Late Lessons, Jensen Huang and AI
Markdown version

LL2-14 hindsight check: Climate change: science and the precautionary principle#

Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013), Chapter 14 (Hartmut Grassl and Bert Metz; Panel 14.1 by Malcolm MacGarvin), report pp. 308–346 (PDF pp. 310–348). Hindsight check carried out on 25 September 2026.

Method note. The web search tool’s session budget was used up before this check began, so all evidence comes from direct fetches of known primary sources and bibliographic databases (Crossref, OpenAlex, the Federal Register API, the UN Treaty Collection, NOAA data files, IPCC and EU pages). Three sources blocked automated access and could not be read directly: - the UNEP Emissions Gap Report 2025: its figures are taken from the UN’s own summary page; - the ICJ advisory opinion text: it is cited through a peer-reviewed commentary (Foster 2025); - EUR-Lex: one EU regulation is cited without re-reading it (flagged where it appears).

This chapter was new in 2013. Climate change was left out of the 2001 volume (p. 308), so Annex 3 of the 2013 report has no update for it.

Overview#

The chapter was written in 2011–12. It makes three kinds of claim: 1. A science-and-targets snapshot, drawn from AR4 (2007): emissions must peak by about 2015, concentrations must stay below about 450 ppm CO2-eq, and mitigation costs are modest. 2. An account of the policy record to 2011: Kyoto compliance, the Clean Development Mechanism (CDM), Copenhagen, Durban, the US, and the effect of attacks on the science. 3. An interpretive thesis. The physical science is settled, so precaution now does little work. Risk management and “green growth” framing drive action.

What happened next divides these claims sharply.

On several points the later evidence is harsher than the chapter was. The CDM and HFC‑23 weaknesses that were “being addressed” largely persisted: atmospheric measurements still show HFC‑23 emissions five times higher than reported in 2021. - The science-confidence claim was strengthened. AR5 (2014) found human influence “extremely likely” to be the dominant cause of warming, and AR6 (2021) found it “unequivocal”. In 2025 the US National Academies judged that the 2009 US endangerment finding “has stood the test of time”. - The quantitative targets were a gloss on AR4 and have been overtaken. - Emissions did not peak by 2015. - CO2 passed 400 ppm in 2015–16 and reached 425.6 ppm (global mean) in 2025. - Total human-caused forcing (3.10 W m⁻² in 2025) is now above AR4’s lowest stabilisation band. - The policy frame has moved from concentration ceilings to temperature goals, carbon budgets and net zero. - The cost optimism was strengthened. In AR6, options costing USD 100/t or less could halve 2019 emissions by 2030, and the benefits of holding warming to 2 °C exceed the costs in most of the literature assessed. - The claim that precaution no longer matters is contested. Precautionary reasoning came back in: - the tipping-point and low-likelihood, high-impact risk framing; - the Paris 1.5 °C goal; - the ICJ’s unanimous 2025 advisory opinion, which treats precaution as an element of due diligence. - The green-growth prediction held up only in part. The paradigm was widely adopted (the European Green Deal as a “growth strategy”), and dozens of countries have decoupled emissions from GDP. But decoupling rates fall far short of what the Paris goals require, and in 2025–26 the US made the largest federal reversal of climate policy yet. - The “manufactured doubt” analysis held up. But the chapter’s evidence that public concern collapsed is selective. The Commission’s own 2011 climate survey showed concern rising.

Claim 1: Peak by 2015 for 2 °C; a 2025 peak means at least 3 °C; mitigation cheap, costs and benefits comparable (p. 319)#

Original claim. “In May 2007 Working Group III … made it clear that global emissions need to peak not later than 2015 to have a 50 % chance of keeping long-term temperature increase below 2 °C … If peaking were delayed to 2025 the world would be committed to at least 3 °C warming in the long term. The report confirmed that ample options for reducing GHG emissions are available at modest costs … costs and benefits of mitigation are broadly comparable, even for the most stringent stabilisation … (i.e. 450 ppm CO2-eq.)” (p. 319).

What the source actually said. The AR4 WG III Summary for Policymakers (Table SPM.5) links peaking years to equilibrium warming, calculated at the best-estimate climate sensitivity: - Category I (445–490 ppm CO2-eq; 2.0–2.4 °C) peaks between 2000 and 2015. - Category II (2.4–2.8 °C) peaks between 2000 and 2020. - Category III (2.8–3.2 °C) peaks between 2010 and 2030.

The chapter’s “50 % chance” and “at least 3 °C if 2025” wording does not appear in the SPM table; it is the authors’ gloss. On benefits, AR4 said early integrated analyses found costs and benefits “broadly comparable in magnitude, but do not as yet permit an unambiguous determination of an emissions pathway or stabilization level where benefits exceed costs”. That is more guarded than “even for 450 ppm” (IPCC AR4 WG III SPM, 2007).

What happened next. - Emissions did not peak by 2015 and have not clearly peaked yet. - Global GHG emissions excluding land use rose to 54.1 Gt CO2-eq in 2025, up 0.7 % on 2024 (EDGAR 2026 report, JRC, September 2026). - Fossil CO2 reached a record 38.1 Gt in 2025, up 1.1 % (Global Carbon Budget 2025, 13 November 2025). - AR6 found that 2010–2019 had the highest decadal average emissions on record (IPCC AR6 WG III SPM B.1, April 2022). - The 2 °C “deadline” was later revised, not confirmed. AR6 (2022) says pathways holding warming to 2 °C (>67 %) with immediate action see emissions peak “between 2020 and at the latest before 2025”. These pathways need GHG cuts of 27 % by 2030 and 63 % by 2050 relative to 2019, and net-zero CO2 in the early 2070s, often followed by net-negative emissions (SPM C.1, C.1.1, C.2). - Part of the change is methodological. AR6 uses transient peak warming and assumes large-scale carbon dioxide removal; AR4 used equilibrium stabilisation. - So the exact AR4 timing-to-temperature mapping in the chapter should not be read as a fixed physical threshold. - Implications of a peak around 2025. AR6: without stronger policies than those in force at the end of 2020, emissions rise beyond 2025, with median warming of 3.2 °C by 2100 (range 2.2–3.5 °C). Current projections that assume emissions level off and then decline slowly are somewhat lower: - Climate Action Tracker (13 November 2025): 2.6 °C under current policies. - UNEP Emissions Gap Report 2025, as summarised by the UN: 2.8 °C under current policies and 2.3–2.5 °C if NDCs are fully implemented.

