labs.llc DomainChaosHistory
Light — the edition you are reading
Look up

Domain history · from the archive

What it was,
before it was this.

A domain's past, from the Internet Archive's Wayback Machine: when its home page was first captured, how often it changed, what the crawler was served each year — set on one scale beside the date the registry says it was last registered, so a site that predates its own registration shows.

DC‑H1 Archive receiver

Wayback CDX · Registry

The archive usually takes eight to sixteen seconds to answer; the registry, under two. Both are asked at once.

Instrument
Try

Source The Internet Archive's Wayback Machine (its CDX index at web.archive.org), and the registry's own record for the registration date. The archive is partial: it holds what its crawlers and visitors happened to save, not a complete history.

What a reading holds

Six panels for any name the archive has seen. A name it has never captured gets a plain answer saying so — which is not the same as a name that was never used.

01

The tuning scale

Captures per year as a bar band, content changes as ticks, and the registration date as the needle — one time scale for all three.

02

Before and after

Whether any capture predates the current registration — the sign of a previous owner, or a lapse.

03

Captures by year

How many, what the crawler was served — pages, redirects, refusals, errors — and one snapshot a year.

04

Change points

The captures whose content differs from the one before, each linked to the archived page.

05

Silent stretches

Runs of a year or more with no capture — which is nobody saving it, not the site being down.

06

Sources

The archive's index and the registry, when each was asked, and whether the answer came from the cache.

Reading the archive

What the archive remembers — and what it cannot.

The Wayback Machine is the nearest thing the web has to a public record of itself, and it is an extraordinary one. It is also an accident of what was crawled and what somebody chose to save. Here is how to read it, and how this page reads it for you.

01A public archive, and a partial one

The Internet Archive, a non-profit library in San Francisco, has been crawling the public web since 1996 and opened the Wayback Machine to everybody in 2001. Its collection is made by its own crawlers, by partner crawls, and — since 2013 — by anyone who presses Save Page Now. That is why it is so uneven: a famous home page is captured many times a day, a small business's a handful of times a decade, and plenty of names never at all.

So an empty year means nobody saved the page that year. It does not mean the site was down, and a name with no captures is not proof that it was never used. Site owners can also ask for their pages to be excluded, pages behind a login are never captured, and a site built entirely in script is sometimes captured as an empty shell. Every count on this page is a count of what the archive holds — never of what existed.

02What a capture is, and what a change is

Behind the Wayback Machine sits an index — the CDX server — with one line per capture: the moment it was taken, to the second and in UTC; the HTTP status the crawler received; the kind of content; and a digest, a fingerprint of the exact bytes that came back. This page asks that index about the name's home page.

A change point is a capture whose digest differs from the one before it. That is a change in bytes, not necessarily in meaning: a redesign is a change, and so is a new date in a footer, a rotated advert or a fresh security token. Many changes in a year says the page was busy, or dynamic; few says it stood still — or that few people saved it.

03Captures from before the registration

The creation date a registry publishes is the date of the current registration, not the first one ever. When a name lapses and is registered again — by somebody new, or by the old owner after letting it go — the date starts over. So when the archive holds captures older than that date, the name was in use before this registration: a previous owner, or a lapse. The archive cannot say which, and neither can this page.

It is worth knowing before you buy a name, or when you are trying to understand one. A name with a past can arrive with backlinks and search reputation, good or bad, with old email still being sent to it, and occasionally with a trademark dispute attached. The change points either side of the gap are usually the quickest way to see what it was: a business, a blog, a parking page with a price on it.

Registration histories — who held a name, and when — are not public. Commercial databases sell reconstructions of them; this page does not use them, and does not guess.

04What the crawler was served

Each year's bar is split by the class of answer the archive recorded. Read together they tell a story: a long run of pages, then redirects, is often a name folded into another brand; refusals and missing pages are often a site winding down, or blocking crawlers.

StatusWhat it means
200A page was served — this is the capture you can read.
301 302 307 308A redirect: to https, to www, or to another name altogether. The snapshot shows where it went.
403 404 410Refused or missing. A server that turned the crawler away, or a page that was not there.
5xxThe server failed at the moment of the capture.
-No status recorded — usually a “revisit” record, which points back to an earlier capture with identical content.

05How this page works

  1. The name you give is reduced to its registrable domain — https://www.bbc.co.uk/news becomes bbc.co.uk — using the Public Suffix List.
  2. This site's server asks the archive's CDX index for every capture of that home page, collapsed where the content did not change, and counts them by year and by status.
  3. At the same moment the WHOIS engine asks the registry for the name's current record — the creation date, the registrar and the expiry.
  4. Both are drawn on one time scale. Captures older than the creation date are drawn hatched and counted, and said so in words.

The archive is slow — eight to sixteen seconds is normal — so each answer is cached for a day, each address is limited to six readings a minute, and the archive itself is asked no more than ten times a minute from here. No archived page is embedded or shown as an image: every snapshot is a link to web.archive.org, where the archive shows it in its own frame, on its own terms. No affiliate links, no account, and nothing you look up is kept beyond the cache.

What it will never claim

Four answers this page refuses to give.

01

It will not say who owned it.

Captures from before the registration say the name was in use before — a previous owner, or a lapse. Which one, and who, is not in the archive, and this page will not guess it from anywhere else.

02

A gap is not downtime.

A year with no capture is a year nobody saved the page. The site may have been up every day of it. The page says no capture, never “offline”.

03

A failure is not an empty history.

When the archive does not answer — it is often slow, and sometimes down — the reading is Unknown, with a way to ask again. It is never shown as “no captures”.

04

A missing date stays missing.

Some registries publish no creation date. Then the scale carries no registration needle and the page says so, rather than borrowing the first capture as a stand-in.