A saved link depends on someone else keeping that page online and unchanged. To make your reading archive more resilient, keep each link's context with it (title, author, date, address, and your reason), check whether an archived copy exists, and accept one limit: no copy, printed or not, preserves everything.
This guide is part of A guide to saving articles and reading them later, which organizes every guide on the topic.
How much disappears: what studies measure
There are two different problems. Sometimes a link stops working: the page or the whole site no longer exists. Other times the address still loads, but the content has changed and is no longer what you read.
For the first, a 2024 Pew Research Center analysis took almost a million pages from the Common Crawl archive, captured between 2013 and 2023, and checked whether they were still online. As of October 2023, 25% were no longer accessible; among pages from 2013 the share was 38%, and among pages from 2023 it was 8%. The same study found at least one broken link on 23% of the news pages and 21% of the government pages it examined, and in the references section of 54% of English-language Wikipedia pages.
Pew counted only pages that no longer exist, not pages that changed, and its news and government samples come from U.S. sites. Treat it as an order of magnitude, not a forecast for your own list.
For the second problem, a 2016 study by Jones and colleagues in PLOS ONE examined web references in science, technology, and medicine articles published between 1997 and 2012. Among the references the authors could compare with a snapshot from the time, the live content had drifted from what was originally cited in more than 75% of cases. That is evidence about scholarly publishing; it does not measure what happens to your saved reading.
What to keep with each link
This is an original worksheet proposed by KeepAnalog. You can fill it in a note, a spreadsheet, or a notebook:
- Exact title and author or outlet. If the address changes, these details still let you search for the text.
- Publication date and the date you read it. If there is no date, write "undated".
- Original address. Copy it in full, without link shorteners.
- Archived copy, if one exists. Include the capture date.
- Your reason, in your own words. One or two sentences about why you saved it and which idea you want to remember. If you copy a sentence from the text, keep it short and in quotation marks.
The fifth item is the one no outside archive can keep for you. If the text disappears, your note keeps why it mattered; if it changes, your note helps you notice the difference.
How to use an archived copy without overtrusting it
The Internet Archive's Wayback Machine help page explains that you can link to archived pages and that the "Save Page Now" feature saves a specific page one time: it does not add the page to future crawls or save the rest of the site.
The same page spells out the limits. Some pages are missing because they were password protected, blocked by robots.txt, or never found by the crawlers; site owners can request exclusion; and elements that depend on JavaScript are often archived poorly. Before you rely on a capture, open it and confirm that it shows the text you read, not an empty front page or a later version.
Keeping context is not copying everything
Recording a reading's details and your own ideas does not require reproducing the full text. If you want to keep a complete copy, check the site's terms; this guide is not legal advice and does not authorize redistributing what you save. A link to a public archived copy is usually a more careful reference than passing the whole file to other people.
What if you print it?
A printed page preserves what was printed that day. It does not update if the original changes, it does not pick up later corrections, and it does not replace the reference to the source: without title, author, date, and address, it becomes hard to tell where it came from.
KeepAnalog turns public links chosen by one person into a printed edition for Argentina, and each link must resolve to an HTML page or a PDF to be included. A link that no longer works cannot be added. That is one more reason to record the context while the source is still available.
Next time you save something, take a minute to fill in the worksheet. If the link fails one day, you will know where to start looking.
For now, KeepAnalog's pilot is in Argentina only, with no delivery in the United States.