LL1-16 digest — Ch16 Twelve late lessons#

Source: EEA, Late lessons from early warnings: the precautionary principle 1896–2000 (2001), Ch 16, report pp. 168–191 (PDF pp. 168–191). Authors: the editorial team: Harremoës (chair), Gee, MacGarvin, Stirling, Keys, Wynne, Guedes Vaz. Four of them also wrote case chapters. The “initial framing” came from the ESTO project on technological risk and uncertainty, whose final report was written by editor Stirling (pp. 168, 190–191). There are no panels or dissenting voices.

Core argument#

The synthesis chapter returns to the four questions put to every case author: first credible warning, actions taken, costs and benefits, and lessons (p. 168).

The gap between warning and effective action was “many years or decades, and in some cases over a century”. Unequivocal precautionary action stayed scarce even after the principle was articulated (p. 168).

The central diagnosis: in many cases adequate information existed but was not delivered in time, or was discounted, sometimes because of “short-term economic and political interactions” (p. 168). Genuine society-wide ignorance also features in “many” cases (p. 171).

The editors say the case histories “suggest another story” than an EU-versus-North America battle: acceptance of precaution varies between institutions on both sides of the Atlantic (p. 168).

Costs and benefits could not be assessed systematically; there is no credible “single figure”, and a general analysis “lay beyond the scope” of the report (p. 168).

The twelve lessons (pp. 168–169, elaborated pp. 169–182)#

1. Respond to ignorance as well as uncertainty and risk.
2. Monitor and research for early warnings.
3. Seek out blind spots.
4. Reduce interdisciplinary obstacles.
5. Account for real-world conditions.
6. Scrutinise claimed benefits as well as risks.
7. Evaluate alternatives and promote robust, diverse, adaptable technologies.
8. Use lay and local knowledge.
9. Account for different groups’ values.
10. Keep regulators independent of interested parties.
11. Reduce institutional obstacles.
12. Avoid “paralysis by analysis” by acting on reasonable grounds for concern.

Section 16.3 goes explicitly beyond the cases (p. 169) in three areas: - Science (pp. 183–186): Type I/II error bias; how publics and institutions handle ignorance; stakeholders involved from the start. - Innovation (pp. 186–187): alternatives assessment, lock-in, diversity as insurance. - Governance (pp. 187–189): evolving existing tools, early local participation, enlarging responsibility to cover foreseeable-in-principle surprises.

Key evidence (all second-hand from the case chapters)#

• Early warnings: benzene 1897, asbestos 1898, PCB animal tests 1937, DES trial data 1953, Swann Committee 1960s, the CFC mechanism 1974 (pp. 169, 171, 173, 176, 179).
• BSE: reassurance cited “absence of evidence, when no evidence was actually being sought” (p. 172). There was a 17-month lag before MAFF told the Department of Health (p. 180). The Southwood committee self-censored on what was “politically feasible” (p. 179). The US and UK reached different decisions on the same scrapie evidence (p. 181).
• “Controlled use” failed: PCB “closed systems”, MTBE tank leaks, and the 2001 WTO asbestos ruling (pp. 174–175).
• Substitution traps: asbestos-like fibres, HCFCs, MTBE, tall stacks (pp. 173–177).
• Table 16.1 (p. 184): graded levels of proof. Swedish 1973 law sets a high bar for manufacturers to show safety and a low one for regulators to act.

Main mechanisms#

• Institutional ignorance: knowledge exists but does not reach decision-makers (p. 171); the chapter also applies the term to disciplinary capture where available information was disregarded (p. 174). Separately, warnings that did arrive were discounted (p. 168).
• Absence of evidence read as evidence of absence (p. 172). Low-power studies generate false reassurance (p. 184).
• Disciplinary capture: acute versus chronic effects, health versus ecology, veterinary versus human (p. 174).
• The “problem already solved” belief (p. 173).
• Idealised performance assumptions (pp. 174–175).
• Information dependence on the firms being assessed (p. 179). Weak “refutations”, suppression, and “shooting the messenger” (p. 179).
• Institutional friction: dual mandates, political transitions, departmental silos, agencies defending past decisions (pp. 179–181).
• Unpriced harms and lock-in favouring incumbents (pp. 176–177, 186–187).
• Distribution: whether demands for evidence count as “prudence” or “paralysis” depends on who bears costs (p. 182).

