There is a phrase that has been repeated ever since we began leaving ever larger parts of our lives online: the Internet never forgets.
It is normally used as a warning. Be careful what you publish. A photograph, an opinion or an unfortunate comment can follow you forever.
It sounds true. But it isn't.
The Internet forgets things every day. Servers are switched off. Companies disappear. Domains expire. Platforms close. Links stop working. Images remain hosted on services that no longer exist. Videos are removed. Disks fail. Backups we were certain we had made turn out to be surprisingly imaginary.
I discovered this in the least theoretical way possible.
Having been the founder and CEO of BySide, it was natural for my personal blog to be hosted on the company's servers. BySide and I grew together over many years, and completely separating my personal infrastructure from the company's never seemed particularly urgent.
After BySide was acquired by CoreMedia, my blog disappeared.
Deleted by mistake. Along with the backups.
It wasn't just a WordPress installation, a database and a collection of files that disappeared. Texts written over many years vanished. Opinions, stories, images, videos and moments I had chosen to preserve.
A part of me had been erased.
For a while, I accepted that part of me was lost. The blog had already been practically abandoned, and my writing had moved to BySide's weekly newsletter, published every Monday at nine in the morning. I wrote every week for years. Never just for myself.
Only after the company was sold, when that routine ended, did I return to those old memories. And I decided to try to recover them.
The idea seemed relatively simple: find the pages saved by the Wayback Machine, recover the texts and republish them on a new site, without WordPress and without the fragility of the old infrastructure.
The word “recover” made it all sound like a mechanical process. Like restoring a backup.
It wasn't.
I wrote a script to query the Internet Archive index, list the saved addresses from my domain and download everything it could find. The first run returned 127 records that appeared to correspond to texts.
It was a large enough number to feel complete. And numbers have this strange power: they give us a sense of rigour, even when they merely measure what we knew how to look for.
The 127 records corresponded to pages that followed the URL structure used by the old WordPress site — precisely the pattern I had told the code to search for.
It seemed reasonable to assume that was everything.
Until I found a capture of the homepage from 26 August 2024 showing content that wasn't on that list.
The 127 were not the archive. They were simply the 127 my code had known how to find.
There is an important difference between looking for what we know exists and discovering what we do not yet know how to look for. It is a difference that applies to technology, science, business and, apparently, the recovery of old blogs.
I went back and rewrote almost the entire process.
Instead of starting from a URL pattern, I began with the pages themselves: homepages saved in different years, monthly archives, categories, tags, pagination, feeds and internal links. Every page I found could contain the address of another I did not yet know existed.
I also had to teach the code something that was obvious to me but not to it: that http, https, www, ports, trailing slashes and redirects could all represent the same page dressed in different ways.
And that was when recovery began to feel less like computing and more like archaeology.
In the Wayback Machine, a page may be complete while its image has disappeared. Another keeps the image but has lost the text. A video survives as an empty rectangle. An article retains its title and first few paragraphs, ending at precisely the point where the old WordPress site said “continue reading”.
Sometimes we have the door, but not the house. At other times, we have a photograph of the house but can no longer get inside.
The Wayback Machine is not a perfect reproduction of the Internet. It is an extraordinary collection of fragments. And it is precisely because it is imperfect that its importance becomes clear.
Today, we produce more information than at any other point in history and, paradoxically, we may be creating some of the most fragile memories ever.
A photo album had to be lost, soaked or burnt. A digital photograph can disappear because someone stopped paying an account, forgot a password or saved it in a format no one can open any more.
The Internet Archive and the Wayback Machine therefore do something that seems increasingly essential to me: they preserve context.
They do not just preserve words. When they can, they preserve the page those words appeared on, the images that accompanied them, the links they made, what the Internet looked like at that moment and even the mistakes we made.
That 2024 capture did not contain everything. But it contained enough to say: “look again”.
Two of the articles I was looking for also demonstrated the limits of any archive. For both, the Wayback Machine had preserved references, titles and excerpts, but never the full text.
