Corrections

Every correction to published analysis, with the original wording kept legible.

This site's argument is that claims should be sourced and checkable. That obligation applies hardest to its own mistakes, so corrections are published rather than quietly edited: each entry records what the text said, what it says now, and why the original was wrong.

Entries are never removed. Where a correction weakens a point the original was making, it says so.

  1. Agricultural land grading paragraph

    englands disappearing farmland

    Was
    Grades 1 and 2 together comprise approximately 21% of agricultural land in England [linked to the GOV.UK guide to assessing development proposals on agricultural land], with Sub-Grade 3a adding a further proportion.
    Now
    Natural England’s 2012 note on the classification gives “current estimates” that Grades 1 and 2 “together form about 21% of all farmland in England”, and that Subgrade 3a “also covers about 21%”; it does not say how the estimates were made. [Linked to Natural England’s technical information note TIN049.]
    Why
    The figure is right, but it was cited to a page that does not contain it: the GOV.UK guide has no “21” in any form, on the live page or in five archived copies from 2019 to 2026. The 21% is Natural England’s, from its technical information note TIN049 of 19 December 2012, where it is a “current estimate” with no method given. The sentence now cites that note, dates it, and gives the note’s matching estimate for Subgrade 3a in place of “a further proportion”.
  2. Solar farms section, “Where the panels actually go”

    englands disappearing farmland

    Was
    CPRE found that 59% of England's largest operational solar farms are located on productive farmland [linked to CPRE's press release], with almost two-thirds built on Best and Most Versatile land. Government policy states that best agricultural land should be prioritised for food production [linked to a GOV.UK news story of 15 May 2024].
    Now
    CPRE’s study of the 38 solar farms over 30 MW operating in England in June 2025 found that 59% of the land they use is productive farmland: 31% is Best and Most Versatile land and 28% is Grade 3b. Counted by farm, 25 of the 38 sit at least partly on one or the other, which is the “two-thirds” in CPRE’s press release. In May 2024 a ministerial statement said that “the highest quality agricultural land is least appropriate for solar development”.
    Why
    Three errors in two sentences. In CPRE’s own report 59% is a share of the LAND, “nearly two thirds (59%) of all the land used … is productive farmland”, and it counts Grade 3b with the best grades; the sentence repeated CPRE’s press release, which made it a share of the farms. The “two-thirds” is the count of farms, 25 of 38, with any Best and Most Versatile or Grade 3b land; only 20 of the 38 have any Best and Most Versatile land, so “almost two-thirds built on Best and Most Versatile land” was wrong. And “should be prioritised for food production” is a news story’s paraphrase of a minister, not policy text, so the statement it reported is now quoted instead.
  3. What Would Actually Help, point 2 (sequential testing for solar)

    englands disappearing farmland

    Was
    Government policy already states that best agricultural land should be prioritised for food production [linked to a GOV.UK news story of 15 May 2024]. Planning decisions should reflect this.
    Now
    Government policy already leans this way, though less firmly than it is often quoted. Since 17 August 2026 the National Planning Policy Framework has told planning decisions to take the quality of agricultural land into account and, where significant development of farmland is necessary, to “use areas of poorer quality land where possible”. For the largest schemes, the national policy statement for renewable energy says solar on the best land is “not prohibited” but should be avoided “where possible”. Planning decisions should reflect this.
    Why
    The words presented as policy were not policy. They were a GOV.UK news story’s summary, on 15 May 2024, of a written ministerial statement made the same day (HCWS466), and no policy text we read says the best land “should be prioritised for food production”. Since 17 August 2026 the National Planning Policy Framework (Annex A) has listed that statement as either incorporated into the Framework or no longer up-to-date policy, without saying which. The sentence now quotes what is in force: the Framework’s policy N2(1)(b), and the 2025 renewable energy policy statement (EN-3), in force from 6 January 2026, paragraphs 2.10.21 to 2.10.22.
  4. Description, CPRE key findings, and The Honest Assessment

