On September 1, the Internet Archive published a blog post with a title that reads like a distress signal: “Keep Our Servers Running.” The pitch is straightforward — start a recurring donation of $25 or more this month, and a matching donor will triple your first gift. The Archive, which holds 210 petabytes of books, websites, audio, and software, is powered by donations averaging about $25. The Hacker News thread has 959 points and 247 comments, most of them variations on “this is important, please donate.”
Here’s what almost nobody in that thread is saying: the fact that this campaign exists at all is the story. Not the generosity. The fragility.
One Nonprofit, One September, No Plan B
The Internet Archive is not a government agency. It is not a university consortium. It is not backed by a sovereign wealth fund or a tech billionaire’s foundation. It is a 501(c)(3) in San Francisco that has to run a matching campaign every September to keep its storage arrays spinning.
Think about what that means. The Wayback Machine — the closest thing humanity has to a public record of the web — depends on whether enough people sign up for $25 monthly gifts during a four-week window. If the campaign underperforms, the Archive doesn’t shut down overnight. But it does defer maintenance, delay ingest, slow crawls. The margin between “preserving the web” and “falling behind the web” is a fundraising target.
A systems librarian at a state university put it to me this way, standing in a server room while a tape backup whirred behind her: “Every time the Archive runs one of these campaigns, I think about what happens the year they miss. There’s no fallback. There’s no other copy of most of this.”
She’s right. The Archive’s redundancy is technical — multiple copies, multiple locations. Its institutional redundancy is zero.
The Lawsuits and the Hack Didn’t Change the Model
The Archive has already been through two near-death experiences this decade. In Hachette v. Internet Archive, a federal judge ruled that its controlled digital lending program was not fair use, and the Second Circuit affirmed. The Archive removed hundreds of thousands of books from its lending library.
Then, in October 2024, the Archive was hit by a DDoS attack and a data breach. The site went down for days.
Neither event changed the fundamental structure. The Archive is still a single nonprofit, still dependent on individual donations, still one bad year away from having to make cuts that would be invisible to the public until a link goes dead and a citation breaks.
The Hachette ruling was a warning about legal exposure. The hack was a warning about operational exposure. The September campaign is a warning about financial exposure. Three warnings, same lesson: this institution is load-bearing, and it is not built to bear the load.
Civilization’s Memory Should Not Be a Charity Case
Here’s the uncomfortable question the Hacker News thread doesn’t ask: why is the preservation of the public web a charity?
When the Library of Congress needs funding, it goes to Congress. When the National Archives needs a new facility, it gets a line item. When the Internet Archive needs to keep 210 petabytes online, it posts a blog entry and hopes for the best.
The conventional wisdom treats this as a heartwarming story about the generosity of ordinary people. And it is — the $25 average donation is genuinely moving. But it’s also a structural failure. We have decided, by default, that the historical record of the internet age should be maintained by a nonprofit that has to run a bake sale every September.
The Archive’s founder, Brewster Kahle, has been making this argument for decades. He’s right. But the argument has not produced an endowment, a government partnership, or a consortium of institutions willing to share the load. It has produced a matching campaign.
So donate. The 3x match is a good deal, and the Archive does irreplaceable work. But don’t mistake the campaign for a solution. It’s a stopgap. The real question is what happens when the stopgap stops working — and nobody has an answer.