I tried every capture, every variation of the address and every page that referred to them.
There was nothing else.
I gave them up for lost.
I found them on my Mac, in iA Writer, the Markdown editor I have used for more than ten years.
They were simply where they had been written. They had survived several computers and several migrations without my ever thinking about it, because I had continued to carry the same files with me and had never changed editor.
Digital archaeology had failed. An old habit had solved the problem.
One was “Man and machine: Creativity and creative AI”. The other, “All flowers in time bend towards the sun”.
When I began using Markdown, it was a niche choice and I had to explain several times why I wrote in text files with asterisks. Today it has become almost a lingua franca of digital writing.
It did not survive because it was modern. It survived because it was simple — which is often the same reason things last.
The Internet's memory turned out to be distributed.
One part was in the Wayback Machine. Another was on one of my disks. Other parts were in images, videos, archive and aggregation pages, and links created by people who probably never imagined they were taking part in a preservation effort.
No single source had everything. Together, they had almost everything.
That “almost” contains very specific things.
I permanently lost the image of the business cards I ordered online for an article called “Personal Branding 2.0” — a title that dates the text better than any timestamp.
The article survived. The image did not.
I also discovered that several links and videos I had published had simply ceased to exist — with no connection at all to the deletion of the blog. They disappeared on their own, at the normal pace at which the Internet forgets.
Recovering an archive does not make it safe. It merely returns it to the same process.
In the end, the historical archive contains 127 texts.
The number is the same as when I started. The arithmetic isn't.
The first 127 were recovered files, but they represented only 125 distinct articles: two captures were duplicate versions of the same content.
The broader search revealed the two missing articles from 2019. When I found them on my Mac, the total returned to 127 — this time, 127 different texts.
Had I stopped after the first search, I would have arrived at exactly the same number and ended up with a different archive, without ever realising it.
Then came the strange part: reading what I had written fifteen or eighteen years ago.
I found opinions about Google, Twitter, the iPhone, Microsoft, marketing, innovation and entrepreneurship. Some have aged well. Others are, let's say, interesting when seen from a distance.
I also found texts that had nothing to do with technology. Music. People. Moments. My cat Clara. A song by Jeff Buckley and Elizabeth Fraser. Things I remembered and others I no longer knew I had written.
The first temptation is to correct them.
Not just the English or a typo, but to correct the past. Refine a prediction. Remove a naïve sentence. Make the person who wrote the text seem a little more intelligent than he really was.
But an archive is not a CV.
The value of those texts lies precisely in the fact that they were written by the person I was at that moment. I may disagree today. I may smile. I may even cringe a little at my former self.
But I should not rewrite that person simply to make him more like the person I am now.
Memory only has value if it is not retrospectively improved.
The images and videos made this even clearer. A text without the photograph that accompanied it is not always the same text. A cushion, a cat, a comic strip or a seemingly irrelevant video may be the key that gives meaning back to a page.
Recovering those elements was not decoration. It was recovering part of the memory.
The site that exists now no longer aspires to be a monument to the recovery of an old WordPress site.
It is a personal website again. A place where I can bring together old and new texts, technology, music, culture, work and a few additional ramblings.
The archive is there, but it does not need to dominate the house.
Preserving is not embalming, and an archive no one ever opens again is merely a better-organised box.
There is a beautiful irony in the fact that one of the hardest texts to recover was “All flowers in time bend towards the sun”.
It is an article about a song that was never officially released but continued to circulate online. The song survived because someone copied it, someone shared it and other people continued to listen to it.
The text about that song survived in much the same way: an excerpt in an archive, a reference on another page and a copy in an editor I never stopped using.
None of this happened because the Internet never forgets.
It happened because people, machines and institutions made the effort not to let everything disappear.
The Wayback Machine does not prove that the Internet has an infinite memory. It proves that we can resist forgetting.
And perhaps that is the true importance of an archive: not to allow us to return to the past, but to prevent the past from disappearing completely.
Now that the blog is alive again, I am left with the other part.
To keep writing.