    solar land use nuclear comparison

    Was
    With 59% of large UK solar farms on productive farmland … 59% are located on productive farmland … CPRE's finding that 59% of large solar farms occupy productive farmland suggests market forces drive developers toward agricultural land
    Now
    With 59% of the land under England’s largest solar farms classed as productive farmland … 59% of the area they cover is productive farmland, counting Grade 3b with the best grades; by farm, 25 of the 38 have some … CPRE’s finding that 59% of the land under England’s largest solar farms is productive farmland suggests market forces drive developers toward agricultural land
    Why
    In CPRE’s report 59% is a share of the LAND, not of the solar farms: “nearly two thirds (59%) of all the land used … is productive farmland”, counting Grade 3b alongside the best and most versatile grades. Counted by farm, 25 of CPRE’s 38 have any such land. The 38 are the farms over 30 MW operating in England, so “UK” in the description was wrong as well.
  5. Cleve Hill: Britain’s Largest Solar Farm

    solar land use nuclear comparison

    Was
    … generates 373 MW from 360 hectares (900 acres). It covers marshland rather than prime farmland, but the scale illustrates the land requirement: nearly 1 hectare per MW.
    Now
    … generates 373 MW. How much land that takes depends on what is counted. The government’s renewable energy planning database records a solar site area of 364 hectares (900 acres); the decision letter gives around 490 hectares for the whole development and around 176 hectares under panels. Almost all of its farmland is Grade 3b, below the best and most versatile grades, on marsh drained since medieval times. On the database’s area the scale is nearly 1 hectare per MW; counting only the land under panels, it is about half that.
    Why
    “360 hectares” did not say which area it meant, and three are on record: 364 hectares of solar site in DESNZ’s Renewable Energy Planning Database, and around 490 hectares of development with around 176 hectares under panels in the Secretary of State’s decision letter of 28 May 2020. The land per MW depends on the choice: about 1.0 hectare on the database’s area, 1.3 on the whole development, and 0.5 on the panels alone. “Marshland rather than prime farmland” was right about quality and loose about land type: the examining authority records arable farmland, almost all Grade 3b, on marsh drained since the medieval period, with 35.1 hectares of grazing marsh inside a 491.2 hectare site.
  6. Description, opening, and the classification paragraphs

    english rivers 16 percent good ecological status

    Was
    Only 16% of English rivers meet good ecological status. … Fifteen years on, the figure hasn't improved. … Sixteen percent of English rivers and canals meet "good ecological status." That figure comes from the Environment Agency's most recent full classification under the Water Framework Directive, completed in 2019. Of 3,900 river and canal water bodies assessed, just 621 achieved high or good status. … The 2019 classification is the most recent full cycle. … The next full classification is due in 2025, with results expected in 2026.
    Now
    In 2019 only 16% of English rivers met good ecological status; the 2026 classification puts it at 15.2%. … Of 3,900 river and canal water bodies assessed, just 620 achieved high or good status. … A newer classification was published on 6 August 2026. Counted from the Environment Agency’s own data, 594 of 3,905 river and canal water bodies are now at high or good status, 15.2%, against 620 on the same count in 2019. … The Agency says the two rounds are “not directly comparable”.
    Why
    The 16% is from the 2019 classification, in a Defra indicator that “will not be updated again”, and the article presented it in the present tense. The Environment Agency published a newer classification on 6 August 2026, so the 2019 figure is now dated and the newer one sits beside it; the title keeps 16% because it is the figure the article was written on. Separately, the 2019 datafile gives 617 good and 3 high, which is 620, not 621.
  7. Opening section, spending sentence

    english rivers 16 percent good ecological status

    Was
    After fifteen years, hundreds of millions in spending, and endless strategy documents, English rivers are no cleaner.
    Now
    After fifteen years and endless strategy documents, English rivers are no cleaner. Money labelled as river restoration runs to hundreds of millions of pounds, not billions: Ofwat allowed water companies £170.0 million for restoration lines over 2020 to 2025, and £231.2 million for 2025 to 2030, both across England and Wales. These are allowances, not money spent.
    Why
    “Hundreds of millions” had no source, and it sat in a sentence about all spending on rivers, where the right order is billions (the water industry’s environment programme for 2025 to 2030 alone is about £22.1 billion for England). For money labelled as restoration it is the right order, so the sentence now says which money is meant and cites Ofwat’s line items for it.
  8. The Price: A 57% Premium, first paragraph