Transferable insights (technology-neutral)#

• Strong:
• Many failures were failures to deliver or use existing knowledge, though society-wide ignorance also mattered in many cases (pp. 168, 171, 179).
• Risk, uncertainty and ignorance need different tools (p. 170).
• Reassurance is only as good as the search behind it (pp. 172, 184).
• Real-world use departs from assumed performance (pp. 174–175).
• Standards and burdens of proof are explicit design choices (pp. 170, 184).
• Moderate to strong: disciplinary capture (p. 174); the “problem solved” blind spot and substitutes sharing hazardous properties (pp. 173–174, 177); institutional design driving delay, especially BSE (pp. 179–181).
• Moderate:
• Properties that raise the cost of being wrong (persistence, irreversibility, ubiquity, scale) as proxies for unknown harm (pp. 170–171).
• More research can increase uncertainty; prospective versus retrospective appraisal (pp. 172–173, 181).
• Scrutinise benefits and efficacy; prior “justification” rules are rare and only partly effective (p. 176).
• Acute harms often preceded chronic ones, but chronic harm can arrive with no acute warning (BSE, halocarbons); wildlife as possible “sentinels” (p. 171; sentinel role only suggestive).
• Lay knowledge, used with scrutiny (pp. 177–178).
• Information dependence (p. 179).
• Same evidence, different decisions (pp. 180–181).
• Costly early action plus high uncertainty tends to produce foreseen larger costs (p. 176).
• Suggestive:
• Unpriced harms delay substitutes (p. 177).
• Diversity as insurance (p. 187).
• Public intuition as a rational response to the unfamiliar (p. 178).
• Institutional reassurance-by-research breeds mistrust (p. 185).
• Advisers self-censor towards what is politically “realistic” (Southwood, p. 179; one instance).
• Asserted:
• Lock-in can be arbitrary (p. 187).
• Precaution channels rather than stops innovation (p. 182).
• Participation must come at the framing stage (pp. 186, 188).
• Responsibility should cover foreseeable-in-principle unknowns (p. 189).

Main caveats#

• Advocacy. The chapter is a programme for implementing precaution. The ESTO/STS framework came first, and cases illustrate it rather than being systematically coded (pp. 168–169). Section 16.3 draws heavily on the editors’ own work.
• Selection and hindsight. All fourteen cases are confirmed hazards; none are false alarms. Costs of action are unassessed (p. 168), and Ch16 barely considers the costs of over-precaution (the report’s Conclusions give it one line: Ch 17, p. 194). Hindsight colours “could have been deduced” claims, though the chapter hedges (pp. 170–171, 180).
• Tensions. “Know more” sits against “don’t wait” (acknowledged, p. 181). Lifting restrictions needs research that “genuinely reveals” a concern is unfounded, while maintaining them needs only unresolved uncertainty (pp. 173 versus 181). Public intuition is credited when it proved right (p. 178), but public rejection of irradiated food, Brent Spar dumping and GMOs is cited only as a “cost of failure” of traditional approaches (p. 188), without asking whether it was well founded (outside knowledge: expert bodies judged food irradiation safe).
• Dating issues to verify against case chapters:
• Swann given as 1967, 1968 and 1969.
• “PCB-induced chloracne” in 1899, which predates commercial PCBs (1929).
• The 1974 DES reversal attributed to the farm lobby (probably a court ruling).
• Halocarbon “regulatory neglect”, which understates the late-1970s aerosol bans and the 1987 Montreal timing.
• The ozone discovery framed as “accident” rather than long-term monitoring.
• The ICRP “justification principle” dated to the 1950s (usually dated to its 1977 formalisation).
• Evidence for claims about the public rests on two GMO focus-group studies, one co-authored by an editor (p. 185).
• Fairer than its reputation. It rejects blanket opposition to innovation (p. 169), concedes research can increase uncertainty and that restricting the wrong agent is not precautionary (p. 173), scrutinises lay knowledge (p. 178), credits cost-benefit analysis (p. 187), and calls the choice of response “an essentially political business” (p. 183).