Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →A README link that fails today may have once led to a real page—or it may never have worked. A current error alone cannot tell you which. There is no README-specific study in the available evidence showing that half of dead README links were never alive, so treat that figure as a hypothesis, not a measured fact.
What a dead README link can—and cannot—tell you
Two different problems can look identical when you click a link and get an error:
As an Amazon Associate I earn from qualifying purchases.
- Link rot: the destination existed but later moved, changed, or disappeared.
- A URL with no evidence it ever worked: the address may have been mistyped, invented, or pointed at a page that was never available.
A failed request today does not distinguish between them. A 404 is a clue about the current response, not proof of the URL’s history. An archived copy can show that a page existed, but finding no archive is not proof that it never did: archives do not capture every URL. A 2026 study of model-generated citation URLs used Wayback Machine snapshots to distinguish non-resolving “stale” URLs from “hallucinated” ones, while cautioning that archive coverage makes the distinction imperfect. Read the study.
Does the evidence support “half” of README links?
No README-specific prevalence study in the reviewed evidence establishes that half of dead README URLs were never alive. Studies of other kinds of links show that dead links are a real maintenance problem, but their populations are not interchangeable with README links.
| Study population | Reported finding | What it does—and does not—show |
|---|---|---|
| Source-code comments; 9,654,702 links from 25,925 Git repositories, studied in 2019 | Almost 10% of links were dead. | This concerns links in comments, not README files, and does not establish how many dead links had never worked. 2019 study. |
| Commit messages; 18,201,165 links in 23,110 GitHub repositories, studied in 2023 | 14% of links described as prone to evolve became unavailable over time. | This is a finding about a specific subset of commit-message links, not all README links. 2023 study. |
| Model-generated citation URLs, evaluated in a 2026 study | Across evaluated systems, 3–13% were classified as hallucinated and 5–18% were non-resolving. | These figures concern selected model and research-agent outputs, not repository READMEs. The hallucination classification relied on archive evidence and the study’s method. 2026 study. |
These results support a narrower conclusion: links can go dead, and some generated URLs may lack evidence of ever having resolved. They do not tell us the share of README failures that fall into either category.
How to investigate a broken README link
- Record what failed. Keep the URL, the response or browser error, and the date you checked it. A timeout, access-denied response, or blocked automated request is less conclusive than a repeatable not-found response.
- Check for a redirect or replacement. Visit the site’s current documentation, search for the page title or product name, and inspect any redirect destination. A site redesign or version change may have moved the material.
- Look for historical evidence. Search an archival service for the exact URL and, where useful, its domain or page title. A snapshot is evidence that the address once resolved; no snapshot does not settle the question.
- Inspect the repository context. Review the README’s Git history and nearby text, code, or issue references to infer the intended destination. A typo or obsolete version path may become apparent from context.
- Repair only when the destination is clear. Update the link to a verified successor, replace it with a relevant current source, or remove it if it no longer serves the reader. If you cannot establish the intended target, do not invent a replacement.
How to check README links automatically
A Markdown link checker can find current response problems and redirects across README or documentation files. The lychee repository documents Markdown scanning, dead, redirected, and slow-link classifications, plus CI and pre-commit examples. It is one example of the category, not a guarantee that every link can be checked reliably.
Rank #2
Use automated results as a triage list, not as a verdict about a URL’s past. Some sites block automated requests, respond slowly, or behave differently depending on authentication or rate limits. Verify ambiguous failures manually before changing the README, and retain enough detail to reproduce the check.
Free tools Windows power users keep installed
One-click scans. No signup required.
How to reduce future link rot
- Prefer stable, canonical documentation pages over temporary previews, session-specific URLs, or links to frequently changing search results.
- For material within the same repository, use relative links where appropriate. GitHub resolves relative links against the repository and adjusts them for the branch being viewed; this helps internal navigation but cannot keep an external site online. See GitHub’s Markdown guidance.
- Run a link check when documentation changes or in CI on a schedule that fits the project. Review redirects and inaccessible results rather than automatically treating every non-success response as a broken link.
- For essential external references, consider recording a page title and enough context that a maintainer can locate a successor if the URL changes.
GitHub’s README guidance says, “Longer documentation is best suited for wikis.” Moving substantial material to a wiki can organize documentation, but it does not itself prevent links from becoming stale.
Quick Recap
Best Value
Rank #3
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
