The Internet Forgets, and It's Not Random

Key Takeaways

• Research by the Internet Archive documents that large portions of web content from a decade ago are no longer accessible — and the loss is not evenly distributed.
• Platform closures, link rot, legal decisions, and commercial priorities determine what survives online; grassroots and community content is the most vulnerable.
• For communities whose histories were erased or suppressed in physical archives, digital loss continues rather than corrects that pattern.
• UNESCO's Recommendation on documentary heritage recognizes digital preservation as a matter of cultural rights — meaning what disappears online is a governance question, not just a technical one.

Today is the International Day for the Remembrance of the Slave Trade and its Abolition, observed every August 23rd to honor the uprising that began in Saint-Domingue — now Haiti — on the night of August 22, 1791. The day was established by UNESCO to ensure that the history of the transatlantic slave trade remains visible and remembered. 

That word "ensure" is doing a lot of work. Memory doesn't preserve itself. Someone has to decide what survives, where it lives, and who can find it. Online, those decisions are being made constantly — and the outcomes are not neutral.

Online Memory Isn’t as Permanent as It Looks

The internet created a widely held assumption that everything posted online exists forever. In practice, a significant portion of what was published a decade ago is already gone.

Websites close. Platforms change their policies or shut down entirely. Links break, content gets removed under copyright claims, and hosting fees go unpaid. The result is a web that is constantly losing pieces of itself.

Most of those losses are invisible. The people who would notice are often the ones whose content disappeared.

This isn't a fringe technical problem. According to the Internet Archive's Vanishing Culture project, documented in research published in April 2026, the gap between how much content is created online and how much is actually preserved has become a genuine crisis. 

An earlier phase of the same research, published in October 2024, examined the systems and incentives behind digital loss. What both found is that preservation is a choice, not a default — and the choices being made are not made equally.

What Determines What Survives

The content most likely to be preserved online shares certain characteristics. It's produced by institutions with resources, hosted on stable platforms, and covered by legal frameworks that create an incentive for someone to maintain it. Major news archives, government records, and commercially valuable media have the best survival rates. Community content, activist documentation, and archives from grassroots organizations have the worst.

In September 2024, the Second Circuit Court of Appeals upheld a ruling against the Internet Archive in Hachette Book Group v. Internet Archive, limiting the Archive's ability to lend digitized books through its digital lending program. The practical effect was that one of the few organizations that has made digital preservation its explicit mission was legally constrained in what it can preserve and share, while the publishers that sued it — already equipped to maintain their own archives — were protected.

How digital platforms systematically exclude content from marginalized communities plays out across multiple systems. The archiving gap is one of its least visible forms.

The Communities That Bear the Cost of Digital Loss

When platforms shut down or change their policies, the losses fall unevenly. Tumblr's 2018 content policy change erased a substantial archive of queer creative work and community history. Twitter's ownership changes hollowed out what was collectively called Black Twitter — a decade of cultural commentary, political organizing, mutual aid, and documentation of events that mainstream media covered poorly or not at all. Neither loss was random. Both reflected whose content the platforms' business models had prioritized and whose they hadn't.

For communities whose histories have been systematically erased in physical archives — enslaved people whose names weren't recorded, diaspora communities whose documentation was never collected, activists whose records weren't considered worth keeping — the digital era represented a chance to build something that the official record ignored. A platform closure, a link that rots, a legal decision that limits what can be archived: these aren't only technical events. They're continuations of a much older pattern.

UNESCO's Recommendation on the Preservation of and Access to Documentary Heritage frames documentary heritage — including in digital form — as part of cultural heritage that member states have an obligation to protect. The digital content created by communities, civil society, and marginalized groups falls within that framework. Treating its loss as a technical inevitability rather than a governance failure is the wrong response.

The Right to Remember Is a Digital Rights Issue

The observance today asks something specific of anyone who cares about how information is governed online. The slave trade wasn't forgotten accidentally. It was hidden, minimized, and sanitized through deliberate choices about what got recorded, where it was kept, and who had access to it. Memory has always been political.

The digital layer doesn't change that. It adds new mechanisms — platform decisions, copyright enforcement, hosting economics — through which some histories persist, and others disappear.

I think a free and open internet has to include the freedom to remember: to create records that survive, to access archives that haven't been selectively pruned, and to find the histories of communities that mainstream institutions have not prioritized preserving. An internet that technically allows you to publish anything but structurally ensures that certain content doesn't last is not a free internet. It's a selective one.

The observance of August 23rd is built on the recognition that forgetting is not neutral. Neither is digital forgetting. And neither is the silence about who bears the cost of it.