The Wayback Machine is not a search result that forgot to die. It is a library that meant to keep the page. Australians who have just won a Google Search removal discover this in the worst way: they search their name, the live blue link is gone, and the second result is web.archive.org with a 2019 capture of the same article. Writing to Google about that URL is a category error. Google did not store it.

If you remember one distinction from this guide, make it this one: site owner versus subject. The Internet Archive will, in defined cases, exclude a host at the request of the person who publishes it. It will almost never unpublished a news story because the person named in the story has moved on. The rest of the search and archives hub is about leftovers. This piece is about the leftovers collected on purpose.

What the Archive actually is

The Internet Archive is a United States non-profit. The Wayback Machine is its web collection. Crawlers visit public URLs, store timestamped copies, and serve them on web.archive.org. Anyone can request a fresh capture of a page that is still live. Anyone can browse captures of pages that are not. That is the point of the project, and it is why reputation advice copied from a Google help page fails here.

The Archive is not a party to your dispute with the Herald. It did not write the article. A publisher’s 410 does not reach a WARC file written in 2016. It is not bound by an Australian court order unless the order names it, or unless the publisher uses the order in an exclusion request. Serving Google and assuming the copies will cascade is how people spend a year watching the same paragraph reappear in a different wrapper.

It is also not the National Library of Australia. Trove and the Australian Web Archive are a separate collection with a separate mandate — the archive Australians actually have a statutory relationship with, and the one most people email second, after they have already written the wrong first letter.

How to write to archive.org

The public door is still an email, not a polished portal. The address the help pages have pointed at for years is info@archive.org. That is a ticket queue. It is not a solicitor’s office and it is not a right-to-be-forgotten form.

Build a short file before you send anything.

  1. The exact capture URL, including the timestamp. A homepage on web.archive.org is not a request.
  2. The live URL, and a screenshot of what it does now — 404, 410, login wall, or still published.
  3. Who you are in relation to the host. Registrant, counsel for the publisher, or the person in the photograph. Those are three different jobs. Do not blur them.
  4. The hook in one sentence. “Please exclude this host, as the site owner.” “This capture displays my current home address.” “This capture is a non-consensual intimate image.” If you cannot write that sentence, you do not yet have an Archive job.
  5. Proof. A WHOIS record, a Search Console screenshot, a redacted bill, the order. Keep it off a compromised cloud account.

Site-owner requests are the ones that move. The Archive has long been willing to exclude a domain, or a path, when the publisher asks — particularly where the live site now blocks crawlers, has gone offline, or is being rebuilt after a hack. That is an operational courtesy to the person who publishes, not a privacy right for everyone mentioned on the site.

Subject requests move only in a narrow band. Personal data that is plainly out of place — a scanned identity document, a home address on a forgotten blog, an intimate image — can be put to them. Copyright owners can use the Archive’s DMCA path for their work, not for an article about them. Child sexual abuse material and non-consensual intimate images should go to the specialist channels first; the Archive is a second stop, and you should not attach the file to the email.

What does not move: “This old news story is unfair.” “The charges were dismissed and the publisher will not update the headline.” Those are news-archive jobs. Do them at the masthead. Then decide whether the publisher, as site owner, will write to the Archive. Most will not. Newspapers treat the Wayback copy as someone else’s problem, and in a narrow legal sense they are right.

The robots.txt myth, after 2022

For years, a Disallow in robots.txt could hide existing Wayback captures of that host. The mechanism was real. It was also a gift to anyone who wanted to erase a paper trail after the fact — including public authorities. The Archive changed the practice. The help text in force since the post-2022 revisions is unambiguous: robots.txt is not a deletion switch for captures that already exist.

You will still find SEO threads that tell a client to add User-agent: ia_archiver and wait. That is cargo-cult work. It may discourage future crawls. It will not purge 2014. If the captures are already in the collection, you write to info@archive.org as the site owner, or you do not write at all.

Two related myths travel with that one. A meta noindex tag on the live page will not reach back into the Archive. A Google removal does not “propagate” to Wayback. These are different institutions with different jobs. Confusing them is how a successful Search takedown still leaves a pink calendar of captures under the first result.

Google cache is not the Wayback Machine

The Google cache is a convenience copy the search engine keeps so a result can be shown when the host is slow. It is short-lived, tied to the index, and it usually dies after a real de-index or a hard 410. It is not a library.

The Wayback Machine is the opposite. It stores on purpose. It serves on purpose. It will recapture a page you have just restored. A cache clear and an Archive exclusion are not the same form, not the same email, and not the same standard. If the leftover is webcache.googleusercontent.com, you are in the Search chapter. If it is web.archive.org, Google’s outdated-content tool will do nothing.

Other public web archives exist — archive.today and its mirrors, Ghostarchive, various national libraries. Each has its own mailbox and almost no appetite for subject requests, except for image-abuse and court orders that name them. Name the capture. Name your standing. Accept a no.

Trove and the National Library, lightly

The National Library of Australia collects Australian websites under a legal-deposit and web-archiving mandate. Parts of that collection surface in Trove. The Australian Web Archive is the Library’s own playback. This is not San Francisco, and it is not a Google product.

The Library will restrict or remove material in defined cases — privacy, cultural sensitivity, a court order, a publisher’s own request — and it will not do so because a name search is inconvenient. The process is slower than an archive.org email and more formal. For some Australian news and government material it is the copy that actually matters.

Treat Trove as a separate ticket. Do not attach it to the archive.org email. Do not assume a Google de-index has any effect on an NLA identifier. If the publisher has already corrected or taken the piece down, say so, and send the Library to the news-archive outcome rather than to a speech about reputation.

A sequence that does not waste a week

  1. Identify every capture that is referring people — Wayback, archive.today, a Trove work, a Google cache. They are different jobs.
  2. Finish the source work. A live 200 with the old text will be recaptured the next time anyone clicks “Save Page Now”.
  3. If you control the host, write to info@archive.org as the site owner, with the domain, the paths, and proof. Ask for exclusion of existing captures and of future crawls.
  4. If you are only the subject, write only where you have a personal-data, copyright, or image-abuse hook. Otherwise take the request to the publisher and stop expecting San Francisco to edit Melbourne.
  5. File Trove or NLA only for the items they actually hold, with the identifier, not with a Google screenshot.
  6. Recheck in a fortnight, then again in ninety days. New captures appear when the live page comes back, or when a third party saves it.

Agencies that invoice for “archive removal” and then paste your URL into a general info@ mailbox are doing work you can do. What they cannot honestly sell is a subject’s right to be unpublished from a library that was built to remember.

If you want the desk to look at a general question about archives, use contact. Do not send the intimate image, the full capture, or a zip of every timestamp. We publish method. We do not run your case.