Washington's carbon market has added 24 to 46 cents a gallon to gasoline.

1. Pass Title matches the evidence

   Title says the carbon market added 24 to 46 cents a gallon; evidence items 1 and 4 report premiums of 24.2 to 46.1 cents per year and a finding of False against Ecology's 1-3% estimate. Consistent.

2. Pass No obvious contradictions

   Finding False agrees with the answer and conclusion, which say the effect has been several times the 1-3% estimate. Minor rounding differences (13.6 vs 14 cents) are not contradictions.

3. Pass Plausible justification

   Conclusion gives a clear reason for False: measured premium of 24 to 46 cents, about 6 to 12% of pump price, tracking auction carbon costs with slope near one, and matching the state consultant's estimates.

4. Pass Sources are relevant

   Ecology regulatory analysis and letter, EIA weekly prices, Ecology auction report, EPA emission factors and TRFC fuel volumes are all plausibly on topic for the gasoline price claim.

5. Pass Common questions answered

   Five common questions (Oregon comparison, usual premium, refinery outages, 2025 gas tax, program worth) each receive a direct response.

Coverage: Read the full article payload; sources judged by name/title/description/role only, per First Look's rules; no source snapshots read. startedAtUtc equals the submission time because the article was fetched earlier and the start time was not recorded then.

1. Concern (minor) Claims are supported and labeled

   Say in the title and keyPoints that the 24-46 cents is the Washington-specific premium, mostly attributed to the program, and label the 'most cost would not reach drivers' reading as inference.

   Major claims are cited: each evidence item names its sources and a saved dataset analysis, and the article separates measurement from inference in most places ("The gap appears to come from pass-through"; chart caption: "does not by itself prove causation"). Two inferences are stated a little too much like fact: (1) the title "Washington's carbon market has added 24 to 46 cents a gallon" attributes the full difference-in-differences premium to the program, while limitation e1 says the method "measures every Washington-specific change... not only the Climate Commitment Act"; the auction tracking supports attribution of most of it, so the impression is broadly right. (2) The keyPoint "The 1-3% estimate implied that most of that cost would not reach drivers" is an inference about Ecology's model, while limitation e4 admits "Ecology's own conversion is not reproduced here." Minor: peripheral inferences, not unsupported central claims. Confidence: Medium.

   About conclusion, evidence:estimate-vs-measured, limitation:annual-premium The finding

2. Concern (minor) Context is complete

   Explain what Ecology's 1-3% measured (base price, years covered, pass-through assumed) and why it differs from the ~46 cents its own $58.31 allowance assumption implies.

   Context is generally strong: the article states the program's mechanism, the September 2023 letter's position, the Clean Fuel Standard ("under 1 cent"), the missing Oregon series, the 2023Q1 near-zero premium, and California's small effects with the reason (its own carbon market). It also limits scope (no judgment on benefits). One gap: the article reports that Ecology's primary scenario assumed $58.31, which it computes as about 46 cents (about 10% of a $4.53 price), yet Ecology published 1-3%, and the article does not explain the discrepancy beyond "Ecology's own conversion is not reproduced here." The article does not address whether the 1-3% referred to a different base, period or modeled quantity; a reader would want this to judge whether the comparison is like-for-like. Because Pub. 22-02-047's Table 78 could not be read (only its summary page), I could not check this myself. Given the size of the measured gap, I do not expect it to change the overall impression, hence Minor. Confidence: Medium.