So the chapter’s direction holds: a peak around 2025 puts 2 °C “likely” out of reach without overshoot. But “at least 3 °C” now looks too pessimistic for current-policy trajectories. - Cost claim: strengthened. - Options costing USD 100 tCO2-eq⁻¹ or less “could reduce global GHG emissions by at least half the 2019 level by 2030”. More than half of that potential costs under USD 20. - Unit costs fell by 85 % for solar and 55 % for wind between 2010 and 2019. - GDP in 2050 is 1.3–2.7 % lower in 2 °C pathways, before counting avoided damages. - “The global economic benefit of limiting warming to 2 °C is reported to exceed the cost of mitigation in most of the assessed literature (medium confidence)” (IPCC AR6 WG III SPM B.4.1, C.12, C.12.1–C.12.2).

Verdict: partly held up. The urgency and cost messages held up, and the cost message was strengthened. The specific peak-year thresholds were a loose paraphrase of AR4 and were superseded by later modelling.

Implications for weight. The general lessons carry full weight: - delay narrows options; - inertia means that waiting for observed harm locks in more harm (pp. 314, 337); - remedies were affordable while action lagged.

The dated thresholds should not be cited as if they were precise physical tipping dates. They were model-dependent readings that later assessments redrew. The pattern itself is instructive: a precise-looking deadline, set by an assessment body, is missed, and the deadline is then moved in the next assessment.

Claim 2: 2 °C requires below about 450 ppm CO2-eq (about 400 ppm CO2) and CO2 cuts of 50–85 % globally (80–95 % in industrialised countries) by 2050 (p. 320)#

Original claim. “Staying within this temperature limit requires greenhouse gas concentrations to remain below about 450 ppm CO2‑eq., which in turn implies a maximum concentration for CO2 alone of about 400 ppm (compared to measured 380 ppm in 2007). To achieve that, CO2 emissions would have to be reduced by 50–85 % from 2000 to 2050 globally, and by 80–95 % in industrialised countries” (p. 320). This matches AR4 Category I: 350–400 ppm CO2, 445–490 ppm CO2-eq, and CO2 in 2050 at 50–85 % below 2000 levels.

What happened next. - The 400 ppm CO2 line was crossed within a few years. - The Mauna Loa annual mean first exceeded 400 ppm in 2015 (401.01 ppm). - The global annual mean first exceeded it in 2016 (403.07 ppm). - The global mean was 425.62 ppm in 2025, and NOAA’s daily global trend estimate was 427.98 ppm on 24 September 2026 (NOAA GML). - The Global Carbon Budget 2025 puts 2025 concentrations 52 % above pre-industrial. - The CO2-eq and forcing thresholds have been passed. - NOAA’s CO2-equivalent for long-lived gases only (no aerosol offset) was already 475 ppm in 2007, 490 ppm in 2012 and 539 ppm in 2024 (NOAA AGGI, updated autumn 2025). - AR4 defined its categories by total radiative forcing; Category I is 2.5–3.0 W m⁻². - Total human-caused effective radiative forcing was 2.72 W m⁻² in 2019 (AR6 WG I SPM A.4.1) and 3.10 W m⁻² in 2025 (Forster et al., Indicators of Global Climate Change 2025, ESSD, June 2026). - Allowing for the difference between RF and ERF, the world is now above AR4’s lowest band. - The frame shifted from concentrations to temperature, budgets and net zero. - The Paris Agreement (adopted 12 December 2015, in force 4 November 2016) set the goal of “well below 2 °C” while “pursuing efforts” towards 1.5 °C. - AR6 frames what is needed as a remaining carbon budget and net-zero CO2. - The remaining 1.5 °C budget (50 %) is about 130 Gt CO2 from the start of 2026, “a little more than 3 years” at 2025 emission rates (Forster et al. 2026). The Global Carbon Budget 2025 gives about 170 Gt, or about four years. - Industrialised-country targets tightened, and some delivery followed. The EU’s “80–95 % by 2050” objective became: - legally binding climate neutrality by 2050 (European Climate Law); - a 55 % net cut by 2030; - an amendment adopted in March 2026 setting a 90 % net cut by 2040, made up of 85 % domestic and “up to 5 % of international carbon credits” (European Commission 2040 target page).

EU net emissions were 37 % below 1990 in 2024 (EEA indicator, 6 November 2025). EDGAR puts EU27 emissions excluding land use about 36 % below 1990 in 2025. Globally, fossil CO2 emissions were 76 % above 1990 in 2025 (EDGAR 2026). - Warming observed. - Human-induced warming reached 1.37 °C in 2025 (Forster et al. 2026). - Copernicus found the 2023–2025 average (1.52 °C in ERA5) to be the first three-year average above 1.5 °C, and projects the 1.5 °C long-term level being crossed around 2029 (Copernicus Global Climate Highlights 2025, 14 January 2026).

Verdict: partly held up. The direction and scale of the cuts held up, and later science made them stricter: net zero rather than 50–85 %. The specific concentration ceilings (450 CO2-eq and 400 CO2) were crossed within a few years. Later modelling, which allows overshoot and net-negative emissions, treats 2 °C as still barely achievable above those ceilings, so the ceilings were not literal limits.

Implications for weight. The chapter’s own point holds and is reinforced: “dangerous” thresholds and 2 °C are value-laden political choices (pp. 322, 327). The thresholds were translated into metrics (concentrations, then budgets, then net-zero dates), and each metric was crossed or redefined in turn.

The transferable pattern is technology-neutral. A numeric safety threshold, set to operationalise a precautionary goal, can be quietly overtaken while the goal is kept in name. Later rules can also shift the burden onto assumed future remedies (net-negative emissions) or offsets (the EU’s up-to-5 % international credits). That pattern is well evidenced.

