18 October 2026

📌 Small hours of Sunday in Gourock, Take Me Back to Eden in the headphones, the whole album on repeat since midnight. Tonight's thing was small and I liked it: every article on La Bottega now has a button that saves the page to the Internet Archive. One click, and a copy of the page sits in the Wayback Machine, outside our hands and outside the hands of anyone who might want it gone.

The idea

A collective blog that has been online since 2001 has, by now, a fair number of pieces that someone somewhere would rather not exist. The usual way to protect them is "we have backups", which is true and also beside the point - a backup proves nothing to anyone but us. A copy in the Wayback Machine is a different thing: it is timestamped, it is public, and it is kept by a library that does not answer to us.

Internet Archive already lets anyone do this by hand, through a "Save Page Now" form. The idea was simply to put that one step at the bottom of every article, next to the share button, so a reader who thinks "this one should not disappear" can act on it without leaving the page.

Three ways to do it

There were three options, and I went through them in order.

The first is a plain link to https://web.archive.org/save/<url>. Thirty lines of PHP, no credentials, the reader's browser does the work. Honest and boring.

The second is a button that calls an endpoint on our own site, and the site talks to the archive on the reader's behalf using the collective's own archive.org account. Nicer for the reader (they get the link to the copy right there), and it lets the site keep a record of what was saved and when.

The third is to archive every article automatically when it is published, and leave the button as a "see the copy" link.

I picked the second. Partly because the reader's click is the point - it is their act of care, not an automated one - and partly because the third can be bolted on later with the same code.

How it works

The site now has a mu-plugin, bottega-archivio.php, in the same public repo as the others (bottega-mu-plugins, GPLv3). It adds a box under each article - "Conserva questo articolo" - with a button and a link to the previous copies.

The button talks to a REST endpoint on the site. The endpoint accepts only the ID of a published post, works out the URL itself, and asks the archive's Save Page Now 2 API for a capture. The API answers with a job id; the page then polls the site every five seconds, and the site polls the archive, until the capture comes back as done or failed. When it is done, the reader sees "Salvato: copia del ..." with the link.

The account keys live on the server, in the same place as the other secrets, and never reach the browser. The endpoint cannot be used to archive other people's pages, because it never takes a URL from outside. And it is throttled: at most five requests a minute in total (the archive allows seven per authenticated account), and one copy per article per day - a second click just returns the copy already made.

"It doesn't seem to save"

First test, on a real article. The button spun for 75 seconds, which was my chosen limit, then gave up with "still being processed, check back in a few minutes". The archive.org profile page showed nothing. My first thought was that the thing did not work.

The server log said otherwise: no errors, no refusals, nothing at all. So the job had been accepted and every status poll had simply come back as "pending". A minute later a look at the transients on the server showed the capture had completed at 03:06 UTC - about two minutes after the click. Save Page Now, at three in the morning, is just slower than my patience. The limit is now two minutes, and the message says plainly: the copy will be ready in a few minutes, press again to see the link.

Contro l'oblio

The button lives at the bottom of the article, where only people who read to the end will find it. So the sidebar got a small box too, between the RSS feed and the contact details, titled Contro l'oblio - against oblivion. Four lines: every article has a button, a copy stays on archive.org, out of reach of anyone who wants it gone, use it.

While I was there the contact box got a rewrite as well. "Write to us in the clear, a normal email is fine. For delicate tips or documents that should not travel, use the editorial PGP key, so what you send is read by us and no one else." It had been three words shorter and said less.

and then...

It is past four. The album has looped more times than I counted. What I like about this one is that it does almost nothing - it moves one existing step a few centimetres closer to the reader - and yet it changes what the blog promises. Until tonight we could say "we keep backups". Now a reader can say "I kept a copy", and point to it. That is a different sentence, and I think the better one.