Hydropower generates most of Washington's electricity; Grand Coulee leads U.S. plants by capacity.

1. Pass Title matches the evidence

   Title says hydro generates most of Washington's electricity and Grand Coulee leads by capacity; evidence items 1 and 5 support exactly those qualified statements, including the nuclear-output caveat.

2. Pass No obvious contradictions

   Mostly true rating agrees with the conclusion, which accepts the generation claim and qualifies the supply and output readings. Drought-year figures (59.7%, 59.3%) match across the evidence and key points.

3. Pass Plausible justification

   Conclusion explains the rating by separating generation, delivered power and capacity versus output, and shows which meanings hold; a reader would accept that as the reason for Mostly true.

4. Pass Sources are relevant

   EIA generation and plant data, Washington Commerce fuel mix, Bureau of Reclamation, Ecology drought declarations and EIA drought analysis are all on topic for hydro share and Grand Coulee.

5. Pass Common questions answered

   Five common questions (imports, Palo Verde, capacity figure, decline, unspecified power) each get 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) Context is complete

   Add Commerce's delivered fuel mix for non-drought years (e.g. 2019-2022) so readers can tell whether the 2024 48.73% hydro share is typical.

   The article handles the main ambiguities well: generation vs delivered power, capacity vs output, trade (net importer in 2001, 2010, 2023, 2024), unspecified power, and the declining trend. One gap: the delivered-to-customers reading, which the conclusion calls "the weaker part" of the claim, rests on 2024 alone, a drought year with the lowest in-state share on record. Commerce's page says it publishes "Fuel mix data and reports" (a multi-year series), but the article does not report any non-drought year, so a reader cannot tell whether hydro is typically above or below half of delivered power. The article does flag "One year only" and the drought, so the overall impression (generation majority; delivered plurality in 2024) is unchanged; hence Minor. Confidence: Medium.

   About evidence:delivered-fuel-mix, limitation:delivered-fuel-mix, source:wa-commerce-fuel-mix The finding

2. Not checked Central claims are accurate

   Summary: hydro produced most of Washington's in-state generation every year 2001-2025, Washington leads all states in hydro, Grand Coulee is the largest U.S. plant by capacity but 27th by 2025 output, and hydro was 48.73% of 2024 delivered power. Ratings against readable sources: (1) Grand Coulee 6,809 MW and largest by capacity: True for the operator figure; Reclamation says "generating capacity of more than 6,809-megawatts" and calls it "the largest hydropower generating complex in the United States". (2) EIA cited about 6.7 GW: True ("net summer capacity of 6.7 gigawatts (GW)"). (3) Reclamation max about 21 billion kWh: True ("up to 21 billion kilowatt hours"). (4) Drought declarations: True; Ecology declared drought for "most of Washington" on April 16, 2024, continuing the 2023 emergency that "covered 12 watersheds". (5) EIA's 2023 monthly hydro near/below low end of 10-year range: True ("close to or below the lower end of the 10-year range, except for a brief spike in May"). Unverifiable against primary sources: the central figures, i.e. the 59.3%-79.7% share series, state rank, EIA-860 capacity sums (7,079 MW summer, Vogtle 4,658 MW), EIA-923 output ranks (Vogtle 37.1 TWh, Palo Verde 31.3 TWh, Grand Coulee 16.2 TWh, 27th), supply/disposition trade figures and the Commerce 48.73% figure. Those EIA files and the Commerce chart could not be read. The FactHub dataset reproduces every one of these figures, and I recomputed some from it (2001-2010 mean 71.0%, 2016-2025 mean 65.1%, hydro-to-retail-sales range 67.5%-102.2%, net importer only in 2001, 2010, 2023, 2024), but that dataset is FactHub's own derivation, so it confirms internal consistency rather than accuracy against EIA. Following the auditor rule for unreadable sources, the central-claims accuracy check is NotChecked rather than rated from my own knowledge. Confidence: High in the readable-source ratings; central figures unexamined.

