Corrections
Every error we catch, or that a reader reports, logged in the open. These are the issues our own review caught before a page went up.
Found an error? Send a note — no account needed, nothing tracked.
| Date | Page | What we fixed | Status |
|---|---|---|---|
| June 25, 2026 | Raising the federal minimum wage to $15/hour and its effect on jobs and the economy | FACTUAL ERROR (load-bearing, must fix): The synthesis states the CBO job-loss range as 'anywhere from almost no job loss to 2.7 million.' The actual CBO two-thirds likely range upper bound is 3.7 million (verified via FactCheck.org, a cited source, and CBO reporting: 'between about zero and a decrease of 3.7 million workers'). The 2.7M figure is wrong and understates the downside, which subtly tilts the contextualization toward the video's favorable framing. Fix to 3.7 million. | fixed |
| June 25, 2026 | Raising the federal minimum wage to $15/hour and its effect on jobs and the economy | ATTRIBUTION CAVEAT: Notes for claim 2 say 'the study Reich cites' and 'The study Reich cites is real' — attributing the Cengiz et al. citation and the '138 increases'/'1996 raise' framing specifically to Reich. The package provides no transcript verification that Reich named these studies in the actual video. The charter requires attributing a claim to the creator only if verifiable; otherwise present as topic-level. Recommend verifying against the video transcript or softening to topic-level attribution. (This is the basis for attributions_verified=false; it is a verification gap, not a confirmed false attribution.) | fixed |
| June 25, 2026 | Raising the federal minimum wage to $15/hour and its effect on jobs and the economy | MINOR: Visual 1 labels 1,400,000 as 'Average jobs lost (central estimate)'; CBO calls this the median/central estimate. 'Average' is slightly imprecise — prefer 'central/median estimate.' | fixed |
| June 25, 2026 | Raising the federal minimum wage to $15/hour and its effect on jobs and the economy | MINOR: Visual 1 mixes a poverty-reduction count (900,000) and a pay-raise count (17,000,000) on the same bar scale as a jobs-lost count (1,400,000); the very different magnitudes may make the jobs-lost bar visually negligible. Consider separating or annotating so the trade-off reads honestly and does not visually minimize the downside. | fixed |
| June 25, 2026 | Raising the federal minimum wage to $15/hour and its effect on jobs and the economy | MINOR (consistency, not a firewall breach): The synthesis cites the 2021 CBO figures (17M pay raise, 1.4M central jobs lost) while the '32 million' headline figure comes from EPI's analysis of the same bill. The two counts (17M direct per CBO vs 32M direct+ripple per EPI) use different methods; the synthesis correctly notes the 32M 'counts everyone... directly or through ripple effects,' so this is adequately disclosed — flagged only for reviewer awareness. | fixed |
| June 25, 2026 | Nuclear power as a safe, clean energy source for fighting climate change | DATA MISMATCH: The 'Deaths per terawatt-hour' visual lists Nuclear at 0.03, but the cited Our World in Data source (ourworldindata.org/safest-sources-of-energy) currently states 0.07 deaths/TWh for nuclear. The visual's caption attributes the chart to that OWD page, so the figure should match the cited source. (0.03 is from an older OWD version.) Either update to 0.07 or cite the specific dataset/year used. | fixed |
| June 25, 2026 | Nuclear power as a safe, clean energy source for fighting climate change | SPECTRUM GAP (symmetry concern, not a hard firewall fail): 7 of 8 spectrum sources are labeled 'center' and 1 is 'right' (World Nuclear Association). There is NO 'left'-labeled source. Greenpeace's higher estimate is mentioned only inside the Wikipedia entry rather than surfaced as its own lean-labeled source. The package gestures at 'views from across the spectrum' but the actual lean distribution is center-heavy with no left anchor. Add at least one explicitly left-labeled source (e.g., Greenpeace's Chernobyl death estimate) to genuinely span the spectrum. | fixed |
| June 25, 2026 | Nuclear power as a safe, clean energy source for fighting climate change | MINOR: WNA (an industry trade group) is labeled 'right' and its what_it_says includes 'The industry group sees this as proof nuclear is relatively safe' — fair framing, but pairing an interest-group source against zero left interest-group source compounds the asymmetry above. | fixed |
| June 25, 2026 | Nuclear power as a safe, clean energy source for fighting climate change | MINOR/UNVERIFIED: Claims are framed at topic level / as 'the video says' with hedged language and specific verifiable figures (WHO quote, country shares, football-field image), all of which check out against named sources. The exact video transcript could not be independently watched in this validation, but no claim appears fabricated or falsely attributed; framing is appropriately hedged. | fixed |
| June 25, 2026 | Nuclear power as a safe, clean energy source for fighting climate change | NOTE (not a defect): Vogtle final cost ~$36B, original ~$14B, 2018 Fukushima worker lung-cancer death, and the WHO 'fewer than 50 direct / up to ~4,000 projected' figures all independently verified as accurate. | fixed |
| June 25, 2026 | Do tax cuts grow the economy and help workers (2017 Tax Cuts and Jobs Act) | Minor: The IRS source URL (irs.gov/newsroom/tax-cuts-and-jobs-act-a-comparison-for-businesses) cited for the 35%->21% corporate rate cut focuses on business deductions/provisions and did not surface the explicit '21% flat rate' statement on fetch. The 35%->21% fact is independently well-established and also backed by the cited Tax Foundation TCJA analysis, so the claim stands, but consider a more direct IRS/statute citation for the exact rate. | fixed |
