When you click on an outdated link and end up on a 404 page, you feel a certain kind of unease. These days, it happens all the time: a government form you bookmarked, a news article you intended to read again, or a tweet someone cited during a disagreement three years ago. The page has simply vanished. This sentiment has been measured by Pew researchers: almost 25% of all webpages created between 2013 and 2023 are no longer accessible, and for the oldest content, the figure rises above 33%. It’s easy to dismiss this as a small annoyance. Most likely it isn’t.
It doesn’t really matter if you call it digital decay or digital rot. The mechanism is what counts. Entire categories of human expression simply cease to exist in any retrievable form, servers are shut down, businesses are sold, and archives are quietly pruned to save storage costs. The fact that over half of Wikipedia’s reference links now lead to dead pages makes it an excellent gauge in this regard. A website built on citations is gradually turning into a website built on citations that have fallen short of expectations. That has a melancholy quality that is almost poetic.
Not only is the scale unsettling, but so is the speed. One in five of the pages that were gathered just two years ago are already disappearing. From an institutional perspective, that is the blink of an eye rather than centuries of erosion. In contrast, a printed newspaper or a clay tablet in a library basement might last a millennium without any upkeep. It turns out that rather than being a technical guarantee, digital permanence was always more of a marketing pledge.
Sociologist Dean Curran of the University of Calgary has been uncomfortably equating the current digital economy with the period leading up to the financial crisis of 2008. His argument is based on the idea of “tight coupling,” which he borrowed from sociologist Charles Perrow. This refers to systems that are so interconnected and devoid of redundancy that a single failure can lead to a much larger problem. The classic example is the CrowdStrike outage in 2024, when a defective software update caused airports to suddenly ground flights and television stations to go dark. It wasn’t a hack. It was simply exposed fragility.
It’s difficult to ignore how casually we’ve come to accept this vulnerability as a necessary expense of conducting business online. In an effort to expand as quickly as possible, businesses frequently buy out or destroy rivals before any obsolete, analog alternatives remain. That is definitely efficient. Additionally, it’s the exact type of fragile architecture that crumbles under sudden stress. In 2017, WannaCry and NotPetya caused billions of dollars’ worth of damage while barely altering the process of developing digital infrastructure. In Curran’s words, “the warning signs are sitting in plain sight,” which somehow makes the inaction seem stranger rather than safer.
From a different perspective, Peter Krapp, a professor of film and media at UC Irvine, makes the strangely eerie observation that the very technologies designed to preserve our digital past frequently hasten its obsolescence. In the past, CD-ROMs were advertised as being almost unbreakable. They are no longer even readable by the majority of contemporary laptops. According to Krapp’s research, in order to retrieve data from the 1950s, an insurance company in 1978 had to run seven layers of emulation software. This was similar to a digital archaeology dig, but the artifact was only thirty years old at the time.
Despite the marketing language suggesting otherwise, there’s a feeling that AI is making things worse rather than better. Corrections take longer to spread than hallucinogenic facts. Human moderators are unable to keep up with the scale of misinformation. Furthermore, rather than becoming less complex, the underlying infrastructure—servers, data centers, and fragile platform dependencies—keeps getting more complicated. It is genuinely unclear if this will be addressed in a meaningful way before a more significant systemic failure forces the problem. Regulation usually comes after the harm, never before.
It’s hard to get rid of the impression that we constructed a filing cabinet the size of a civilization and neglected to make sure the drawers locked.