3. Pass Claims are supported and labeled

   Every evidence item lists sourceIds, and each figure traces to a named saved analysis in the dataset (e.g. wa-hydro-share-summary: "below 60% only in 2023 and 2024; never at or below 50%"). Inferences are labeled as such: the drought item's limitation says "the timing match is descriptive and does not by itself measure how much drought reduced generation", and EIA's Today in Energy article separately attributes reduced Pacific Northwest hydro output to drought. The conclusion separates fact ("never less than 59.3%") from interpretation ("A careful version of the claim is...") and declines to claim a delivered majority ("the data cannot show it"). No central claim rests on assertion. Confidence: High.

4. Pass Sourcing is strong and transparent

   Central claims rest on primary sources: EIA's state generation tables, EIA-860 and EIA-923 census files, EIA State Electricity Profile Table 10, Washington Commerce's statutory fuel-mix compilation, Reclamation as operator, and Ecology's drought declaration. Source-to-claim mapping is explicit via sourceIds and dataset analysis ids. The article discloses that most figures rest on one institution (EIA) and that Reclamation "is not independent of" the plant. The 2024 delivered mix is transcribed from Commerce's own chart, a primary state source; the dataset notes the underlying spreadsheets sit in a Box folder. No partisan, anonymous or circular sourcing. Confidence: High.

5. Pass Framing is fair and neutral

   The original claim is quoted as it circulates and each reading is tested, including the less favorable ones (delivered power, annual output). The title leads with the true part and carries the capacity qualifier; the summary names both qualifications. Objections, including "Unspecified power is really fossil fuel", get evidence-based answers. No loaded language or selective emphasis that changes the impression. Confidence: High.

6. Pass Uncertainty is stated

   Uncertainty is stated wherever it matters: 48.73% is "a floor rather than an estimate"; the trade figure is a residual that "absorbs errors in estimated losses"; capacity definitions differ (6.7 GW vs 6,809 MW vs 7,079 MW) and the conflict is explained and weighed ("All three figures rank Grand Coulee first"); the conclusion says "Nothing here establishes how much drought specifically reduced output, and the 2024 consumption-side figure is a single year"; the editorial note says EIA revises data. High confidence is reasonable for a Mostly true rating built on census data. Confidence: High.

Coverage: Read the full article payload. Read source snapshots: usbr-grand-coulee (archive), eia-tie-2024-drought (archive), ecology-drought-2024 (archive), wa-commerce-fuel-mix (archive). Read the cited FactHub dataset dataset:wa-electricity-generation-by-source:1 (analyses, structured results, replications, limitations) and recomputed several article figures from it. UNREADABLE: eia-state-generation and eia-sep-washington (xlsx not readable by the snapshot tool), eia-860-2025 and eia-923-2025 (zip over 5,120 KB), wa-commerce-fuel-mix-chart (PNG not readable), facthub-chart (no URL; it charts the dataset analysis that was read). The primary EIA data behind the hydro-share, state-rank, plant-capacity and plant-output figures, and the Commerce chart behind the 48.73% delivered figure, could therefore not be checked against the originals; claims-accurate is NotChecked for that reason.

1. Pass Rates the real claim

   The claim "Hydropower supplies most of Washington's electricity, and Grand Coulee is the largest U.S. power plant" is a compound claim as it commonly circulates, and the article rates both halves under each plausible reading: generation vs delivered power, and capacity vs annual output (method step 1). It neither strawmans (it tests the harder delivered-power and output readings) nor overreaches: the conclusion stays within Washington and the U.S. plant fleet, 2001-2025. Confidence: High.

