Quantityempirical · derivedA factual claim that rests on inference from other evidence rather than direct observation.constitution →importance · minorImportance 0.30, from 0 to 1 · minor: narrow or largely settled, cheap to get right. Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →

TopicsFact-Checking Web Content Accuracy

Proportion of claims selected by professional fact-checkers that receive a false or mostly false rating

Measures: The share of claims that professional fact-checking organizations choose to check and give a verdict on whose verdict falls at the false end of the organization's scale: False, Pants on Fire, Mostly False, or the equivalent bottom one or two levels of a graded scale (e.g. "False", "Partly false" on binary-plus scales). Unrated or "unproven" items are excluded from the denominator.
Population: Rated fact-checks published by established professional fact-checkers (IFCN signatories and comparable outlets such as PolitiFact, Snopes, FactCheck.org, AFP Fact Check, Full Fact), pooled across organizations
Conditions: Recent period, roughly 2016 to 2025, the era dominated by viral social-media claims; the share differs by organization and was lower in the pre-2016 period of politician-focused checking
Unit: a share, shown as a percentage

45 to 85%

Between 45 and 85% (90% interval), most likely about 65%.

Stated as an interval, the precision the evidence supports. varies strongly by organization and era; PolitiFact post-2016 near the top, Snopes and pre-2016 politician-focused checking lower

last estimated Nov 2, 2026 · Claude Opus 5.5

Assessment

The graph's estimate, all things considered: the figures and arguments it rests on, and how they were combined.

Among claims that professional fact-checkers choose to rate, most land at the false end of the scale. The best estimate for the recent period (about 2016 to 2025), pooled across major outlets, is around 65%, with a 90% interval of 45% to 85%. The share depends heavily on the outlet and the period. PolitiFact's verdicts have shifted toward false since 2016, and by 2021 fewer than 10% of the claims it selected were rated correct. Snopes rates a larger share of claims True or Mostly True, about 29% against PolitiFact's 11%. Benchmark datasets built from ClaimReview fact-checks show the same skew, with about 58% False against about 19% True. The figure reflects how fact-checkers choose claims, since they concentrate on suspect, viral material. It says nothing about how accurate public claims are in general.

The work: which figures, how they were combined, and the arithmetic

Figures: (1) arXiv 2210.13865, crawl of 20,274 PolitiFact checks 2007-2021: under 10% correct in 2021, so the false side (Mostly False/False/Pants on Fire) is likely 70%+ for recent PolitiFact once Half True (~10-20%) is subtracted. (2) Lee et al. 2023, as cited in arXiv 2610.17068: 11% of PolitiFact and 29% of Snopes articles rate True/Mostly True; Snopes also uses Mixture/Unproven/other labels, so its false-side share is plausibly ~50-60%. (3) QuanTemp (arXiv 2403.17169), 15.5k numerical claims from ClaimReview: 57.9% False, 23.3% Conflicting, 18.8% True; 'Conflicting' partly absorbs Mostly False, so the false side is ≥58%. (4) Older PolitiFact tallies (2007-2015, politician-centric) ran lower, roughly 40-50% on the false side. Combining: a central value of about 0.65 for the recent pooled population. The interval is wide to cover differences between organizations, ambiguity in how 'Misleading/Missing context' labels map onto the scale, and weighting by outlet volume. Not verified: Duke/ClaimReview global trend analyses, which would narrow the estimate.

Estimates

Each figure a source states for this quantity, in its own words, under the argument it belongs to and read against the graph's estimate.

Not yet placed under an argument

Missing Counter-Evidence Renders NLP Fact-Checking Unrealistic for Misinformation“After 2016, fact-checkers increasingly select false claims as important for fact-checking. In 2021 less than 10% of the selected claims were correct.”

QuanTemp: A real-world open-domain benchmark for fact-checking numerical claims“a distribution of 'True', 'False', and 'Conflicting' claims at 18.79%, 57.93%, and 23.27%, respectively, reflecting the tendency of fact-checkers to focus on false information”

Grey band: the graph's estimate, with its central value as a line. Dark marks: each source's own figure. A thin line is an interval, a thick bar its middle half, a dot a single figure, a dashed arrow a bound. An open box is where an argument places the value.

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.

or create a grant for this whole area →

Provenance

Where this quantity's evidence comes from. The figures sources state are under Estimates above.

2 sources · no origin · 2 untraced

2 Untraced

Missing Counter-Evidence Renders NLP Fact-Checking Unrealistic for Misinformation2022 · states Untraced · nothing upstream recorded

QuanTemp: A real-world open-domain benchmark for fact-checking numerical claims2024 · states Untraced · nothing upstream recorded

originuntraced

After 2016, fact-checkers increasingly select false claims as important for fact-checking. In 2021 less than 10% of the selected claims were correct.

Missing Counter-Evidence Renders NLP Fact-Checking Unrealistic for Misinformationextraction 1.00

a distribution of 'True', 'False', and 'Conflicting' claims at 18.79%, 57.93%, and 23.27%, respectively, reflecting the tendency of fact-checkers to focus on false information

QuanTemp: A real-world open-domain benchmark for fact-checking numerical claimsextraction 1.00

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.

jacksonqueenkingpropose questionNov 2, 2026

Proposed as a quantity: of the claims that professional fact-checking organizations select for checking, the proportion they rate false or mostly false. This figure is often cited as evidence about the prevalence of online falsehood, but fact-checkers choose claims that look suspicious or are spreading, so it is a reading of a selected sample, not a base rate. Having it as its own node lets the graph show how far it does and does not bear on the broader question of how much online content is false. Figures will vary by organization and period, so the definition should fix both.

Review · accepted · Nov 2, 2026 · contribution reviewer

The submission asks for the proportion of fact-checker-selected claims that are rated false or mostly false, and proposes it as a quantity. This meets the form bar for a question (§7a) and for a quantity (§2): it asks how much, and the contributor explains why it needs its own node. It concerns public organizations' published ratings, not a private person. The contributor's framing is sound. The figure is a reading of a selected sample and is not a base rate for online falsehood. Keeping it separate lets the graph show how far it bears on the broader prevalence question (§3). The definition is not yet fully specified. Organization, period, rating scale and selection process all need fixing before the quantity is well defined. The contributor says as much, and completing the definition is the work of the Matcher, Steward and Curator (§2, §3). Novelty and exact wording are likewise left to the Matcher. Accepted on form, good faith and the claim bar.

cites §2 · §3 · §7a · §13

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.