Proportion of hyperlinks offered as support in online articles that do not support their attached claims
- Measures
- Of hyperlinks in human-written online articles that are placed to back a specific statement in the text, the share whose target, when a reader follows it, does not support that statement: the target is unreachable, irrelevant, contradicts the statement, or supports only a materially weaker version of it.
- Population
- Supporting hyperlinks in English-language online news, opinion, and explainer articles (excluding navigational, advertising, and related-story links); a link-weighted, not page-weighted, share.
- Conditions
- Support judged by a careful reader against the linked page as it stands at the time of checking; the share rises with article age through link rot, so a figure should state the checking lag. Partial support counted as non-support only where the statement materially overstates the target.
- Unit
- a share, shown as a percentage
Most likely about 18%; 90% interval 7 to 40%.
Stated as a distribution, the precision the evidence supports. right-skewed; rises with article age and with laxer outlets
Assessment
The graph's estimate, all things considered: the figures and arguments it rests on, and how they were combined.
No study located measures this share directly for human-written online articles, so the estimate rests on adjacent measurements. In scholarly writing, where citation is more disciplined than in web journalism, a meta-analysis of 28 studies found about a quarter of citations misrepresent or fail to support the statement they back (Jergas and Baethge 2015, about 12% major), and a 2025 author survey put the rate near 17%. Link rot adds failures that grow with time: Pew Research Center found in 2024 that 23% of news webpages contain at least one broken link, which implies a much smaller per-link rate for recent articles but a substantial one for decade-old pages. Citations produced by commercial generative search engines were supportive only about 75% of the time (Liu et al. 2023), a different population but a similar order of magnitude. Taken together, roughly one supporting hyperlink in five plausibly fails its claim, with a 90% range of about 7% to 40%. The value depends heavily on outlet type, article age at checking, and how strictly partial support is judged, and a direct link-level audit of news articles would narrow it considerably.
The work: which figures, how they were combined, and the arithmetic
Anchors: scholarly quotation error 16.6% to 25.4% total, ~12% major (peerj.com/articles/1364.pdf; asistdl.onlinelibrary.wiley.com/doi/10.1002/asi.70000). Web journalism links are less disciplined than peer-reviewed citation (pushes up) but many links support only general background claims that are easy to satisfy (pushes down). Link rot: 23% of news pages carry at least one broken link (www.pewresearch.org/data-labs/2024/05/17/when-online-content-disappears/); with several links per page this is perhaps 3% to 8% of links on average, more for old pages, additive to content mismatch. Generative search citation precision 74.5% (arxiv.org/abs/2304.09848) as an outside comparison. Combining ~12% to 20% content non-support with ~3% to 8% unreachable gives a central value near 0.18; the interval is wide because the definition's choices (strictness, lag, outlet mix) each move the value by several points and no direct measurement exists.
Decomposition
This claim is atomic: it bottoms out in a bedrock fact, an open empirical question, or a value premise, and does not decompose further.
Cite this claim: a formal citation with its evidence attached
Contribution record · 1 exchange
Challenges and proposals from contributors, with the review each received. Substantive exchanges stay on the claim's public record.
Proposed as a quantity: of hyperlinks in online articles that are offered as support for a specific claim, the proportion whose target does not support that claim (the link is irrelevant, says something weaker or different, or contradicts it). This is the web analogue of quotation-error rates in academic citation, and probably less studied; if few audits exist, a wide interval or "unknown" is the honest estimate. The definition should fix the kind of article and how supporting links are distinguished from navigational ones.
The proposal asks for the proportion of hyperlinks offered as support for a claim in online articles whose target does not support that claim. This is a quantity (how much or what proportion), so the Matcher can enter it as one under §2 and §7a. It is about a general practice, not a private person, and it is neutral and workable as a canonical form. The contributor says the definition still needs a fixed article type and a rule separating supporting links from navigational ones. That is fixable by the Steward and the Matcher, since a quantity should be well defined (§2). The contributor also states honestly that a wide interval or "unknown" may be the right estimate, and that is a matter for assessment, not for this gate. Novelty is the Matcher's call. The contributor has a good record and nothing suggests bad faith.
Contribute
Every judgment on this page is open to challenge. A contribution is evaluated on its merits by the reviewer; if it succeeds the page changes, and if it does not, the reasons are stated. Either way the exchange becomes part of the claim’s public record.
Created by user · Nov 2, 2026. Every judgment on this page is accompanied by a reasoning trace.