2. Pass Rating follows from the evidence

   Before reading the verdict I would have chosen Mostly true: in-state generation majority every year 2001-2025 ("never less than 59.3%") is true; the delivered 2024 mix (48.73% with 15.44% unspecified, a floor) qualifies "supplies"; Grand Coulee is first by nameplate (6,809 MW) and net summer capacity (7,079 MW), the usual meaning of "largest" for a power plant, but 27th by 2025 output. Both halves hold under their most common meanings with a qualifier each, which is what Mostly true denotes. The answer ("Grand Coulee leads by capacity, not annual output") matches the label. Mixed would be defensible only if the output reading were the dominant one, which the article reasonably treats it as not. Confidence: High.

3. Pass Confidence matches the evidence

   High confidence rests on census-type federal data (EIA-923 state totals, EIA-860 generator census) with an operator cross-check (Reclamation's 6,809 MW) and an independent consumption-side source (Commerce). The title states the finding with its capacity qualifier, and the summary names both qualifications. Material limitations are stated: in-state vs delivered (limitation:in-state-generation-share), the single-year, utility-reported delivered mix with unspecified power (limitation:delivered-fuel-mix), the residual trade estimate, and differing capacity definitions. The conclusion also says the drought effect is unmeasured and the 2024 consumption figure is one year. I would give High for the Mostly true label. Confidence: High.

4. Pass Strongest objections answered

   The strongest objections I would raise are: (1) Washington trades power heavily, so in-state generation is not what Washington uses; (2) "largest" usually means output, where nuclear leads; (3) capacity figures conflict (6.7 GW vs 7 GW); (4) the share is declining; (5) unspecified power may be fossil. All five appear in the objections with evidence-based answers: the trade objection cites 67.5% hydro-to-retail-sales in 2024 and the delivered mix; the Palo Verde objection gives 2025 output and capacity figures; the 6.7 GW objection explains definitions; the decline objection gives decade averages; the unspecified-power objection treats 48.73% as a floor and explicitly declines to claim a delivered majority. None is answered by assertion. Confidence: High.

5. Pass Evidence is independent

   Several items (in-state-generation-share, drought-years, generation-versus-sales, national-hydro-rank, plant-annual-output) rest on EIA-923-derived data, so they are not independent of one another, and the article says so plainly (limitation:in-state-generation-share: "this line depends on one institution"; editorial note: "Most figures rest on EIA survey data"). The article does not present them as independent confirmations. Independent support exists where it matters: Commerce's utility disclosures for the delivered reading, Reclamation's operator figure for capacity (independent of EIA though not of the plant, as disclosed), and Ecology's drought declarations. For a census-based question, reliance on the official census is appropriate rather than a gap. Confidence: Medium-High.

Coverage: Reviewed the full article payload: claim, finding, answer, confidence, title, summary, context, conclusion, all seven evidence items with methods, institutions, years and limitations, five objections, method steps, editorial note and source list. Per the Skeptic's division of labor, evidence was taken as stated and no source snapshots were read.

1. Concern (minor) Title states the finding

   Make seoTitle state the finding, e.g. "Hydropower Generates Most of Washington's Electricity; Grand Coulee Leads U.S. Plants by Capacity".

   Title "Hydropower generates most of Washington's electricity; Grand Coulee leads U.S. plants by capacity." states the finding positively, names both subjects and resolves the claim with its key qualifier: passes. The seoTitle, "Does Hydropower Supply Most of Washington's Electricity? Is Grand Coulee the Largest U.S. Power Plant?", only poses the questions. Rule broken: "seoTitle may use the wording people search with as long as it states the finding" (Minor: a seoTitle that does not state the finding).

   About conclusion The finding

2. Concern (minor) Each section does its job

   Delete the hand-written independence note from editorialNote (the page generates it) and cut the summary to one sentence naming the strongest evidence.

   editorialNote contains "Independence note: the evidence comes from one country, four institutions ..., four methods ... and 2001-2025", a hand-written independence note that repeats what the page generates. Rule: "The independence note is generated by the page; the author must not write one" (Minor). Also, the summary runs three sentences, while the rule says "summary is one sentence naming the strongest independent lines of evidence". Otherwise the fields do their jobs: keyPoints state what the evidence establishes, the conclusion weighs the generation/delivered and capacity/output readings and states what remains unproven (the drought effect, single-year consumption figure), and editorialNote states scope, caveats and review status.

