LaFrance, The Atlantic
-
We're losing our digital history. Can the Internet Archive save it? — BBC Future
-
The Past Web: Exploring Web Archives — Peter Webster, "Digital archaeology in the web of links"
Filed under Hauntology & the uncanny · Pre-internet technology
This post is timestamped using Blockchain technology. Verify