Why articles disappear from the web: link rot explained
Updated
Articles disappear because nothing on the web is kept by default. Domains expire, sites are redesigned and their addresses change, publishers close, authors delete posts, and pages move behind paywalls. This is called link rot, and it is common: the Pew Research Center found that 38% of web pages that existed in 2013 could no longer be reached a decade later. You can't stop it, but you can keep your own copy of what matters, by archiving public pages with the Wayback Machine or saving the article text where you control it.
How common is link rot?
The best recent measure comes from the Pew Research Center's study When Online Content Disappears (May 2024). Pew took "a random sample of just under 1 million webpages from the archives of Common Crawl" and checked which ones still worked. Its main findings:
| What Pew checked | What it found |
|---|---|
| Pages that existed in 2013 | 38% no longer accessible |
| Pages that existed in 2023 | 8% no longer accessible |
| All pages that existed at some point from 2013 to 2023 | A quarter no longer accessible, as of October 2023 |
| Government web pages | 21% contained at least one broken link |
| News web pages | 23% contained at least one broken link |
| Wikipedia pages | 54% had at least one broken link in their References section |
| Tweets | 18% of tweets from its collection window were no longer publicly visible, just months after posting |
Older research found the same pattern in places where links are meant to last:
- Court opinions and law journals. A 2014 Harvard study found that "more than 70% of the URLs within the Harvard Law Review and other journals, and 50% of the URLs within United States Supreme Court opinions, do not link to the originally cited information" (Zittrain, Albert and Lessig). The paper introduced Perma.cc, a service that lets authors archive the pages they cite.
- News articles. A 2021 study of links in New York Times articles from 1996 to mid-2019 found "a near linear increase of linkrot over time", and that "over half of articles containing at least one URL also contained a dead link" (Harvard Berkman Klein Center).
These numbers are a floor, not a ceiling
Pew counted a page as gone "only if they returned one of nine error codes that definitively indicate that the page and/or its host server no longer exist or have become nonfunctional." A page that still loads but now shows something else would not count as gone. So the share of articles you can no longer read as you first read them is likely higher.
Link rot and content drift
There are two ways to lose an article:
- Link rot. The address stops working. You get an error, a "page not found" message, or a domain for sale.
- Content drift. The address still works, but the page has changed. The article was edited, replaced, merged into another page, or redirected to a home page. The Harvard study of the New York Times looked at both, because a link that loads is not proof that the article is still there.
Content drift is harder to notice, because nothing looks broken. A bookmark that opens a page about something else has failed just as surely as one that shows an error.
Why pages disappear
Most pages are not deleted on purpose to hide anything. They disappear because keeping old pages online costs time and money, and nobody is responsible for keeping them.
- Domains expire. When a site's owner stops paying for its domain, every page on it goes at once. Old blogs and small publications are especially exposed.
- Sites are redesigned or move platforms. A new content system often means new web addresses. Unless someone sets up redirects from every old address, older links break.
- Publishers close or merge. When a publication shuts down or is bought, its archive may be moved, trimmed, or switched off.
- Authors and platforms delete posts. People remove old posts, accounts are closed, and platforms change what is public. Pew's tweet figure shows how quickly this happens.
- Paywalls and logins. An article that was free to read can later require a subscription or an account. The page exists, but not for you.
- Pages are updated in place. News stories, product pages, and documentation are often rewritten under the same address.
What this means for your saved links
A bookmark, a browser reading list entry, or a read-later list that stores only links is a list of addresses. Every one of them depends on someone else keeping a page online. The longer the list has been growing, the more of it has rotted. Pocket users saw this first-hand: the 2025 Pocket export contained the links but, in Mozilla's words, "does not extract the text of saved links" (Mozilla), so every dead link in an export is an article that is gone. See what to do with your Pocket export if you have one.
What you can do about it
Check your bookmarks for dead sites
The free bookmark dead-link checker reads a bookmarks.html file in your browser and checks whether each bookmarked site still answers. For the ones that don't, it links to the Wayback Machine so you can look for an archived copy, and it gives you a clean file without them. It can't detect "page not found" pages on a site that still answers, so it catches dead domains, not every missing article.
Archive public pages with Save Page Now
The Internet Archive's Wayback Machine lets anyone save a public page. Paste the address into Save Page Now and it captures the page, "including the images and CSS", and gives you a permanent address for the copy (Internet Archive). It saves one page at a time, and some sites block it. The copy is public, and site owners can ask for their pages to be excluded, which the Internet Archive reviews case by case (Internet Archive).
Keep your own copy of the text
The most reliable copy is one you keep yourself. You can print an article to PDF, save the page to your computer, or clip it into a note app. How to save an article before the page disappears compares these methods and their trade-offs.
Save to a read-later app that stores the text
Some read-later apps keep a copy of each article you save, so your reading list doesn't rot with the web. Reading Rewind does this on every save: when you click the toolbar button, the page goes to the top of Read Later, and in the background the server extracts the article and stores its text. If the original page later changes or disappears, the text you saved is still there to read. Images are not stored; they load from the original site, so they can go missing when the site does.
Reading Rewind is a desktop read-later app for Chrome, with Firefox in beta, and it is free to use today.
Frequently asked questions
What is link rot?
Link rot is when a link stops leading to the page it was meant to point to, because the page or its whole site no longer exists. It affects bookmarks, citations, and reading lists alike.
How fast do web pages disappear?
Faster than most people expect. Pew found that 8% of pages that existed in 2023 were already inaccessible by October 2023, and 38% of pages from 2013 were gone a decade later.
Can I recover an article that has already disappeared?
Sometimes. Paste its address into the Wayback Machine to look for an archived copy, search for the title in case it moved, or try the site's own search. If nobody archived it, it may be gone for good.
Does a bookmark keep a copy of the page?
No. A browser bookmark stores the address and title only. If the page disappears, the bookmark leads nowhere.
Is the Wayback Machine enough on its own?
It is a good public backstop, but it isn't a personal library. Not every page has been captured, some sites block it, and archived pages can be excluded on request. For the articles you care about most, keep a copy of your own as well.