Claim 3: Emissions rose from about 38 Gt CO2-eq (1990) to about 50 Gt (2010), driven by developing countries; Kyoto deliberately accepted growth to 2012, expecting all countries to act later (p. 325)#

Original claim. Global emissions rose “from about 38 Gt CO2‑equivalent in 1990 to about 50 Gt CO2‑equivalent in 2010 … largely because developing countries increased their emissions … The global increase till 2012 was deliberately accepted (although underestimated at the time) in the design of the Kyoto Protocol … The expectation was that in subsequent periods all countries would strengthen their actions” (p. 325).

What happened next. - The numbers were confirmed. AR6 gives 59 ± 6.6 Gt CO2-eq in 2019, “about 12 % (6.5 GtCO2-eq) higher than in 2010 and 54 % (21 GtCO2-eq) higher than in 1990”. That implies about 38 Gt in 1990 and about 52.5 Gt in 2010 (IPCC AR6 WG III SPM B.1.1). The chapter’s figures are consistent with this; 2010 was slightly higher than it stated. - The drivers were confirmed. EDGAR (2026) attributes growth since 2000 “mainly” to China, India and other emerging economies: - China: 3.69 Gt in 1990, 15.98 Gt in 2025 (29.5 % of the global total); - India: 1.32 Gt in 1990, 4.51 Gt in 2025.

Note that the European Commission says Kyoto covered only about 18 % of global emissions (EC Kyoto page). - The expectation that all countries would act was met in form, with a different architecture. - The Paris Agreement brought in near-universal participation (194 Parties) through self-set NDCs rather than Kyoto-style negotiated caps. - AR6 says Paris “has led to policy development and target-setting at national and sub-national levels” (SPM B.5.1). - But the Kyoto track that was meant to strengthen action withered. The Doha Amendment (the second commitment period, 2013–2020) entered into force only on 31 December 2020, the last day of the period it covered (UN Treaty Collection). - Global emissions kept growing until at least 2025.

Verdict: held up. The facts were accurate. The expectation that all countries would strengthen action later was realised only in the weaker, bottom-up form of Paris, and too late to bend the curve before 2020.

Implications for weight. This is well evidenced. The lesson that a phased design accepting near-term growth (on grounds of equity and “polluter pays”) relies on later tightening that may not come is a strong, general lesson about sequencing and deferred commitment. The chapter’s parenthesis, “underestimated at the time”, also holds: the growth of emerging economies was larger than the architecture assumed.

Claim 4: Most Annex B Parties except Canada and New Zealand likely to meet Kyoto targets; Annex B about 22 % below base year in 2009, partly because of recession; EU‑15 on track for –8 %; former Soviet bloc about 35 % below 1990 (p. 324)#

Original claim. “Other than Canada and New Zealand, most Parties are likely to meet their targets … Emission levels in 2009 of all Annex B Parties … were about 22 % below the base year, in part due to the economic recession of 2008 … The EU‑15 is on track … Collectively [former Soviet and eastern European countries] are at around 35 % below their 1990 level” (p. 324).

What happened next. - Final compliance. The first ex post assessment using final 2008–2012 data found the following (Shishlov, Morel and Bellassen, Climate Policy, 17 August 2016): - Of the 36 countries that fully took part, only nine emitted more than their commitment and had to use flexibility mechanisms. - After those mechanisms, “all Annex B Parties are in compliance”. - New Zealand therefore complied. - Canada escaped non-compliance by withdrawing. It gave notice in December 2011, which the chapter’s Table 14.1 records. - Over-achievement was mostly windfall. The same paper finds aggregate over-achievement of 2.4 Gt CO2-eq a year. Of this, “‘hot-air’ was estimated at 2.2 GtCO2e yr–1”, and land-use accounting removed another 0.4 Gt. Hypothetical participation by the US and Canada would have cut the surplus by 1 Gt a year. None of these factors alone would have caused global non-compliance, “even without use of the 0.3 GtCO2e of annual emissions reductions generated by the Clean Development Mechanism”. The authors flag domestic policy and carbon leakage as unquantified. - The EU over-delivered. The EU‑15 achieved “an overall cut of 11.7 % domestically” against its –8 % target, before sinks and credits (European Commission, Kyoto page). In the second period the EU cut emissions by 31 % by 2020 against a 20 % target. The EU and other buyers pledged not to buy surplus first-period units (“AAUs”) (same source). - The broader verdict. AR6: Kyoto “led to reduced emissions in some countries and was instrumental in building national and international capacity for GHG reporting, accounting and emissions markets”. It adds that at least 18 countries with first-period targets sustained absolute cuts for at least a decade from 2005 (SPM B.5.1).

Verdict: held up. The projection was essentially right. The chapter’s warning that compliance was partly windfall (the UK coal closures, German reunification, the post-Soviet collapse and the recession) is strongly vindicated: formal compliance was dominated by hot air.

Implications for weight. This supports the digest’s transferable insight 11 as strong: formal compliance with a target can diverge sharply from real, policy-driven reductions when baselines, windfalls and accounting rules do the work (pp. 324–326). It also supports a quieter lesson: frameworks that “fail” on outcomes can still build durable measurement and reporting capacity (AR6 B.5.1).

Claim 5: The CDM would deliver about 0.3 Gt a year (2008–12) and 0.7 Gt a year (2013–20), roughly halving Annex I domestic reductions; additionality and HFC‑23 weaknesses “being addressed” (pp. 325–326)#

Original claim. CDM projects “equate to a reduction of about 0.3 Gt CO2‑equivalent per year in the period 2008–2012 and about 0.7 Gt CO2‑equivalent per year from 2013 to 2020. Given their relatively low price, it is very likely that Annex I countries will buy all the CERs … domestic emissions reductions in these countries will be only half of what they would have been without the CDM”. The CDM “is one of the successes of the Kyoto Protocol”. Additionality and HFC‑23 weaknesses “are being addressed in current negotiations on a new international agreement for the period after 2012” (pp. 325–326).