    interconnector economics britain pays france nuclear

    Was
    Britain's wholesale electricity prices are structurally higher than France's, so the cables carry power eastward across the Channel for most of the year.
    Now
    Britain's wholesale electricity prices are structurally higher than France's, so the cables carry power northward across the Channel, from France to Britain, for most of the year.
    Why
    Wrong direction. France lies south of the Channel, and the power flows from France into Britain: in 2024, 20,274 GWh went from France to the UK and 824 GWh the other way (DESNZ Energy Trends, table 5.6.A), and every quarter of 2024 and 2025 was a net import from France. The sentence’s own reasoning (power moves from the cheaper market to the dearer one) says the same; only the compass word was wrong.
  9. System cost chart, key insight

    wind system costs germany energiewende

    Was
    System costs add 40-60% to base LCOE.
    Now
    For wind, system costs add 40-60% to base LCOE.
    Why
    The chart’s own figures put the uplift over base LCOE at 55.9% for wind, 44.2% for gas as then published, and 14.3% for nuclear (12.0 against a base of 10.5). Stated for every source, the range was false for the nuclear row directly above it, and the next sentence half conceded as much by saying nuclear’s base “includes most system costs already”. The article’s description had always applied 40-60% to wind alone, where it holds.
  10. Gas section, conclusion and the system cost chart

    wind system costs germany energiewende

    Was
    Natural gas faces the largest system cost additions, primarily from fuel price volatility and carbon pricing. … Adding fuel volatility and carbon costs to gas's base LCOE of approximately 11 cents (already including current carbon prices) produces a system cost of 15-17 cents per kWh, making gas clearly the most expensive option. … Gas rises from approximately 11 cents to 15-17 cents per kWh (largest increase from carbon costs and fuel volatility) [chart: Carbon Costs +2.5 ct/kWh on the 11.3 base, total 16.3 ct/kWh] … [chart text] Gas faces the largest system cost increase from carbon pricing and fuel volatility. … System costs capture integration expenses not reflected in standard LCOE: grid reinforcement, backup capacity, profile costs (timing mismatch), and externalities like waste management or carbon pricing. … Gas system costs include: Fuel price volatility premium (2022 showed 300%+ price spikes); EU ETS carbon costs at ~€80/tCO2; import dependency security premium.
    Now
    Natural gas’s system cost additions come from fuel price volatility and import dependency. Carbon pricing is a cost too, but it already sits inside gas’s base LCOE. … Adding fuel volatility and import dependency produces a system cost of about 14 cents per kWh, which still makes gas the most expensive of the three. [chart: no carbon component, total 13.8 ct/kWh]
    Why
    Carbon was counted twice. The article said gas’s base of about 11 cents was “already including current carbon prices”, and the chart described its 11.3 base as Fraunhofer ISE’s figure with the CO2 price, then added 2.5 ct/kWh of carbon costs on top. Counted once, gas is 13.8 ct/kWh, not 16.3. It is still the dearest of the three, but its addition (2.5 ct/kWh) is now smaller than wind’s (3.8), so the claim that gas faces the largest system cost addition is withdrawn too.
  11. Opening paragraph and timeline

    fca crypto exodus hostile environment

    Was
    On March 5, 2025, Gemini Trust Company announced it was halting operations in the UK, European Union and Australia. The exchange … cited "evolving regulatory landscapes and strategic realignment" as it cut 25% of its global workforce.
    Now
    On 5 February 2026, Gemini … announced it was exiting the UK, European Union and Australia … Its own explanation was commercial, not regulatory: these markets had "proven hard to win in for various reasons", and "we don’t have the demand in these regions to justify them." Gemini did not blame the regulator.
    Why
    The date was wrong by eleven months: Gemini announced its exit on 5 February 2026, and 5 March 2026 is when UK accounts went withdrawal-only. The quoted reason appears in no source we can find, and the cited Yahoo Finance page no longer exists. Gemini’s own announcement gives commercial reasons. The article used an invented regulatory quote to open its argument, so the correction says plainly that Gemini did not blame the regulator.
  12. Exchange exits section