   About conclusion, evidence:estimate-vs-measured, source:ecology-fra-2022 The finding

3. Not checked Central claims are accurate

   Summary: the article says Ecology's pre-program 1-3% fuel-price estimate was far too low, because measured Washington gasoline prices have carried a 24-46 cent (6-12%) premium since 2023 that tracks allowance prices. Claim ratings: (1) Ecology's FRA estimated "a 1-3% increase in retail transportation fuel prices": Unverifiable; the FRA summary page names Pub. 22-02-047 (Sept 2022) but not the figure, and the letter is an unreadable PDF. (2) 7-state premium "41.6 cents in 2023 (SE 12.0), 24.2 in 2024 (4.9), 46.1 in 2025 (6.1) and 44.4 in 2026": True against the dataset's annual-effects table (exact match); the EIA source page shows only current weeks (Washington $5.424 vs U.S. $4.354 on 10/05/26), consistent but not a check of the history. (3) Other comparisons "between 20 and 56 cents", 2025 "41 to 56": True (dataset: 20.2 to 55.9; 2025 41.2, 53.2, 55.9). (4) Percentages 9.2/6.0/11.6/9.7%: True arithmetic (41.6/453, 24.2/402, 46.1/397, 44.4/459). (5) Pass-through: means 38.7 vs 38.9, slopes 1.03 and 1.02, correlation 0.62: True against pass-through-summary; the 2023Q4-to-2024Q1 drop (premium 56.1 to 23.9, carbon cost 41.2 to 20.3) is in quarterly-comparison. (6) Placebo "up to about 21 cents" and baseline sensitivity "under 6 cents": True (placebo max 21.1; 2023 7-state 41.6 vs 36.6). (7) Consultant estimates 51.4/36.0/24: match the dataset's replication-check, but Pub. 26-14-043 is unreadable, so Unverifiable at source. (8) "$58.31 allowance... about 46 cents": arithmetic True (58.31 x 7.902 kg = 46.1 cents); the $58.31 figure and EPA factor are Unverifiable (sources unreadable). (9) Cost ~$3.8 billion and $5.1 billion: match cost-estimates; TRFC volumes Unverifiable. (10) Letter's Oregon comparison: Unverifiable. Every figure I could check is accurate and the False finding follows from them, but the claim under test (Ecology's 1-3% and what it measured) and the consultant replication rest on sources that could not be read, so this check is NotChecked per the auditor rules. Confidence: High on the dataset matches; none on the unreadable primary documents.

4. Pass Sourcing is strong and transparent

   The article relies on primary sources for each part: the estimate (Ecology Pub. 22-02-047, Table 78, and Ecology's own letter), prices (EIA weekly retail series), allowance prices (Ecology auction reports), emission factor (40 CFR 98 Table MM-1), volumes (TRFC forecast), and the replication (Ecology Pub. 26-14-043). It discloses weaker links instead of hiding them: the letter copy "is hosted by the Washington Policy Center, which opposes the program; the letter itself is on Ecology letterhead", and "Six of the fifteen auction prices come from secondary reports". Numeric claims map to named dataset analyses, so which source supports which claim is clear. The FactHub dataset is the article's own analysis, but it is built from and reproducible against primary data. Several sources could not be read (see coverage), so this rests on how they are described, not their content. Confidence: Medium-High.

5. Pass Framing is fair and neutral

   The claim is attributed to Ecology's own published estimate and director's letter, and the opposing view ("a large majority of Washington's 2023 price increase could not be attributed to the program") is stated in context and answered in the objections with evidence, including a concession that Oregon is "the most important missing comparison." The article discloses the advocacy host of the letter and states it has no affiliation with campaigns for or against Initiative 2117. It explicitly declines to judge the program's merit ("says nothing about whether the program's benefits justify the cost"). Language is neutral. Confidence: High.

6. Pass Uncertainty is stated

   The article separates what is firm from what is not: "High that the effect has exceeded 1-3%... moderate on its exact size, because single-year estimates carry about 20 cents of noise and the comparison captures all Washington-specific factors." Limitations state the placebo noise (dataset placebo-2019: -20.6 to +21.1 cents), the 15-quarter sample and 0.62 correlation, the missing Oregon series, secondary auction prices, forecast volumes and diesel upper bounds. Conflicting evidence (Ecology's Oregon comparison) is presented and weighed with reasons. The tightest year is 2024 (24.2 cents vs a ~12-cent ceiling), but the 2024 carbon cost of 20-31 cents in the dataset corroborates it, so High on exceedance is defensible. Confidence: Medium-High.