What happened next. - The first-period volume was confirmed, but it did not decide compliance. Shishlov et al. (2016) also cite about 0.3 Gt CO2-eq a year of CDM reductions. But hot air, not the CDM, dominated Annex B’s surplus, so the claim that domestic cuts would be “halved” is moot in aggregate. - The market collapsed. A study for the Commission’s climate directorate (DG CLIMA) describes “the collapse in pricing and the petering out of the classical CDM market in 2011-2012, whereby CER prices below marginal transaction costs make renewal of crediting economically non-viable for most projects” (Cames et al., Öko-Institut, How additional is the CDM?, March 2016). The expectation that Annex I would buy all the credits did not hold. - Additionality turned out worse than the chapter implied. The same study estimates that “85% of the projects covered in this analysis and 73% of the potential 2013-2020 [CER] supply have a low likelihood that emission reductions are additional and are not over-estimated”. Only 2 % of projects had a high likelihood. It concludes that “the large majority of the projects registered and CERs issued under the CDM are not providing real, measurable and additional emission reductions” (Cames et al. 2016). - The HFC‑23 problem was only partly fixed. - The EU restricted the use of HFC‑23 and adipic-acid N2O credits in its emissions trading system from 2013. This was Commission Regulation (EU) No 550/2011, adopted June 2011, before the chapter was published; it was not re-read for this check because EUR-Lex blocked automated access. - HFC‑23 destruction was later required under the Montreal Protocol’s Kigali Amendment from 2020. - Atmospheric measurements then showed that HFC‑23 emissions rose to a record in 2018, although reported measures in China and India should have cut global emissions by 87 % between 2014 and 2017 (Stanley et al., Nature Communications, 21 January 2020). - Emissions fell to 14.0 Gg a year in 2023 but “remained five times higher than reported in 2021” (Adam et al., Communications Earth & Environment, 21 December 2024). - Offsets returned in new form. The EU’s 2040 target, adopted March 2026, allows “up to 5% of international carbon credits” (European Commission). The Paris Article 6 mechanisms carry the CDM’s integrity questions forward.

Verdict: weakened. The 0.3 Gt a year figure held up. The chapter’s reassurances did not: that Annex I would buy all the credits, that the CDM was a success, and that its weaknesses were “being addressed”. Later independent evidence found the integrity problems to be systemic, and measurements show under-delivery on HFC‑23 continuing years later.

Implications for weight. The critical half of the chapter’s CDM discussion carries more weight than the chapter gave it; its optimistic half carries less. Metz had written on the CDM (Metz 2010 is cited as the source here), and the digest notes the chapter’s insider tone. The CDM record is a strong, well-documented case of three mechanisms: - baselines that are counterfactual and cannot be verified invite over-crediting; - a remedy mechanism can create perverse incentives, as with HFC‑23; - self-reported abatement needs independent checking against real measurements, because reports and atmospheric data can diverge for years.

Claim 6: Copenhagen pledges imply 2.5–5 °C, Durban-era pledges 3–4 °C; COP17 launched negotiations for a legally binding post-2020 agreement by 2015; stronger pre-2020 action needed “if the 2 °C limit is to be taken seriously” (p. 328)#

Original claim. Copenhagen-era pledges “imply a 2.5–5 degree trajectory” (p. 328). Durban-era commitments “would lead to a 3–4 °C trajectory, making it impossible to meet the 2 °C limit”. COP17 decided “to start negotiations for a new legally binding agreement that would cover the period after 2020 and to complete negotiations by 2015”. There is “a need for more ambitious action in the period up to 2020 if the 2 °C limit is to be taken seriously”.

What happened next. - The 2015 deadline was met. The Paris Agreement was adopted on 12 December 2015 and entered into force on 4 November 2016. It has 194 Parties (UN Treaty Collection). - It is a binding treaty, but its NDC targets are self-set. - The ICJ’s unanimous advisory opinion (23 July 2025) held that 1.5 °C is “the agreed primary temperature goal” of the Parties. It found that States’ Paris obligations carry a stringent due-diligence standard, and that breaches engage State responsibility (Foster, ICLQ, October 2025, summarising the opinion). - Projected warming fell but stayed well above 2 °C. - Current-policy projections are now about 2.6 °C (Climate Action Tracker, November 2025) to 2.8 °C (UNEP EGR 2025). - Full NDC implementation gives 2.3–2.5 °C (UNEP), or 2.2 °C median with net-zero pledges (CAT). - This is an improvement on the chapter’s 3–4 °C, and cheaper clean energy (AR6 B.4.1) is a large part of the reason. - Pre-2020 action was not strengthened enough. - Emissions rose through the 2010s (AR6 B.1). - The Doha Amendment came into force only on 31 December 2020. - AR6 found that NDCs announced before COP26 made exceeding 1.5 °C “likely” and left 2 °C dependent on “a rapid acceleration of mitigation efforts after 2030” (SPM B.6). - The consequences of that are now observed. - The 2023–2025 average was above 1.5 °C (Copernicus, January 2026). - The 1.5 °C budget is nearly used up (Forster et al. 2026). - UNEP says temporary breaches of 1.5 °C are now “more likely” (UN summary of EGR 2025). - Participation became reversible. The US left Paris twice: - the first withdrawal took effect on 4 November 2020; - it rejoined on 20 January 2021; - the second withdrawal was notified on 27 January 2025 and took effect on 27 January 2026.

On 27 February 2026 the US notified withdrawal from the UNFCCC itself, effective 27 February 2027 (UN Treaty Collection).

Verdict: held up. The pledge-gap diagnosis and the warning about the pre-2020 period were right. The 2015 agreement was delivered as mandated. Later pledges narrowed the gap somewhat, but not enough to keep 2 °C “likely” without overshoot.

Implications for weight. Strong support for the digest’s transferable insight 5: framework-level agreement is easier than binding, effective measures, and the delay lies in implementation (pp. 321–328). A new lesson the chapter could not see: universal participation, won by making commitments self-set, came with reversibility. The largest historical emitter left, rejoined and left again within a decade.

Claim 7: After Obama’s election there was hope US federal GHG policy would change, “but this has not happened”; US net emissions about 15 % above 1990 in 2008 (p. 326)#

Original claim. “After the election of President Obama at the end of 2008 there was hope that federal US GHG emission reduction policy would change, but this has not happened.” US “net emissions were about 15 % above 1990 levels in 2008” (p. 326).

What happened next. - Obama-era executive action (2013–2016). - President’s Climate Action Plan, June 2013. - Clean Power Plan final rule, 80 FR 64661, 23 October 2015. The Supreme Court later limited the approach behind it in West Virginia v. EPA, 30 June 2022. - Paris Agreement participation from 2016.