    fca crypto exodus hostile environment

    Was
    It would not return for two years, re-entering in December 2025 after investing in compliance infrastructure. … Bybit invested heavily in compliance before returning.
    Now
    It would not return for two years, re-entering in December 2025, and even then not under its own FCA authorisation: it operates under the auspices of Archax, a London-based exchange that approves its financial promotions.
    Why
    Neither the dead Cointelegraph link nor the CoinDesk and Bybit sources describe an investment in compliance infrastructure. CoinDesk reports that Bybit returned under Archax’s auspices, which is a different and weaker thing.
  13. The Banking Blockade section

    fca crypto exodus hostile environment

    Was
    One exchange reported approximately £1 billion ($1.2 billion) in declined transactions from UK banks alone. … Brian Armstrong, CEO of Coinbase, stated in January 2026 that major banks view cryptocurrency as an "existential threat to their business". … HM Treasury’s response came on January 28, 2026.
    Now
    One exchange reported nearly $1.4 billion in declined transactions in 2025 … Armstrong said that a top executive at one of the world’s 10 largest banks had told him crypto was now the bank’s "number one priority" and an "existential" issue. … HM Treasury’s response, given to CoinDesk on 27 January 2026.
    Why
    CoinDesk reports "nearly $1.4 billion"; "$1.2 billion" appears in no source. The words in quotation marks were CoinDesk’s headline, not Armstrong’s, and he relayed what one bank executive told him rather than describing "major banks". The Treasury spokesperson spoke to CoinDesk on Tuesday 27 January; the story ran on the 28th.
  14. Investment data paragraph

    fca crypto exodus hostile environment

    Was
    Deals over $100 million halved. Between 2022 and 2023, UK fintech investment had already declined 57%, with average deal size dropping from $29.7 million to $18.4 million.
    Now
    The value of deals over $100 million fell 55%. In 2023, UK fintech investment had already fallen 34%, from $18.7 billion to $12.3 billion.
    Why
    The sentence mixed two datasets. 57% was KPMG’s first-half 2023 comparison; the full-year 2022 to 2023 fall was 34% (KPMG, 6 February 2024). The $29.7m and $18.4m deal sizes are FinTech Global’s Q3 2024 and Q3 2025 figures, not 2022 and 2023. "Halved" described the value of $100m+ deals (down 55%), not their number.
  15. Media coverage section

    deaths per terawatt hour nuclear safety

    Was
    Germany’s decision to close its nuclear plants after Fukushima led to increased coal and gas generation and, by extension, increased air pollution deaths.
    Now
    Germany’s decision to close its nuclear plants after Fukushima led to increased coal-fired generation and, by extension, increased air pollution deaths.
    Why
    The cited Carbon Brief page no longer exists and has no archived copy. The peer-reviewed replacement (Jarvis, Deschenes and Jha, JEEA 2022) finds the lost nuclear output was "offset primarily by increases in coal-fired production and net electricity imports", with most of the social cost from mortality. It does not support gas.
  16. Timeline paragraph

    meadow loss 97 percent nobody talks about

    Was
    Between 1965 and 1985, as nuclear output grew from 26 TWh to 127 TWh [linked to a hostile.eco page that never existed] …
    Now
    Between 1965 and 1985, as nuclear output grew from about 15 TWh to about 59 TWh [linked to DESNZ historical electricity data] …
    Why
    DESNZ’s historical series gives 15.5 TWh in 1965 and 58.9 TWh in 1985. 127 TWh is more than Britain has ever generated from nuclear in a year (the peak was 99.5 TWh in 1998). The internal link pointed at a URL that no page on this site has ever had.
  17. Title and body, and the same figure on the homepage and six other articles

    britain imports energy crisis

    Was
    Britain Imports 40% of Its Energy. Here’s Why. … 40% energy dependency (2023) … 2024: 40% overall energy dependency.
    Now
    Britain Imports 43% of Its Energy. Here’s Why. … 43% energy dependency (2025) … 2024: 43.7% overall energy dependency.
    Why
    DESNZ’s Digest of UK Energy Statistics (table 1.1.3) puts net import dependency at 40.35% in 2023, 43.71% in 2024 and 43.33% in 2025. The article was published in November 2025 with the 2023 figure, and presented it as 2024’s. The same stale 40% was repeated as a present-tense claim across the site.
  18. Air pollution section

