Commentary & Analysis
Commentary & Analysis
ItaliaOnLine has pulled the plug on Digiland — the free web hosting service that gave thousands of Italians their first slice of cyberspace. The loss would be manageable. What isn't is discovering that the
Wayback Machine,
humanity's backup of the web, is being quietly dismantled from the inside,
one...
publisher restriction...
at a time.
This is a story about digital memory, the peculiar politics of web archives, and why the most dangerous edits to reality are the ones nobody can prove happened.
ItaliaOnLine — arguably Italy's biggest internet services provider — announced the decommissioning of the Libero community, Digilander included. For many Italians of a certain vintage, this is like losing a photo album. For me, it was the discovery of a much larger problem: the entity we trusted to preserve the web is itself being slowly eroded.
Around the year 2000, I put five or six pages on Digilander. A quarter century later, the content barely mattered — what mattered was that those pages existed at a static, findable URL. Not buried in someone's feed. Not dependent on an algorithm. Just there, like a house with an address.
Status of digilander.libero.it → Decommissioned Wayback Machine copies → Restricted
Archive.org's Wayback Machine has crawled and stored billions of web pages since 1996. For most of its existence, it was the closest thing the internet had to a public library. Then publishers noticed it existed.
A non-profit effort to provide "universal access to all knowledge." Crawl the web. Store every version. Let anyone retrieve any page, any snapshot, any edit — no questions asked. The size of the project was staggering: more gigabytes of stored text than a single day of YouTube uploads, multiplied across decades.
Publishers began blocking archive.org crawlers and requesting removal of stored snapshots. Their concern: a reader might access an archived article instead of the original publisher's site — costing advertising revenue. The archive complied. Entire domains, including major news outlets, began disappearing from the historical record.
During the Senate confirmation hearings for a Supreme Court justice, the nominee used the phrase "sexual preference" — an expression that critics immediately called offensive. The nominee appeared genuinely unaware of any such connotation. So was I, frankly. I checked an online dictionary. Nothing in the entry flagged the term as problematic.
Within hours — perhaps less — I checked again. The same dictionary entry now carried a label: offensive. The definition had been silently updated. No notice. No changelog. No acknowledgment that the entry had just been different.
This is the textbook definition of gaslighting applied to an entire society: alter reality, then deny it was ever different. And the only tool capable of proving the change — the Wayback Machine snapshot of that dictionary page — was exactly the kind of resource that publisher restrictions are quietly eliminating.
"I wasn't imagining things. The dictionary had changed. But without the archive, nobody could prove it — and that's precisely the point."
— Author's note| Resource type | Historical snapshots | Current access | Status |
|---|---|---|---|
| Personal web hosting (Digilander) | Crawled 2000–2024 | Host decommissioned | Lost |
| Major news outlets | Crawled until ~2023 | Publisher-restricted | Restricted |
| Online dictionaries | Intermittently crawled | Partially available | Partial |
| Government & academic pages | Consistently crawled | Generally available | Available |
| Personal blogs & forums | Crawled inconsistently | Often available | Available |
While we're here, let's settle a terminology dispute that has been quietly driving me mad. The word misinformation has become ubiquitous in policy and media circles, deployed as a catch-all for anything someone finds inconvenient. The problem is that it doesn't actually mean anything precise.
This distinction matters in the archive debate because the argument for restricting web archives is often framed as preventing "misinformation." But an unchanged snapshot of what a dictionary actually said on a given date is not misinformation. It's evidence. The entity choosing to suppress it is the one with something to hide.
Every government commission, every think-tank report, every op-ed in 2025 is about regulating artificial intelligence. Meanwhile, the far simpler — and more urgent — problem of how to regulate ordinary web publishing and archiving remains completely unsolved.
Publishers have real interests. A reader accessing a five-year-old snapshot of an article rather than the current version might deprive the publisher of ad revenue, or — more legitimately — consume a version that has since been corrected. These are reasonable concerns that deserve a legal framework.
The purpose of a web archive is not to compete with publishers. It's to show what existed at a moment in time — to make visible edits, corrections, reversals. For news in particular, the ability to prove that a headline changed is not a curiosity. It is a democratic necessity. If I were Prime Minister, I'd have sorted this out before touching AI legislation.
I was lucky. When I learned Digilander was shutting down and found the Wayback Machine copies were restricted, I moved quickly: downloaded my own content and republished it on a more modern platform — which is probably where you're reading this now.
I lost the timestamp. There's no longer any easily verifiable proof that these pages existed 25 years ago rather than last Tuesday. That's a small personal loss. Other people will lose more: those without local backups, those who published on platforms now gone, those who assumed the archive was eternal.
The web was supposed to be different from print — not because it was faster or cheaper, but because nothing was supposed to disappear. That premise is looking shakier by the year.
Digilander's closure is a footnote. The quiet erosion of the Wayback Machine is a chapter. Together they point to the same uncomfortable truth: digital memory is far more fragile than we were told. Publishers restrict archives to protect revenue. Hosts decommission to cut costs. Algorithms bury static pages. And the result is an internet that rewrites itself without leaving evidence.
The fix isn't technically complex — it's politically difficult. A legal framework that protects archival access to news for verification purposes, while giving publishers control over commercial reuse, is entirely achievable. It just requires lawmakers to care about history as much as they care about the next news cycle. Don't hold your breath.
In the meantime: keep local copies of anything you care about. The internet forgets more than it admits.