So the chapter’s statement was accurate for legislation at the time of writing, but federal policy did change through executive action soon after publication. - Reversal and counter-reversal. - The first Paris withdrawal took effect in November 2020. - The Inflation Reduction Act (Public Law 117‑169, 16 August 2022) was the first major federal climate legislation. - Public Law 119‑21 (4 July 2025) amended many of the IRA’s energy provisions; its statutes-at-large citations amend the IRA’s pages. - The second Paris withdrawal took effect in January 2026. - A presidential memorandum of 7 January 2026 directed withdrawal from the UNFCCC and the IPCC. - The formal UNFCCC withdrawal notice followed on 27 February 2026 (White House; UN Treaty Collection). - The endangerment finding. - EPA proposed reconsidering the 2009 finding on 1 August 2025 (90 FR 36288). - It finalised the rescission and repealed all vehicle GHG standards on 18 February 2026 (91 FR 7686, effective 20 April 2026). - The final rule rests on statutory interpretation and the major-questions doctrine. It explicitly is “not finalizing the additional bases for repeal set out in the proposed rule”. - On 17 September 2026, EPA proposed rescinding the power-plant GHG findings as well (91 FR 59002). - Emissions. EDGAR gives US GHG emissions excluding land use of: - 6.22 Gt in 1990; - 7.11 Gt in 2005; - 6.02 Gt in 2025, about 3 % below 1990 and about 15 % below 2005. The 2025 figure was up 2.2 % on 2024, the largest absolute rise among major emitters (EDGAR 2026).

EPA’s own latest final inventory still covers 1990–2022. The 1990–2023 draft went out for comment in January 2025 and, as of this check, no later final edition is listed (EPA inventory page).

Verdict: partly held up. It was accurate at the time for federal legislation. Executive action overtook it within months of publication (2013–2016), and legislation followed in 2022. From 2025–26, federal policy moved decisively the other way. The underlying observation holds, and was reinforced: US federal climate policy is weak and highly reversible.

Implications for weight. The chapter treated the US mainly as a free rider (p. 324 onwards). Hindsight adds a stronger mechanism: in systems where policy rests on executive action or narrow majorities, it can swing sharply, which undermines the long-term signals that the chapter itself says investors need (p. 328). The reporting gap in the official US inventory is also relevant to the chapter’s emphasis on measurement infrastructure (digest insight 2).

Claim 8: Attacks on climate science reduced public concern and political pressure; EU concern fell from 57 % (2007) to 34 % (2011); media coverage declined after 2009; the IPCC’s May 2011 reforms addressed the criticism (p. 336)#

Original claim. “Polls have shown that these attacks and other developments have had an impact on public opinion … In Europe in 2011 only 34 % of citizens consider climate change to be one of their five major concerns, compared to 57 % in 2007 (EC, 2011a) … Media coverage of climate change has also decreased since 2009.” The IPCC’s May 2011 package, including “a conflict of interest policy”, was the response (p. 336).

What happened next. - The chapter’s EU evidence was selective even when it was written. - The source is Special Eurobarometer 365, a survey on environmental attitudes that asks for up to five main environmental worries. - The Commission’s dedicated climate survey from the same year, Special Eurobarometer 372 (fieldwork June 2011, published October 2011), found the reverse. It reported that climate change “remains a key concern … and a greater one than when the last special climate survey was conducted in 2009”. - In SEB 372, 51 % named climate among the world’s most serious problems (up from 47 % in 2009). 68 % called it “very serious” (up from 64 %). - The Commission’s own press release was titled “Survey shows increased public concern”. - Longer run in the EU. The 2025 Eurobarometer found: - 85 % see climate change as a serious problem; - 81 % support EU climate neutrality by 2050; - 84 % agree climate change is caused by human activity; - climate is ranked the joint third most serious world problem; - 49 % find it hard to tell reliable information from disinformation on social media (European Commission, Eurobarometer 2025 summary). - Media. World newspaper coverage did not stay low after 2009. It peaked in 2021. By 2025 it was 14 % below 2024 and 38 % below the 2021 peak, ranking 10th of 22 years (MeCCO 2025 review, University of Colorado). - The effect of attacks on the science: US evidence. Leiserowitz et al. found that “Climategate” “had a significant effect on public beliefs in global warming and trust in scientists”. The loss was concentrated among strongly individualistic or conservative respondents, and the economy, politics, less media attention and cold winters also contributed (American Behavioral Scientist, online September 2012). This supports the mechanism but shows it was one of several factors. - Manufactured doubt: strengthened. - A content analysis of one major oil company’s climate communications from 1977 to 2014 found that internal documents and peer-reviewed papers mostly acknowledged human-caused warming, while 81 % of its public advertorials expressed doubt (Supran and Oreskes, ERL, August 2017). - The same company’s scientists projected warming accurately from 1977 to 2003 (Supran, Rahmstorf and Oreskes, Science, 12 January 2023).

These findings support the chapter’s account of industry-funded doubt (p. 330), which relied mainly on secondary sources. - Institutional reform and its limits. - The IPCC conflict of interest policy was “approved at the Thirty-Fourth Session (Kampala, Uganda, 18–19 November 2011)” and amended in 2014 and 2016 (IPCC policy document). The chapter’s “May 2011” date is right for other parts of the package but slightly off for this element. - AR5 (2013–14) and AR6 (2021–23) were completed without a controversy comparable to “Glaciergate”. - But attacks shifted from the science to the institutions. On 7 January 2026 the US government directed withdrawal from the IPCC. - EPA’s August 2025 proposal to rescind the endangerment finding (90 FR 36288) offered an alternative basis: that “developments cast significant doubt on the reliability of the findings”. It evaluated a draft report from a Department of Energy “Climate Working Group” (May 2025). - A National Academies consensus report, commissioned in response, concluded that the 2009 finding “was accurate, has stood the test of time, and is now reinforced by even stronger evidence” (NASEM, September 2025). - EPA’s February 2026 final rule did not finalise those science-based grounds and rested on statutory interpretation instead (91 FR 7686).

Verdict: partly held up. The mechanism the chapter describes has been reinforced: organised doubt, amplification, and damage to the credibility of assessment bodies. Its specific evidence of a slump in public concern was selectively drawn and was not durable. EU concern stayed high, and media coverage later rose above its 2009 level.