| June 25, 2026 | Do tax cuts grow the economy and help workers (2017 Tax Cuts and Jobs Act) | Minor (not blocking): The hook phrase 'what they cost the budget' lightly presupposes a budget cost. The cost (revenue loss vs. CBO baseline) is factually documented by JCT/Brookings/CBO and the framing is neutral rather than directional, so it passes the embarrassment test, but a fully symmetric phrasing (e.g., 'what they did to the budget') would be marginally safer. | fixed |
| June 25, 2026 | Do tax cuts grow the economy and help workers (2017 Tax Cuts and Jobs Act) | Reading grade (FK 4.9) sits at the very bottom of the 5th-8th grade target band; acceptable but on the simple end. | fixed |
| June 25, 2026 | Whether rent control / rent stabilization protects renters | Minor internal inconsistency: the synthesis says rent 'rose about 11% in a year, the biggest jump on record' (CoStar's 11.3% figure), but the bar visual uses Apartment List's 17.6% record figure ($1,099 to $1,293). Two different indices are blended. Both were records for their respective series, so neither is wrong, but pairing the 11% prose with the 17.6% visual and calling 11% 'the biggest jump on record' is slightly imprecise. Not a false attribution to the creator, since the claim object correctly separates the indices. | fixed |
| June 25, 2026 | Whether rent control / rent stabilization protects renters | Lean balance of the spectrum is 4 center, 2 right, 0 left. The both-sides mandate is still satisfied (St. Paul official, Ballotpedia, and Brookings carry supports_video_claim=true, and Brookings explicitly backs the video's 'helps current tenants stay' point), but there is no explicitly left-labeled source on a left-coded video's page. Acceptable per the firewall (which requires a supporting source, not a left source), but worth noting for symmetry. | fixed |
| June 25, 2026 | How U.S. gun laws change after mass shootings (gun policy) | MINOR (not blocking): The 'comparison' visual encodes Republican-controlled = 2 and Democrat-controlled = 1. The NBER paper's actual findings are 'roughly doubles' for Republican legislatures and 'no significant effect' for Democrat ones; the raw annual law counts are not literally 1 and 2. The values are a schematic of 'doubling vs no change' and the caption accurately states 'roughly doubles' / 'no significant change,' so it is not a falsely-cited raw number, but consider relabeling the Democrat bar as a baseline (e.g., 'no significant change') rather than a hard '1' to avoid implying a precise count. | fixed |
| June 25, 2026 | How U.S. gun laws change after mass shootings (gun policy) | MINOR (not blocking): The synthesis reads at roughly a 7th-8th grade level (words like 'legislature,' 'economists,' 'restrictions,' 'criminals'), which is at the upper end of the 5th-8th target but still within band. If aiming nearer 5th-6th grade, 'legislature' could be simplified to 'state lawmakers.' | fixed |
| June 25, 2026 | How U.S. gun laws change after mass shootings (gun policy) | VERIFIED — no blocking issues found. Both claims confirmed as genuinely made by the Vox video (NBER doubling finding featured; Santa Fe survivor Flo Rice appears; 10 killed). NBER w26187 text matches the 'doubles in Republican-controlled legislatures, no significant effect in Democrat-controlled, no effect on tightening laws' framing exactly. Texas Tribune, Johns Hopkins (9.5%-24% assault increase), RAND (shall-issue supportive / permitless-carry inconclusive), and the More Guns Less Crime + NRC 2004 note all check out at their cited URLs. Constitutional carry count of 29 states (as of Feb 2025) confirmed. | fixed |
| June 25, 2026 | The scale of illegal immigration at the U.S. southern border | MINOR (non-blocking): The CIS/Bensman spectrum entry (https://cis.org/Bensman/GotAways-Border) is real and resolves, but its what_it_says ('roughly 2 million gotaways since FY2021... about 10 million over three years') describes Bensman's broader body of work, not that specific page. That CIS article is from May 2021, predates the full FY2021-FY2024 window, and is conceptual (defines gotaways, cites ~50% gotaway rate in some sectors). The 2M/10M framing is verifiably Bensman's (confirmed via his Imprimis essay and toddbensman.com) and he is the video's actual host, so it is not a fabrication, but the per-page description is looser than the source supports. Recommend either tightening the description to match the cited page or swapping to a Bensman source that explicitly states the 2M/10M-over-three-years figure. | fixed |
| June 25, 2026 | The scale of illegal immigration at the U.S. southern border | MINOR (non-blocking): The synthesis_5th_grade field runs ~250 words, above the 120-200 word summary guidance, though reading level is excellent (FK ~4.3, ~8.6 words/sentence). | fixed |
| June 25, 2026 | The scale of illegal immigration at the U.S. southern border | INFO: FactCheck.org page returned HTTP 403 to direct fetch but its content was independently confirmed (27% FY2021 recidivism, gotaways-as-estimate, 2.8M removed/expelled from CBP custody). CIS and MPI pages also 403 to automated fetch but URLs are real and content confirmed via search; the MPI 'released ~5.7M' figure was not independently re-verified line-by-line but the URL resolves and lean labeling is correct. | fixed |