Here's a small experiment. Go find an article you bookmarked five years ago. Not something from a major newspaper — something from a mid-sized blog, a forum thread, a company's resource page, a government subsite that got reorganized. Click it. There's a reasonable chance you're about to meet a 404 error, a domain parking page with ads for unrelated products, or — worst of all — a completely different piece of content that now lives at that URL with no acknowledgment that anything changed.
This is link rot. It is boring, unglamorous, and relentless. It is also eating the web alive.
What Link Rot Actually Is (And How Bad It's Gotten)
Link rot — sometimes called URL decay — is the process by which hyperlinks stop pointing to their original destinations. It happens for a hundred mundane reasons: a company restructures its website and doesn't set up redirects, a startup shuts down and lets its domain lapse, a platform migrates its content to a new CMS and breaks all the old paths, a social media site redesigns its URL structure. Each event is small. The cumulative effect is enormous.
Researchers at Harvard's Berkman Klein Center found that roughly 50 percent of URLs cited in Supreme Court opinions are no longer accessible. Let that one settle. Half the links in decisions from the highest court in the United States — documents that are supposed to be the bedrock of legal precedent — lead to dead ends. A separate study from the Pew Research Center found that about 38 percent of web pages that existed in 2013 are simply gone a decade later.
The Internet Archive, which runs the Wayback Machine and is essentially a one-organization rescue operation for digital history, estimates it crawls hundreds of billions of pages. It's still not enough. Not even close. The web produces content faster than any archive can capture it, and it deletes that content even faster.
Why Nobody's Fixing This
The frustrating thing about link rot isn't that it's technically unsolvable. Permanent URLs are not some exotic engineering challenge. Content-addressed storage — where a file is identified by what it is rather than where it lives — exists and works. The InterPlanetary File System (IPFS) is built on this idea. Stable redirect infrastructure is cheap and well-understood. The tools to make links durable are available.
The reason link rot persists is that fixing it costs money and earns nothing. There is no revenue model for link maintenance. When a company redesigns its website, setting up comprehensive 301 redirects for every old URL is a line item that gets cut in the third round of budget meetings. When a startup runs out of money, nobody's last act is carefully archiving the content and setting up forwarding addresses. When Twitter became X and shuffled its URL structure, hundreds of millions of embedded links across the entire web broke simultaneously, and the business case for fixing them was approximately zero.
The incentives, in other words, are perfectly structured to produce exactly the problem we have.
The Researchers and Archivists Getting Left Behind
For casual browsing, a dead link is an annoyance. For researchers, journalists, and historians, it's a professional catastrophe.
Academic papers increasingly cite web sources. News articles link to primary documents, government data, and source material. Wikipedia — which functions as the web's informal reference desk for hundreds of millions of people — has editors who spend significant time hunting for archived versions of dead citations through the Wayback Machine, a process that works until it doesn't. Brewster Kahle, who founded the Internet Archive, has described the organization's mission as building "the library of Alexandria version two, except this time we're not going to let it burn." It's a noble goal. It's also a goal being pursued by a nonprofit with limited resources against a process of decay that is structural, distributed, and essentially automatic.
The situation gets thornier when corporate consolidation enters the picture. When Gawker Media was destroyed by the Hulk Hogan lawsuit in 2016, a significant chunk of early internet cultural commentary went offline. When Tumblr purged content in 2018 and again through various ownership changes, entire creative communities lost years of work. When Google shuttered Google+ — a platform nobody much liked but plenty of people used — the links, the posts, and the discussions vanished. When any of the dozens of platforms acquired and quietly killed by larger companies over the last decade went dark, their content went with them, and every link ever written to that content became a door to a wall.
The Domain Squatter Problem Is Its Own Special Nightmare
Dead links don't always just 404. Sometimes they do something worse: they get replaced.
Expired domains are purchased almost immediately by squatters who fill them with ad-laden placeholder pages, sometimes mimicking the original site's topic area closely enough to fool inattentive users. A link to a defunct cybersecurity blog might now lead to a page full of affiliate links for VPN services. A link to an old recipe site might redirect to a domain stuffed with weight loss supplement ads. The URL is technically alive. The content it implies is gone. The replacement is actively misleading.
This creates a security dimension that doesn't get enough attention. Trusted old links — the kind embedded in long-standing Wikipedia articles, academic papers, and news archives — can be hijacked by purchasing the expired domain and pointing it at malicious content. Researchers have documented cases where expired domains cited in security advisories were purchased and turned into phishing pages. The link looks legitimate. The destination is a trap.
What a More Honest Web Would Look Like
A web that took link permanence seriously would look different from what we have. It might use content hashes rather than location-based URLs for long-term references. It might require platforms above a certain size to maintain redirect infrastructure for a defined period after restructuring. It might have funded, distributed archiving built into the infrastructure rather than offloaded to one nonprofit in San Francisco that runs on donations.
None of these things are coming anytime soon, because the web was not built by a single entity with a coherent long-term plan. It was built by millions of entities with short-term incentives, and those incentives do not include maintaining the doors other people walked through years ago.
So the graveyard grows. Quietly, without headlines, without a single dramatic moment of failure — just millions of small disappearances, one broken link at a time, until the web's memory of itself becomes a map with more blank spaces than filled ones.
The 404 page, it turns out, is not an error. It's a feature.