Implications for weight. Treat “manufactured doubt” (digest insight 8) as moderate to strong: well documented through company records, though its measured effect on public opinion is modest and depends on ideology. Treat “authoritative assessment bodies are single points of failure” (insight 9) as reinforced, with a twist. The later threat was less about errors in reports and more about governments withdrawing from, or bypassing, the assessment institution. The poll evidence shows how a chapter can pick the survey that fits its argument; later readers should check which instrument is being cited.

Claim 9: Scientific proof for all five criteria of human-caused climate change is “largely or completely available at a high confidence level”, so precautionary arguments now have “little, if any, effect”; risk-management arguments (feasibility, cost) drive commitments (pp. 336–338)#

Original claim. “For all these criteria scientific proof is by now largely or completely available at a high confidence level. Hence we are far beyond the knowledge level where the precautionary principle would still be needed” (p. 337). “Now precautionary arguments appear to have only little, if any, effect on internationally coordinated climate policy action.” “Arguments based on a risk management framework had a much stronger impact” (p. 338). Countries committed only after “having satisfied themselves that emissions reductions are technically feasible, and … at reasonable costs” (p. 338).

What happened next. - Scientific certainty: strengthened. - AR5 (2014): human influences are “extremely likely to have been the dominant cause of the observed warming since the mid-20th century” (AR5 SYR SPM). - AR6 (2021): “It is unequivocal that human influence has warmed the atmosphere, ocean and land” (WG I SPM A.1). - Human-induced warming was 1.37 °C in 2025 (Forster et al. 2026). - The chapter’s own climate-sensitivity statement is inexact: the AR4 range was a “likely” range, not two standard deviations. AR6 has since narrowed it to a likely 2.5–4 °C with a best estimate of 3 °C (WG I SPM A.4.4). - Precaution came back where uncertainty remains, and it did not disappear. - AR6 states that “low-likelihood outcomes, such as ice-sheet collapse, abrupt ocean circulation changes … cannot be ruled out and are part of risk assessment” (WG I SPM C.3). - The academic Global Tipping Points Report 2025, led by the University of Exeter and not an IPCC assessment, argues that warm-water coral reefs are crossing their thermal tipping point at about 1.4 °C. It says: “If we wait for certainty that tipping points have been crossed before we act, it will be too late.” - Most significantly, the ICJ’s unanimous advisory opinion (23 July 2025) treats “the precautionary principle or approach” as one of the elements that determine the standard of due diligence. Quoting Rio Principle 15, it says “States should also not refrain from or delay taking actions of prevention in the face of scientific uncertainty” (para. 293, as reported in Foster, ICLQ, October 2025). Foster describes the opinion as embracing “the precautionary principle with thoroughness and care”. - Risk management and feasibility: strengthened as descriptions of what drove action. - AR5 recast climate decisions in terms of risk, noting that the risks of mitigation “do not involve the same possibility of severe, widespread and irreversible impacts as risks from climate change” (AR5 SYR SPM 3.2). - AR6 ties the increased deployment to falling costs brought about by “public R&D, funding for demonstration and pilot projects, and demand pull instruments” (SPM B.4.1). - The European Environment Agency’s own European Climate Risk Assessment (11 March 2024) framed 36 climate risks, many at “critical levels”, in risk-management terms. - Tension with the chapter’s own panel. Panel 14.1 warned that expert groups underestimate structural uncertainty (p. 333). Since then: - observed warming in 2023–2025 ran above the assessed human-induced level: a 1.52 °C three-year average in ERA5 (Copernicus 2026), against 1.37 °C of human-induced warming in 2025 (Forster et al. 2026); - the recent rate of warming has been attributed partly to weaker aerosol cooling as well as record GHG emissions (Forster et al. 2026).

This supports the panel’s caution over the main text’s claim that the uncertainty is resolved.

Verdict: partly held up. Two parts were strengthened: the attribution science, and the observation that feasibility and cost drove commitments. The claim that precaution has “little, if any, effect” is contested by later developments. Precaution moved to the unresolved frontier (tipping points, tail risks, the 1.5 °C goal), and it entered international law through the ICJ’s due-diligence reasoning.

Implications for weight. The chapter’s evidence-criteria ladder (insight 4) and its point that action follows confidence in remedies (insight 7) keep moderate weight. But the conclusion that precaution becomes redundant once the core physical question is settled should be treated as time-bound. The later record suggests that precaution shifts to the outcomes still uncertain, such as thresholds, irreversibility and tail risks, and that courts can revive it as a legal standard of conduct. The internal tension the digest flagged (main text versus Panel 14.1) was resolved in the panel’s favour.

Claim 10: Reframing climate action as a co-benefit of development and “green growth” will align economic self-interest with mitigation and make the necessary action more likely; many countries and bodies are adopting the paradigm (pp. 338–339)#

Original claim. “If climate change is integrated into the agenda of development and economic growth, it aligns the benefits for the stakeholders interested in positive economic activities with the benefit of avoiding climate change damage … making it more likely that the necessary action will be taken … climate change risk reduction becomes a co‑benefit of development and economic growth.” “More and more countries see low‑carbon growth … or green growth as a promising way” (pp. 338–339).

What happened next. - The paradigm was widely adopted. - The European Green Deal (2019) is presented as transforming the EU into “a modern, resource-efficient and competitive economy” (European Commission). - The 2025 Eurobarometer reports strong public support for clean-tech industrial policy: 84 % want support for EU firms in clean-tech markets, and 77 % agree that climate damage costs outweigh transition costs. - US climate legislation in 2022 was built around investment incentives rather than carbon pricing (Public Law 117‑169). - Decoupling is real. - “At least 18 countries have sustained GHG emission reductions for longer than 10 years” (AR6 WG III SPM B.3). - The Global Carbon Budget 2025 counts 35 countries that cut emissions while growing their economies, “double the number from a decade ago”. - EU net emissions fell 36 % between 1990 and 2023 “while GDP grew by nearly 70 %” (EEA, November 2025). - But decoupling is too slow for the goal. - A systematic review of 835 studies found examples of absolute long-term decoupling “rare”. It concluded that large, rapid absolute reductions “cannot be achieved through observed decoupling rates” (Haberl et al., ERL, 2020). - For the 11 high-income countries that decoupled in 2013–2019, “at the achieved rates, these countries would on average take more than 220 years to reduce their emissions by 95%”. Decoupling rates would need to rise roughly tenfold (Vogel and Hickel, Lancet Planetary Health, September 2023). - Global emissions reached a record in 2025 (EDGAR 2026; Global Carbon Budget 2025). - Alignment of self-interest proved fragile when costs became salient. - The EU’s new carbon market for buildings and road transport (ETS2) was originally legislated to start in 2027, under Directive (EU) 2023/959; that start date is from background knowledge and was not re-checked. It is now due to become “fully operational in 2028” (European Commission ETS2 page). - The 2040 target includes up to 5 % international credits. - The 2025–26 US reversal shows that co-benefit framing did not prevent a sharp political reversal in the largest historical emitter. - Some of the chapter’s own “co-benefit” questions have been taken up: public-health and energy-security arguments feature prominently in the 2025 Eurobarometer.

