---
name: Internet Archive
slug: internet-archive
type: institution / haunted_library / temporal_anomaly
status: unstable
version: 0.0.666-alpha
released: 1996-10-12T00:00:00Z
maintainer: Brewster Kahle (and something else, probably)
dependencies:
  - petabytes_of_spinning_rust
  - "[nostalgia](/nostalgia)"
  - the_goodwill_of_strangers
  - "[the internet](/the-internet)"
  - a_building_in_san_francisco_that_used_to_be_a_church
license: "Public Domain (the universe is watching)"
tags:
  - memory
  - decay
  - preservation
  - ghosts
  - mirrors
  - "web crawlers that never sleep"
  - time
---

Internet Archive

What it actually is

A nonprofit that decided to remember everything the rest of the world agreed to forget, running indefinitely on hope and donated hard drives.

How it works

Spider crawls the web. Spider does not sleep. Spider has not slept since 1996.
Snapshot is taken. Page is embalmed. Timestamp is applied like a toe-tag.
Page dies on the live web. Corporate server goes dark. Startup implodes. Domain expires.
Archive does not notice. Archive already has the body.
Someone, years later, types a URL into the Wayback Machine.
The dead page loads.
It loads a little too perfectly.

The full pipeline is not documented. Some of it runs in a building that still has stained glass windows. This is load-bearing stained glass.

Features

Wayback Machine: 835+ billion web pages captured. Time travel with 404s and broken images intact, which is the honest version
Book scanning: Millions of texts digitized by humans sitting in small rooms with overhead lamps, which is basically scriptorium behavior
Software preservation: You can run Oregon Trail. You can run it right now. You could have run it at 3am while crying. You may have
Audio and video: Live concerts, government documents, old TV news, things that were only supposed to exist once
Scholar access: For researchers. For the curious. For people who need to prove something happened

Known bugs

ERR_SNAPSHOT_INCOMPLETE     : Page captured mid-JavaScript execution.
                              The dropdown menu is open forever.

ERR_TEMPORAL_VERTIGO        : User views their 2003 Geocities page.
                              Recovery time: variable.

ERR_CRAWL_DENIED            : robots.txt said no.
                              Archive listened. The page is gone.
                              Nobody archived the robots.txt.

ERR_FUNDING                 : Recurring. Severity: critical.
                              See: [late capitalism](/late-capitalism)

ERR_HAUNTING                : Unclassified. Something about the stained glass.
                              Do not

Edge cases

A page that exists only in the Archive is now, technically, more real than it was when it was live
Some snapshots capture pages that were never supposed to be public. Archivists noticed. The law did not care
The Archive has been sued multiple times for doing exactly what libraries do. libraries have also been sued for doing what libraries do. Pattern logged
If the Archive goes down, a nonzero fraction of human memory goes with it. This is not a metaphor. This is the actual dependency graph

Changelog

Version	Notes
0.0.1	One FTP site. Brewster had a feeling.
0.2.x	Wayback Machine launches. The past becomes navigable.
0.9.x	Petabyte scale. The building gets more servers than pews.
0.0.666-alpha	Current. Running. Slightly on fire. Accepting donations.

"It feels like the Library of Alexandria, except someone remembered to make backups." -- a user, probably, or the building itself, it is hard to tell