   About conclusion The finding

3. Concern (minor) Consistent house style

   Spell out EIA at first use, explain nameplate vs net summer capacity, use TWh/MW throughout, and drop the stray "Energy | Fact check." from context.

   Rule: "an abbreviation is spelled out at first use": EIA first appears in description and summary ("Federal EIA data") without being spelled out in reader text. Rule: technical terms explained at first use: "nameplate capacity" and "net summer capacity" are never explained. Rule: one unit system per quantity: energy appears as TWh, "21-29 million MWh" and "1,170,693 MWh" (generation-versus-sales) and "about 21 billion kWh" (limitation:plant-annual-output); capacity as MW and "6.7 GW"/"about 4.5 GW". Rule: same rounding for the same figure: description says "59-80%" where elsewhere it is 59.3% to 79.7%. Names: "Vogtle (GA)" vs "Vogtle (Georgia)". Context opens with stray text "Energy | Fact check.", which reads as a leftover label rather than framing; I treat it as Minor rather than Material drafting text since it is not a placeholder or editor note.

   About conclusion, evidence:generation-versus-sales, limitation:plant-annual-output, evidence:grand-coulee-capacity The finding

4. Pass Verdict fields agree and stand alone

   Read as: FactHub rates the claim "Hydropower supplies most of Washington's electricity, and Grand Coulee is the largest U.S. power plant." Mostly true (High confidence): "Hydropower has produced most of Washington's power since 2001; Grand Coulee leads by capacity, not annual output." The claim is worded as it circulates, the answer stands alone and says what is true, and the title, summary, keyPoints and conclusion all express the same qualified finding (majority of generation, capacity not output, delivered share under half in 2024). Side note, not a finding because changelog is outside the stated fields: the second changelog entry says the rating was standardized and "was 'Mostly true'", which is also the current label, so that entry reads oddly.

5. Pass Limitations are specific

   Every limitation is specific to its item: in-state-generation-share ("not electricity used by Washington customers"), drought-years (declarations describe water supply, timing match is descriptive), delivered-fuel-mix (one year, 15.44% unspecified, transcribed from a chart), generation-versus-sales (residual trade estimate, 2025 not published), national-hydro-rank (upstream water from Canada, Idaho, Montana), grand-coulee-capacity (differing definitions, operator not independent), plant-annual-output (one year, flow-dependent). None is boilerplate.

6. Pass Tone is measured and respectful

   Language is measured and qualified ("Both halves of the claim hold under their most common meanings, and both need a qualifier"; "the data cannot show it"). "EIA's state data are unambiguous" is justified by a census series that never drops below 59.3%. Objection responses are brief, evidence-based and respectful, e.g. "Trade matters, which is why this check reports delivered power separately." No ridicule or unwarranted absolutes.

7. Pass Readers can verify it

   All eleven sources carry meaningful role labels ("Primary federal data: generator capacity census", "Operator statement"), full titles, identifiers or file/table names (e.g. "SEP Tables for WA.xlsx, Table 10"), and descriptions of why each was chosen; key figures have primary anchors (EIA-860/923 files, archived copies, the Commerce chart image, the FactHub dataset with build scripts). Method steps cover claim definition (step 1), sources reviewed (step 2), checks against independent sources (step 5) and recorded limits (step 6).

Coverage: Reviewed only the article payload: title, seoTitle, description, claim, finding, answer, confidence, summary, keyPoints, context, conclusion, all seven evidence items and limitations, five objections, seven method steps, editorialNote, chart caption, eleven sources (role, title, identifier, description) and changelog. No sources or web content were consulted.