    deaths per terawatt hour nuclear safety

    Was
    The 40,000 people who die from air pollution in the UK every year die invisibly, one at a time, their deaths attributed to heart disease, stroke, or lung cancer rather than to the power station emissions that contributed.
    Now
    The 28,000 to 36,000 people who die early from air pollution each year … Power generation is a minority share of that total, roughly 2,500 deaths a year, but it is the share energy policy actually controls.
    Why
    The 40,000 was uncited and sat in the same article as two cited figures it contradicted: 28,000-36,000 for ALL UK air pollution, and 2,500 for power generation specifically. It attached the largest number to the narrowest cause, inverting our own sourcing. The 40,000 traces to the Royal College of Physicians (2016) and refers to all outdoor air pollution, not power generation; the RCP itself now publishes roughly 30,000.
  19. Nuclear accident death tolls

    deaths per terawatt hour nuclear safety

    Was
    Using [official internationally-recognised statistics](world-nuclear.org): Three Mile Island 0 … Chernobyl ~60, up to 4,000 … Fukushima 0 …
    Now
    Each figure is now cited to the body that produced it — Our World in Data for Three Mile Island and Fukushima, the WHO-convened Chernobyl Forum for Chernobyl — and the World Nuclear Association is named as the nuclear industry trade association, with its stated goal of tripling capacity by 2050.
    Why
    Two failures at once. Presenting a trade association as "official internationally-recognised statistics" is exactly the framing we criticise other outlets for, and we apply incentive notes to the RSPB and CPRE while giving our own side none. Worse, the cited page did not support the claims: it does not mention Three Mile Island at all, and gives 46 deaths for Chernobyl rather than the figures we published.
  20. Fukushima section

    deaths per terawatt hour nuclear safety

    Was
    **Radiation deaths**: Zero. No workers or public members died from radiation exposure.
    Now
    **Radiation deaths**: One. In 2018 the Japanese government recognised a single worker’s death from lung cancer as caused by radiation exposure.
    Why
    The sentence contradicted the Our World in Data page cited beside it, which states that one death has been attributed to the disaster. A pro-nuclear claim that overstates the case by rounding one death to zero, against our own source, is the kind of error that costs a site its standing to criticise anyone else.
  21. Raptor mortality section

    wind turbines birds red kites debate

    Was
    Research from the United States has suggested that eagle mortality at wind facilities may be underestimated by a factor of 20.
    Now
    The claim is attributed to the Campaign for the Protection of Moorland Communities, which campaigns for driven-grouse moor management and against the RSPB, and the multiplier is traced to Save the Eagles International, an advocacy group rather than a study.
    Why
    "Research from the United States" concealed a partisan host in the citing sentence itself — the linked post’s own URL ends "yet-rspb-turn-blind-eye". No peer-reviewed source for the 20x multiplier could be found, so the claim is now presented as a claim in circulation rather than as research, and the surviving argument rests on raptor recovery rates instead.
  22. Cost section and opening

    germany energiewende lessons britain ignores

    Was
    According to a 2024 study by Jan Emblemsvåg in the International Journal of Sustainable Energy [linked to foronuclear.org] … the results are in: €696 billion spent … The Fraunhofer Institute criticized aspects of the methodology, arguing the analysis doesn’t adequately account for alternative scenarios.
    Now
    The paper is cited by DOI, Foro Nuclear is named as the Spanish nuclear industry forum and as how we found it, the opening presents €696bn as one prominent accounting rather than a settled result, and Fraunhofer’s actual objection is stated: that the method double-counts subsidies against the expenditure they paid for, leaving €387bn standing and €696bn disputed.
    Why
    Attributing a paper to its author and journal while linking a nuclear industry body concealed the route in the citing sentence. Separately, we had softened a criticism of a source that supports our own argument into something vaguer and easier to dismiss than what Fraunhofer wrote.
  23. Agricultural pollution sections and source list

