Click an old link and you already know the two ways it can betray you. Either the page is simply gone, a blunt error where an argument or a source used to be, or worse, the address still resolves but now points at something else entirely, a parked domain or an unrelated article wearing the same address. Both failures are quiet. Nothing announces that the thing you were promised has vanished. The link just sits there looking healthy until the moment you actually need it.
We treat the web as a place where things stay put. In truth a link is a promise made by a stranger to keep a specific page at a specific address indefinitely, and strangers move on. Sites get redesigned, companies fold, servers get switched off, and every reference pointing at them turns into a dead end. When those references are footnotes in a court ruling or citations in a news story, the decay does not just break a page. It chips away at the public record.
Two ways a link dies
The obvious death is the broken link, the familiar not found error. That one at least tells the truth. You know something is missing, and you can go looking for it elsewhere. Annoying, but honest.
The quieter death is content drift, and it is more dangerous precisely because it hides. The link still works. It returns a real page. But the page no longer says what it said when someone cited it. A news article that once documented a claim now redirects to a section front. A government statistic sits at a new address while the old one shows a friendly redesign with none of the original numbers. The citation looks intact and points at the wrong thing, which is a harder problem to catch than an honest blank screen. A reader who trusts the footnote never learns that the ground shifted, and a reader who checks it walks away confused about who got the facts wrong.
How much of the record has already gone
The scale is not a guess. In 2021 researchers John Bowers, Clare Stanton, and Jonathan Zittrain examined the outbound links in New York Times articles from the launch of the paper's website in 1996 through the middle of 2019. Across more than two million links, they found that a quarter of all the deep links pointing to specific pages were completely dead, and the older the article, the worse the decay, with the earliest years losing most of their links entirely.
Courts fare no better. Studies of United States Supreme Court opinions have found that roughly half of the web addresses cited in the Justices' own rulings no longer lead to the material the Court was pointing at. Sit with what that means. A judge cites a source to support a decision that governs everyone, and within a few years the citation leads nowhere. The reasoning is still binding. The evidence behind it has quietly slipped out of reach.
Why this is a commons problem
It is tempting to file link rot under personal annoyance, the digital version of a misplaced receipt. It is bigger than that. A shared record is what lets people check each other's claims. A footnote is a standing invitation to go verify for yourself. When the footnote leads to a dead page, the invitation is withdrawn, and we are left trusting a summary because the source can no longer be inspected.
That erosion falls unevenly. A well funded institution can pay to preserve its own archives. A small newsroom, a volunteer group, or a public agency on a thin budget usually cannot, so its corner of the record decays first. The material that vanishes is often exactly the local, specific, hard to reconstruct information that no one else bothered to save. What survives is what someone had the money and the motive to keep. Over enough years, that quietly rewrites what the past looks like, because the surviving sources are the ones that happened to have a caretaker, not the ones that mattered most.
What actually holds the record open
The better news is that people saw this coming and built defenses. The Internet Archive has spent decades capturing snapshots of the web, and its Wayback Machine can often show you what a page looked like on the day it was cited, long after the live version has changed or disappeared. Legal and academic tools that freeze a page at the moment of citation do the same job for footnotes, storing a permanent copy so the reference cannot rot out from under the argument.
These defenses are not perfect. An archive can miss a page, and a snapshot captures only what was public at that moment. But they shift the default from everything is temporary toward at least the important things can be kept. The habit worth adopting is to stop assuming a link will be there tomorrow and to save the version you are actually relying on today.
Keeping your own links alive
You do not need to run an archive to stop feeding the rot. When you cite something that matters, whether in a report, a community post, or a message people will act on, take a moment to save a snapshot of the page and note that copy alongside the original. If you publish anything, check your own old links now and then and repair the dead ones before a reader hits them. When you can, point at stable sources, a permanent record or an archived copy, rather than a fragile address that depends on one company staying in business. The public record is not maintained by anyone in particular. It survives because ordinary people, one careful citation at a time, decide it is worth keeping.