Coverage: Read all eight sources with get_source_snapshot. Readable: ecology-fra-2022 (publication summary page only; it does not contain Table 78 or the 1-3% figure) and eia-weekly (current-week table only, no history). Unreadable: ecology-letter-2023, ecology-pub-26-14-043, ecology-auction-15 and trfc-2026-02 (PDF, which FactHub cannot read yet); epa-mm1 (eCFR returned a bot-check page, not the regulation); facthub-dataset source entry has no URL, so the cited dataset node dataset:wa-cca-gasoline-price-premium-2015-2026:2 (digest c77d8375...) was read directly and its annual-effects, quarterly-comparison, pass-through-summary, replication-check, placebo-2019, baseline-sensitivity and cost-estimates analyses were compared with every figure in the article. The Ecology 1-3% estimate, the $58.31 allowance assumption, the letter's Oregon figures, the consultant's Table 2 values, auction prices, the EPA emission factor and TRFC volumes could not be checked against their primary sources.

1. Concern (minor) Confidence matches the evidence

   Reword the title to say the 24-46 cent premium is Washington-specific and mostly the program, and qualify 'every full year' for 2024 given about 20 cents of single-year noise.

   The split confidence ("High that the effect has exceeded 1-3%... moderate on its exact size") is well judged, and I would also give High to the exceedance claim. Two gaps: (1) The title states "Washington's carbon market has added 24 to 46 cents a gallon" as a causal fact, while the conclusion and limitation e1 say the method "cannot fully separate the program from other Washington-specific shocks" and the comparison "captures all Washington-specific factors, not only the Climate Commitment Act". The auction tracking in e2 supports attribution of most of the premium, so the reader's core belief is right, but the title is a step stronger than the article's own caveat. (2) "High that the effect has exceeded 1-3% of the pump price in every full year": for 2024 the point estimate is 24.2 cents against a 1-3% ceiling of about 12 cents, while placebo deviations reach about 21 cents in single years; the auction cost for 2024 supports it, but 2024 is the year where High is least secure. Neither changes the verdict, hence Minor. Confidence in this check: Medium.

   About conclusion, evidence:annual-premium, limitation:annual-premium The finding

2. Concern (minor) Strongest objections answered

   Add an objection addressing what Ecology's 1-3% figure measured (base price, years, modeled pass-through) and why it differs from the ~46 cents its own $58.31 allowance implies.

   The objections answer the main alternative explanations well: Oregon's raw 2023 rise (answered, and the missing Oregon series is acknowledged), Washington's usual premium (removed by baseline, placebo cited), refinery shocks (answered with the quarter-by-quarter auction co-movement including the 2024 drop), and the gas-tax increase (normalized out). One defender's objection is not answered: the article notes that Ecology's primary scenario assumed $58.31, which implies about 46 cents (roughly 10%), yet Ecology published 1-3%, and says only that "Ecology's own conversion is not reproduced here" and that the estimate "implied that most of that cost would not reach drivers". The article does not address whether the 1-3% referred to a different quantity (for example a modeled price-index change, a different base year, or an average over a longer horizon), which a defender would raise to argue the comparison is not like-for-like. Given the size of the gap this would likely not change the rating, so Minor. Confidence: Medium.

   About conclusion, evidence:estimate-vs-measured, limitation:estimate-vs-measured The finding

3. Concern (material) Evidence is independent

   Drop 'several independent lines' from the conclusion: e1, e2, e4 and e5 all rest on one FactHub DiD of EIA prices, and e3 uses the same method and partly the same data.