    english rivers 16 percent good ecological status

    Was
    The NAO found that the EA spent just £8 million per year … The biggest source of pollution receives a fraction of the regulatory attention … The NAO provides the most structurally impartial assessments cited here. [A second NAO claim linked to nao.org.uk, the homepage, with no date.]
    Now
    The 2010 date is stated in the body, the NAO’s own wording is quoted, present-tense inferences are qualified to "when it was last independently audited", and the undated second claim is now cited to the NAO’s April 2025 report on regulating for investment and outcomes in the water sector.
    Why
    A July 2010 audit was carried into present-tense conclusions with its date appearing only in the source list, and a second NAO claim pointed at the organisation’s homepage rather than any report, which made it uncheckable. The 2025 report behind that claim was found and verified, so nothing had to be weakened.
  24. Opening paragraph

    meadow loss 97 percent nobody talks about

    Was
    406 square miles … That's less than the 66,825 hectares we'd need to replace Hinkley Point C with solar panels.
    Now
    That is about 105,000 hectares — more than the 66,825 hectares we'd need to replace Hinkley Point C with solar panels.
    Why
    406 square miles is approximately 105,154 hectares, which is roughly 57% MORE than 66,825 hectares, not less. A unit conversion error reversed the comparison — and the corrected figure weakens the rhetorical point the sentence was making, which is why it is stated plainly rather than quietly dropped.
  25. Title

    solar land use nuclear comparison

    Was
    Solar's Land Problem: Why 382 Square Kilometres Can't Match 1.75
    Now
    Solar's Land Problem: Why 668 Square Kilometres Can't Match 1.75
    Why
    382 is the land-area RATIO between solar and nuclear, not an area. The article body correctly gives the area as 66,825 hectares = 668 km². The title stated the ratio in square kilometres.
  26. Description and conclusion

    offshore wind marine life knowledge gaps

    Was
    Britain's offshore wind capacity could reach 95 GW by 2030 … We can build 95 GW of offshore wind by 2030.
    Now
    Britain's offshore wind pipeline runs to 95 GW … We can build out the 95 GW pipeline.
    Why
    95 GW is the development pipeline, not a 2030 target — as the article's own opening paragraph correctly said ("a pipeline of 95 GW in development"). The description and conclusion coupled the pipeline figure to a 2030 date it does not belong to.
  27. Policy recommendations

    britain imports energy crisis

    Was
    If 95% of fuel is imported, it's import dependency regardless of the carbon accounting gymnastics.
    Now
    If more than 80% of the fuel is imported, it's import dependency regardless of the carbon accounting gymnastics.
    Why
    The 95% figure contradicted this article's own body, which correctly cites ONS for 81% of wood pellets coming from North America. "95% of biomass is imported" is listed explicitly in our editorial guidelines as a known-wrong claim to avoid, and it had been published in the flagship article regardless.
  28. Cumulative cost section

    interconnector economics britain pays france nuclear

    Was
    The lost 64 TWh of British nuclear output, the gap between 100 TWh and 36 TWh, is remarkably close to the 43.7 TWh Britain imported via interconnectors in 2024.
    Now
    British nuclear output fell by about 59 TWh … Britain imported 43.7 TWh … The two are of comparable magnitude — the decline is larger by about a third.
    Why
    Two errors. 64 TWh and 43.7 TWh differ by 46%, which is not "remarkably close". And the 36 TWh endpoint contradicted the 41 TWh (2024) figure used elsewhere in the corpus, inflating the decline.
  29. Cumulative cost section

    interconnector economics britain pays france nuclear

    Was
    Hinkley Point C, at an estimated £33 billion [cited to DESNZ Energy Trends Section 5] for 3.2 GW of capacity …
    Now
    Hinkley Point C, at 3.2 GW of capacity generating roughly 25 TWh per year, looks expensive as a single project on any of the published cost estimates.
    Why
    The £33 billion figure was cited to DESNZ Energy Trends Section 5, which is an electricity-generation statistics release and contains no construction-cost figure. A citation that does not support its claim is worse than no citation, so the figure was removed rather than re-pointed at a source not verified at the time of correction.

Spotted something wrong? badger@hostile.eco.