Verdict: partly held up. The prediction that the paradigm would spread held up. The causal claim, that it would make the necessary action more likely, has only partial support. Green-growth framing coincided with real decoupling in some countries, but not at the pace required. It also did not insulate policy from backlash once the costs became visible.

Implications for weight. This was 2012 advocacy (as the digest notes), and it should keep the weight of a plausible, partly supported hypothesis, not a demonstrated lesson. The lesson that transfers is a narrower one: pairing a diffuse, delayed benefit with concentrated, near-term economic gains can build coalitions, but the coalition is only as stable as the near-term gains. Where the costs fall on identifiable groups before the gains arrive, the alignment can reverse.

What this means for the section as a whole#

Sources#

IPCC and scientific assessments - IPCC (2007), AR4 Working Group III, Summary for Policymakers (Table SPM.5; cost and benefit statements). https://www.ipcc.ch/site/assets/uploads/2018/02/ar4-wg3-spm-1.pdf - IPCC (2014), AR5 Synthesis Report, Summary for Policymakers. https://www.ipcc.ch/site/assets/uploads/2018/02/AR5_SYR_FINAL_SPM.pdf - IPCC (August 2021), AR6 Working Group I, Summary for Policymakers (A.1, A.4.1, A.4.4, C.3). https://www.ipcc.ch/report/ar6/wg1/chapter/summary-for-policymakers/ - IPCC (April 2022), AR6 Working Group III, Summary for Policymakers (B.1.1, B.3, B.4.1, B.5.1, B.6, C.1–C.2, C.12). https://www.ipcc.ch/report/ar6/wg3/chapter/summary-for-policymakers/ - IPCC Conflict of Interest Policy (approved 34th Session, Kampala, 18–19 November 2011; amended 2014, 2016). https://www.ipcc.ch/site/assets/uploads/2018/09/ipcc-conflict-of-interest-2016.pdf - Forster, P. M. et al. (5 June 2026), “Indicators of Global Climate Change 2025”, Earth System Science Data 18, 3889–3933. https://essd.copernicus.org/articles/18/3889/2026/ (doi:10.5194/essd-18-3889-2026) - National Academies of Sciences, Engineering, and Medicine (September 2025), Effects of Human-Caused Greenhouse Gas Emissions on U.S. Climate, Health, and Welfare. https://www.nationalacademies.org/publications/29239 (doi:10.17226/29239) - European Environment Agency (11 March 2024), European Climate Risk Assessment. https://www.eea.europa.eu/en/analysis/publications/european-climate-risk-assessment - Global Tipping Points Report 2025 (University of Exeter-led; released before COP30, 2025). https://global-tipping-points.org/

Emissions, concentrations, temperature - Global Carbon Budget 2025, press release (13 November 2025). https://globalcarbonbudget.org/fossil-fuel-co2-emissions-hit-record-high-in-2025/ ; full paper: https://essd.copernicus.org/articles/18/3211/2026/ - European Commission JRC, EDGAR, GHG emissions of all world countries — 2026 report (September 2026). https://edgar.jrc.ec.europa.eu/report_2026 ; 2025 report: https://edgar.jrc.ec.europa.eu/report_2025 - NOAA Global Monitoring Laboratory, global annual mean CO2 (accessed 25 September 2026). https://gml.noaa.gov/webdata/ccgg/trends/co2/co2_annmean_gl.txt ; Mauna Loa annual mean: https://gml.noaa.gov/webdata/ccgg/trends/co2/co2_annmean_mlo.txt ; daily global trend: https://gml.noaa.gov/ccgg/trends/gl_trend.html - NOAA, Annual Greenhouse Gas Index (updated autumn 2025). https://gml.noaa.gov/aggi/aggi.html - Copernicus Climate Change Service (14 January 2026, updated 28 January 2026), Global Climate Highlights 2025. https://climate.copernicus.eu/global-climate-highlights-2025 ; PDF: https://climate.copernicus.eu/sites/default/files/custom-uploads/GCH-2025/GCH2025-full-report.pdf - UNEP (November 2025), Emissions Gap Report 2025 (not directly accessible; figures from the UN summary page). https://www.unep.org/resources/emissions-gap-report-2025 ; UN summary: https://www.un.org/en/climatechange/reports - Climate Action Tracker, “Temperatures” (update 13 November 2025). https://climateactiontracker.org/global/temperatures/ - European Environment Agency, “Total greenhouse gas emission trends and projections in Europe” (6 November 2025). https://www.eea.europa.eu/en/analysis/indicators/total-greenhouse-gas-emission-trends