   The conclusion opens "Several independent lines agree". Judged from methods and sources: e1 is the FactHub difference-in-differences on EIA weekly prices; e2 regresses that same measured premium on auction-implied carbon cost (adds auction data, but the premium side is e1's series); e4 divides e1's premium by EIA Washington prices (its limitation says the percentages are "FactHub arithmetic from the dataset's effects"); e5 multiplies the same premium or carbon cost by TRFC volumes. e3, the Greene Economics study, is a different analyst, but its limitation says it uses "the same basic method and some of the same data". So the evidence presented as independent comes down to one underlying approach (DiD on retail price data, largely EIA). The auction-cost arithmetic in e2/e4 is the one genuinely separate support (it shows the expected size of full pass-through), and it does reinforce the conclusion, but it is a prediction, not a measurement. I would keep the rating, but describe the case as one measurement method, replicated by the state's consultant and consistent with the auction-cost mechanism, rather than several independent lines. Confidence: Medium-High.

   About conclusion, evidence:annual-premium, evidence:auction-tracking, evidence:estimate-vs-measured, evidence:cost-to-drivers, evidence:state-consultant The finding

4. Pass Rates the real claim

   The claim "Washington's carbon market would raise gasoline prices by only 1 to 3 percent" matches how the estimate circulates (officials' pre-program assurance of a small effect). The article ties it to Ecology's Pub. 22-02-047 estimate of a "1-3% increase in retail transportation fuel prices"; gasoline is the main retail transportation fuel, so the narrowing is reasonable and is stated in method step 1. The conclusion stays within Washington, 2023-2026, and price effect only ("says nothing about whether the program's benefits justify the cost"). Confidence: High.

5. Pass Rating follows from the evidence

   Before reading the verdict I would rate the claim False: evidence e4 converts the 7-state premium to about 9.2%, 6.0%, 11.6% and 9.7% of the pump price, two to four times the top of the 1-3% range, and e2 shows the auction-implied carbon cost (avg 38.9 cents) is of the same size as the premium (38.7 cents), so the large effect is expected from the mechanism, not only from the price comparison. Even the weakest year (2024, 24.2 cents, SE 4.9) is roughly 6%, double the top of the range. The finding False and the answer ("about 24 to 46 cents a gallon, several times the 1-3% estimate") agree with each other and with the evidence. Confidence: High.

Coverage: Examined the full article payload: title, claim, finding, answer, confidence, summary, keyPoints, context, conclusion, all five evidence items with methods, institutions, years and limitations, all five objections, method steps, and source list. Per the Skeptic's instructions, source snapshots and the cited dataset were not read; evidence was taken as stated.

1. Concern (minor) Title states the finding

   Make seoTitle state the finding, e.g. 'Washington cap-and-invest has raised gas prices 24 to 46 cents a gallon, far above the 1-3% estimate'.

   Title: "Washington's carbon market has added 24 to 46 cents a gallon to gasoline." It states the finding in positive form, names the subject, does not repeat the myth, and is neutral, so it meets the TITLE-FORM rule. seoTitle: "How much has Washington's cap-and-invest program raised gas prices?" is a question that does not state the finding; the rule says seoTitle "may use the wording people search with as long as it states the finding". Minor per the rule ("a seoTitle that does not state the finding").

   About conclusion The finding

2. Concern (minor) Verdict fields agree and stand alone

   Make the answer stand alone, e.g. 'Washington's carbon market has added about 24-46 cents a gallon since 2023, 6-12% of the price', and reword the description to the rated claim.

   Read as: FactHub rates the claim "Washington's carbon market would raise gasoline prices by only 1 to 3 percent." False (High confidence): "Since 2023 the increase has been about 24 to 46 cents a gallon, several times the 1-3% estimate." The claim is worded as it circulates and is not its correction, and title, label, summary and conclusion agree. Two lapses: (1) the answer does not stand alone when quoted: "the increase" does not say an increase in what or caused by what, and it compares cents with a percent without the conversion; the rule says the answer "stands alone when quoted and says what is actually true". (2) Stale wording from version 1: the description still reads "Fact check: the Washington Department of Ecology's estimate that cap-and-invest would raise retail fuel prices by 1-3%", and method step 1 "Defined the claim narrowly as Ecology's published estimate", although the changelog says the claim "is now stated as it circulates... rather than as a quotation of Ecology's estimate". Both are Minor under the rule (accurate answer that does not stand alone; stale wording).

