The Vanishing Evidence Problem in Scientific Publishing
This paper demonstrates that ordinary web links cited in scientific articles suffer from predictable and continuous decay over time, with only two-thirds remaining accessible after a decade, thereby posing an ongoing threat to the reproducibility and integrity of the scholarly record that requires both future preservation standards and retrospective recovery efforts.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine the library of human knowledge as a massive, glittering city of books. For centuries, if you wanted to check a fact, you'd walk to the shelf, pull out a book, and read the page. But in the last few decades, scientists started building a new kind of library. Instead of just writing down what they found, they started writing down where to find the actual tools, data, and code they used. They added little notes in their books saying, "Hey, the magic sauce I used is at this specific address on the internet!"
This is the modern scientific record. It's like a treasure map where the "X" isn't just a spot on a page, but a link to a digital cave full of evidence. But here's the scary part: the internet is a bit like a city where buildings get demolished, streets get renamed, and shops close down every day. When a scientist writes a paper today, they might link to a website that looks perfect right now. But what happens in ten years? What happens in twenty? If that website vanishes, the "treasure map" leads to a dead end. The book is still on the shelf, but the proof inside it has disappeared. This is called "link rot," and it's the silent thief stealing the evidence from our scientific history.
The Great Digital Disappearing Act
A team of researchers decided to play detective with a specific group of these "treasure maps." They didn't look at brand-new books; they looked at a fixed collection of scientific papers published in 2015. Think of this group as a time capsule. The researchers had already checked these links back in 2017 and 2018, but they wanted to see what happened a full decade later. They treated these 2,530 links like a group of runners in a race, checking to see how many were still standing and how many had tripped and fallen.
The results were a bit like watching a slow-motion leak in a boat. When they checked the links in 2025, they found that the water was rising faster than anyone hoped. Only about two-thirds (roughly 69%) of the links were still working. That means nearly 30% of the links were completely broken, leading to nowhere. Even worse, the researchers found that this wasn't just a few random accidents; it was a steady, predictable slide. They used a mathematical model to predict the future, and it suggested that if nothing changes, half of all these links will be gone in about 18 years.
The paper makes it very clear that this isn't just a problem for old, dusty books. The researchers argue that even though scientists are getting better at saving their links in new papers, that doesn't fix the old ones. It's like building a new, fireproof library for today's books while the old library slowly burns down. The "new" improvements don't magically repair the "old" damage. The study shows that the decay is smooth and steady, with no sign of stopping or getting better on its own.
Why This Matters to You
You might think, "So what? If the link is broken, I just read the paper without the extra stuff." But that's like trying to understand a magic trick by only reading the description, without ever seeing the sleight of hand. Science relies on being able to double-check the work. If a scientist says, "I used this specific computer program to prove this," and that program's link is broken, no one can verify if the math was right. The evidence vanishes, and the trust in the science starts to crumble.
The authors of this paper aren't saying the whole system is broken beyond repair. They are saying that the current way of doing things—just hoping links stay alive—isn't enough. They suggest that we need to be much more aggressive about saving these links after they are published, not just before. We need to treat these digital links with the same care we treat physical books, making sure they are copied, stored, and protected so that the next generation of curious teenagers can still follow the treasure maps to the truth. The paper concludes that without a coordinated effort to fix these broken links, we are slowly losing the ability to prove our own discoveries.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.