LL2-28 — Ch28 In conclusion (digest)#

Late lessons from early warnings (EEA 2013), pp. 670–684. No named authors: an EEA editorial synthesis that appears (by inference) to draw heavily on Ch 26 (Grandjean), Ch 27 (Gee) and Ch 15 (floods), and explicitly on van den Hove et al. (2012). No panels or dissent. Memorials to Masazumi Harada (p. 683) and Poul Harremoës (p. 684) follow the chapter; annexes and the index come after.

Core argument#

Since 2001, technology adoption has sped up while governance has stayed static. Warnings are “still not being heeded”, harm is found at ever lower exposures, and polluters still do not pay (p. 670). The cases share four features: pathway decisions “made by a few people on behalf of many”, no mechanisms to respond to warnings, misleading prices, and poor accounting across types of capital (p. 671).

The chapter sets out three opportunities (pp. 671–672): 1. Rebalance economic capital against social, human and natural capital through precaution, prevention and polluter-pays. 2. Broaden evidence and public engagement in choosing innovation pathways. 3. Build resilient, adaptive governance.

Six findings follow. The closing section on “governance of innovation and innovation in governance” (p. 680) admits that all of this remains “good intentions” until it is institutionalised.

Key evidence and claims#

• Delay is worsened by three things (p. 672): the moving target (“today’s technology is now safe”), sunk-investment lock-in, and scale that overwhelms monitoring. The barriers are short-termism, the novelty and complexity of the technologies and systems involved, conservative science, evidence-evaluation practices, vested interests and status-quo institutions. Unequal political power is named but declared “well beyond the scope of this report” (p. 672).
• “Harm expansion” (p. 672): asbestos, tobacco, PCBs, lead and radiation all proved harmful in more ways, and at lower doses, than first thought. Often no safe threshold was found.
• False alarms are rare: 4 of 88 alleged cases (p. 673). Ch 2’s caveats are dropped: subjective classification, about a third “jury still out”, and a denominator drawn mostly from a few sources claiming over-regulation (Ch 2, pp. 19–21, 33).
• Error asymmetry under irreversibility (p. 673). A precautionary measure paired with research can be reversed, while missed harm may not be. So tip policy “towards avoiding harm, even at the cost of more false alarms”.
• Monopolies of lead in petrol, asbestos, CFCs and PCBs prolonged harm and created technological, institutional and ideological lock-in. “Cheap” prices stifled substitutes (p. 673).
• Complexity (p. 674): co-causality, timing, low-dose effects, mixtures, susceptibility and system-level effects. So single-cause proof is rare and “inconsistency is to be expected”.
• Epistemic failures (pp. 674–676): “no evidence of harm” misread as “evidence of no harm”; manufactured doubt; high proof demanded for harm, little for safety; research crowded onto known hazards (top ten substances all metals; the “approximately half of all” articles share does not match Ch 26’s figures); premature safety claims; lay warners dismissed; regrettable substitution (HFCs for CFCs); harassment of early warners.
• Risk assessment (p. 677): narrow methods produce “effective denial” of risks outside the frame. Committees diverge on the same evidence without explanation. Funding bias exists.
• Business (p. 678): short-term shareholder focus and “self-serving bias”. A few actors decide for millions (lead in petrol, 1925).
• Research funding (p. 679): about 1% of EU Framework Programme funding for nano-, bio- and ICT product development went to their hazards, 2002–2013.
• Remedies (pp. 679–680): evidence-indexed pollution taxes, natural-capital accounting, pre-funded no-fault compensation and liability bonds, access to private data, FOI and Aarhus, whistleblower protection, and a formal place to analyse value conflicts.
• Memory (p. 680): the “homo-illogical cycle”, in which post-crisis vigilance fades.

Main mechanisms#

Evidentiary asymmetry, with the evidence standard treated as an ethical choice; moving-target technology; capital, monopoly, price and decision lock-in; interest-shaped cognition without bad faith; research-agenda path dependence; frame-induced blind spots; suppression of dissent; regrettable substitution; justice lag; institutional forgetting.

Transferable insights (technology-neutral)#

1. Moving target. Evidence matures more slowly than the technology changes (p. 672). Suggestive: asserted without a worked example.
2. Evidentiary asymmetry. Strong proof is demanded for harm but weak proof accepted for safety; absence of evidence is read as evidence of absence (pp. 674–675). Strong.
3. Harm expansion. Confirmed hazards broaden, and the doses of concern fall (p. 672). Strong for the named agents; moderate as a general prior (selection).
4. Asymmetric error costs. Under irreversibility, favour provisional action plus research (p. 673). Moderate: a sound conditional with untested premises.
5. Dominance breeds lock-in, and artificially cheap incumbents suppress alternatives (p. 673). Moderate.
6. Uncertainty can be manufactured to delay action (p. 675). Strong for documented cases.
7. Research attention is path-dependent, and development funding dwarfs hazard research (pp. 675, 679). Moderate on direction; the magnitude is unsourced.
8. Frames create blind spots, ending in “effective denial” (p. 677). Moderate.
9. Opaque expert divergence and funding bias (p. 677). Moderate–strong.
10. Regrettable substitution (p. 676). Strong.
11. Early warnings from outside the research establishment are dismissed via unexamined expert assumptions (p. 675). Moderate. Per Ch 5 (p. 105), Harada was himself at first among the experts who dismissed the Minamata mother.
12. A few decide for many (pp. 671, 678). Moderate.
13. Justice lags harm by decades (p. 679). Strong; the no-fault remedy is suggestive.
14. Post-crisis vigilance decays (p. 680). Moderate: closely paraphrased from Ch 15 on floods (p. 360); the extension to all hazards is asserted.
15. Diversity of options preserves exit and resilience (p. 673). Moderate.

Main caveats#

• This is advocacy in the institution’s voice, with no panels or dissent.
• The case base is dominated by false negatives identified with hindsight. The chapter names factors for choosing an evidence threshold case by case: system complexity, the level of protection required, and the costs of being wrong (p. 676). It gives no method for weighing them, or for telling true signals from false ones in advance.
• The costs of precaution (forgone benefits, risk-risk trade-offs) are under-weighted in the asymmetry argument.
• The p. 670 innovation claim conflates the Porter-hypothesis literature on environmental regulation with precaution.
• Accuracy problems found by cross-checking: the Fukushima quotations are misattributed (p. 677): the first is from the government committee’s 2011 report, not the NAIIC, and the second is Ch 18’s authors’ own words (pp. 432, 449). The 1% funding figure is unsourced and conflicts with Ch 27’s “about 3 %” (p. 646). The GM “threat to human health” claim (p. 674) goes beyond Ch 19 and is uncited. The US CFC tax example (p. 679) is unsourced. The no-fault incentive claim is in tension with Ch 24 (pp. 599–600). The claim that the most-studied substances account for “approximately half” of articles (p. 675) is not supported by Ch 26’s figures from the same study: about 12 % of links for the top-20 (p. 626). The reference list misdates Volume 1 as covering “1986–2000”.
• Credit where due: it admits power asymmetry is out of scope, accepts more false alarms, counts false-alarm costs and acknowledges progress since 2001.