Treaties, Kyoto, CDM, HFC‑23 - UN Treaty Collection, Paris Agreement status (US withdrawals). https://treaties.un.org/Pages/ViewDetails.aspx?src=TREATY&mtdsg_no=XXVII-7-d&chapter=27&clang=_en - UN Treaty Collection, UNFCCC status (US notification of 27 February 2026). https://treaties.un.org/Pages/ViewDetails.aspx?src=TREATY&mtdsg_no=XXVII-7&chapter=27&clang=_en - UN Treaty Collection, Doha Amendment status (entry into force 31 December 2020). https://treaties.un.org/Pages/ViewDetails.aspx?src=TREATY&mtdsg_no=XXVII-7-c&chapter=27&clang=_en - Shishlov, I., Morel, R. and Bellassen, V. (17 August 2016), “Compliance of the Parties to the Kyoto Protocol in the first commitment period”, Climate Policy 16(6): 768–782. https://doi.org/10.1080/14693062.2016.1164658 - European Commission, “Kyoto Protocol” (EU-15 –11.7 % domestically; second-period results; AAU declarations). https://climate.ec.europa.eu/eu-action/international-action-climate-change/kyoto-protocol_en - Cames, M. et al. (March 2016), How additional is the Clean Development Mechanism?, Öko-Institut/INFRAS/SEI study for DG CLIMA. https://climate.ec.europa.eu/system/files/2017-04/clean_dev_mechanism_en.pdf - Commission Regulation (EU) No 550/2011 (June 2011) restricting HFC‑23 and N2O credits in the EU ETS (not re-read for this check; EUR-Lex blocked automated access). https://eur-lex.europa.eu/eli/reg/2011/550/oj - Stanley, K. M. et al. (21 January 2020), “Increase in global emissions of HFC-23 despite near-total expected reductions”, Nature Communications. https://doi.org/10.1038/s41467-019-13899-4 - Adam, B. et al. (21 December 2024), “Emissions of HFC-23 do not reflect commitments made under the Kigali Amendment”, Communications Earth & Environment. https://doi.org/10.1038/s43247-024-01946-y - Foster, C. E. (October 2025), “The 2025 International Court of Justice Advisory Opinion on Obligations of States in respect of Climate Change”, International and Comparative Law Quarterly. https://doi.org/10.1017/S0020589325101292 ; ICJ case page: https://www.icj-cij.org/case/187

EU policy - European Commission, “2040 climate target” (amendment to the European Climate Law adopted March 2026). https://climate.ec.europa.eu/eu-action/climate-strategies-targets/2040-climate-target_en - European Commission, “ETS2: buildings, road transport and additional sectors”. https://climate.ec.europa.eu/eu-action/carbon-markets/ets2-buildings-road-transport-and-additional-sectors_en - European Commission, “European Green Deal”. https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/european-green-deal_en

United States - Executive Office of the President (June 2013), The President’s Climate Action Plan. https://obamawhitehouse.archives.gov/sites/default/files/image/president27sclimateactionplan.pdf - US EPA (23 October 2015), Clean Power Plan final rule, 80 FR 64661. https://www.federalregister.gov/documents/2015/10/23/2015-22842/carbon-pollution-emission-guidelines-for-existing-stationary-sources-electric-utility-generating - US Supreme Court (30 June 2022), West Virginia v. EPA. https://www.supremecourt.gov/opinions/21pdf/20-1530_n758.pdf - Public Law 117‑169 (Inflation Reduction Act, 16 August 2022). https://www.govinfo.gov/app/details/PLAW-117publ169 - Public Law 119‑21 (4 July 2025). https://www.govinfo.gov/app/details/PLAW-119publ21 - US EPA (1 August 2025), proposed rule, “Reconsideration of 2009 Endangerment Finding and Greenhouse Gas Vehicle Standards”, 90 FR 36288. https://www.federalregister.gov/documents/2025/08/01/2025-14572/reconsideration-of-2009-endangerment-finding-and-greenhouse-gas-vehicle-standards - US EPA (18 February 2026), final rule, “Rescission of the Greenhouse Gas Endangerment Finding and Motor Vehicle Greenhouse Gas Emission Standards”, 91 FR 7686. https://www.federalregister.gov/documents/2026/02/18/2026-03157/rescission-of-the-greenhouse-gas-endangerment-finding-and-motor-vehicle-greenhouse-gas-emission - US EPA (17 September 2026), supplemental proposal on power-plant GHG findings, 91 FR 59002. https://www.federalregister.gov/documents/2026/09/17/2026-19072/rescission-of-the-greenhouse-gas-findings-for-fossil-fuel-fired-power-plants-and-repeal-of - The White House (7 January 2026), Presidential Memorandum, “Withdrawing the United States from International Organizations, Conventions, and Treaties that Are Contrary to the Interests of the United States”. https://www.whitehouse.gov/presidential-actions/2026/01/withdrawing-the-united-states-from-international-organizations-conventions-and-treaties-that-are-contrary-to-the-interests-of-the-united-states/ - US EPA, “Inventory of U.S. Greenhouse Gas Emissions and Sinks” (status page, accessed 25 September 2026). https://www.epa.gov/ghgemissions/inventory-us-greenhouse-gas-emissions-and-sinks

Public opinion, media, doubt - European Commission (October 2011), Special Eurobarometer 372, Climate change (fieldwork June 2011). https://climate.ec.europa.eu/document/download/75a64bd4-4542-4ea6-bcf0-a49b6c603208_en?filename=ebs_372_en.pdf - European Commission, “Citizen support for climate action” (2025 Eurobarometer summary and archive of earlier surveys). https://climate.ec.europa.eu/citizens/citizen-support-climate-action_en ; survey page: https://europa.eu/eurobarometer/surveys/detail/3472 - Media and Climate Change Observatory (MeCCO), University of Colorado, “2025 special issue”. https://mecco.colorado.edu/summaries/special_issue_2025.html - Leiserowitz, A. et al. (online 13 September 2012; 2013 issue), “Climategate, Public Opinion, and the Loss of Trust”, American Behavioral Scientist. https://doi.org/10.1177/0002764212458272 - Supran, G. and Oreskes, N. (August 2017), “Assessing ExxonMobil’s climate change communications (1977–2014)”, Environmental Research Letters. https://doi.org/10.1088/1748-9326/aa815f - Supran, G., Rahmstorf, S. and Oreskes, N. (12 January 2023), “Assessing ExxonMobil’s global warming projections”, Science. https://doi.org/10.1126/science.abk0063

Green growth and decoupling - Haberl, H. et al. (27 March 2020), “A systematic review of the evidence on decoupling of GDP, resource use and GHG emissions, part II”, Environmental Research Letters. https://doi.org/10.1088/1748-9326/ab842a - Vogel, J. and Hickel, J. (September 2023), “Is green growth happening? An empirical analysis of achieved versus Paris-compliant CO2–GDP decoupling in high-income countries”, Lancet Planetary Health. https://doi.org/10.1016/S2542-5196(23)00174-2