   About conclusion, method:1 The finding

3. Concern (minor) Each section does its job

   Cut the summary to one sentence naming the strongest lines (the 7-state price comparison and its tracking of auction prices); leave the yearly figures to keyPoints.

   Rule: "summary is one sentence naming the strongest independent lines of evidence." The summary is three sentences and repeats the year-by-year figures that keyPoints 2 already lists ("41.6 cents in 2023, 24.2 in 2024, 46.1 in 2025 and 44.4 in 2026"). Other fields meet the rules: context is one paragraph of framing; keyPoints state what the evidence establishes; the conclusion explains how the lines fit, the pass-through explanation, what remains uncertain and the confidence split, rather than restating the verdict; the editorialNote has a scope statement, caveats and external-review status. No hand-written independence note. Minor: one field carrying content that belongs elsewhere.

   About conclusion The finding

4. Concern (minor) Consistent house style

   Explain difference-in-differences, E10 and Newey-West at first use, spell out EIA in the summary, and use one form ('1-3%') and one rounding (4.5 to 13.6 cents) throughout.

   Rules: technical terms explained at first use; abbreviations spelled out at first use; same rounding for the same figure everywhere; consistent formats. Lapses: "difference-in-differences" (summary), "Newey-West standard errors with 12 lags" (e1 method) and "E10 gasoline" (e2 method) are not explained; "EIA" first appears in the summary ("weekly EIA prices") without being spelled out; the 1-3% range is converted as "roughly 4.5 to 14 cents" in the summary but "about 4.5 to 13.6 cents" in keyPoints; and the claim says "1 to 3 percent" while every other field says "1-3%". No leftover drafting text and no conflicting values for the same figure, so Minor.

   About conclusion, evidence:annual-premium, evidence:auction-tracking The finding

5. Pass Limitations are specific

   Each limitation is specific to its item: e1 names the Clean Fuel Standard, refinery outages, ~21-cent placebo deviations and the missing Oregon series; e2 names the 15-quarter sample, 0.62 correlation, the near-zero 2023Q1 premium and six secondary auction prices; e3 states shared method and AAA vs EIA differences; e4 explains the pump-price dependence of the percent conversion; e5 flags forecast volumes and diesel upper bounds. None is boilerplate.

6. Pass Tone is measured and respectful

   Language is measured: "appears to come from pass-through", "consistent with full pass-through but does not by itself prove causation", "this article does not claim the whole premium is the program". Objection responses are short, evidence-based and respectful, e.g. the Oregon response concedes "it is the most important missing comparison". No ridicule or unwarranted absolutes.

7. Pass Readers can verify it

   All eight sources carry meaningful role labels ("The estimate under test", "Claim provenance", "State-commissioned estimate (replication)", "Principal price data"), full titles with publication and table numbers (Pub. 22-02-047 Table 78; Pub. 26-14-043 Table 2; Table MM-1; Vol. II Table A.1), and descriptions of why each was chosen, including a note that the letter copy is hosted by an advocacy group. Key numeric claims anchor to the versioned FactHub dataset with named analyses. Method steps cover claim definition (step 1), sources reviewed (steps 2-3), evidence beyond one source (steps 4-5: placebo, sensitivity and consultant replication) and recorded limits (step 6).

Coverage: Examined only the article payload: title, seoTitle, description, claim, finding, answer, confidence, summary, keyPoints, context, conclusion, five evidence items and limitations, five objections, six method steps, editorialNote, chart caption, sources with roles, titles, identifiers and descriptions, and changelog. No sources or web were used.