LL2-02 digest: The precautionary principle and false alarms (Hansen & Tickner)#

EEA 2013, Ch2, pp. 17–45 (text pp. 17–35; Table 2.3 pp. 35–36; references pp. 37–45). No panels. Authors: Hansen (DTU, risk governance) and Tickner (UMass Lowell, chemicals alternatives assessment; a long-standing advocate of precaution [assessment, not stated in the report]).

Core argument. Critics say precaution produces many “false positives”, meaning regulation of minor or non-existent risks. The chapter tests this claim. When LL1 invited industry to submit examples, “No suitable examples emerged” (p. 18). The authors define a regulatory false positive as a case where a regulator acted on a suspected risk and later evidence gives at least “high confidence” (67–95 %, borrowed from the IPCC scale) of no harm (p. 18). They justify the bar as matching the evidence regulators usually demand before acting (p. 18). Only government regulation counts. Market, liability-driven and advisory responses do not (pp. 18–19). They reviewed 88 cases drawn mainly from the critics’ own literature (p. 19), and found: - about a third were real risks (p. 20); - about a third were “jury still out”, where “lack of evidence of harm has been misinterpreted as evidence of safety” (p. 21); - the rest were unregulated alarms, “too narrow a definition of risk” or risk-risk trade-offs (pp. 22–25); - four were genuine false positives (p. 25): Southern corn leaf blight (USDA, 1971), saccharin labelling (1977), swine flu mass immunisation (1976) and FDA reluctance to allow food irradiation (a roughly 20-year approval pause from the late 1960s; the authors call the 1968 bacon withdrawal itself “completely reasonable”, p. 32). Only swine flu and irradiation are analysed in detail; saccharin and SCLB are referred to Hansen (2004) (p. 25).

Key evidence. - Swine flu (pp. 26–29). A Fort Dix outbreak (12 ill, 1 dead) fitted three theories of influenza cycles. The warning was over-weighted: “Perhaps too much faith was placed on the ability of science to foresee” (p. 31). At the advisory committee, the minutes show a stockpiling option was never really discussed. Consultation felt “pro forma”, and the congressional hearings were allegedly staged (pp. 27–28). About 40 million were inoculated before Guillain-Barré cases (107, with 6 deaths) halted the programme, and the flu never returned. Costs included about USD 124 million direct spending plus about USD 100 million from more than 4,100 lawsuits. Gains included vaccine knowledge and a national surveillance system (pp. 28–29). - Food irradiation (pp. 29–31). Approval for irradiated bacon was withdrawn in 1968 after adverse animal findings, and no new approvals followed for about 20 years. Several international and national committees later found it safe (WHO 1981 was the “turning point”). The FDA’s follow-up review found only 5 of more than 400 studies met 1980 standards well enough to stand alone in support of safety (p. 30); approvals resumed from 1983. Consumer resistance still limits use (p. 29). The cyclobutanone genotoxicity concern is answered only by a purity caveat (p. 30). - Nitrites (p. 25). There was no ban. Instead, lower nitrite plus ascorbate, monitoring and technical assistance made bacon nearly nitrosamine-free within a year. The authors call this “smart management of risks, rather than over-regulation”. - Nuclear power (pp. 23–24). Accident concerns were “not irrational”, and the decline in construction was mainly economic (public opposition was “only partly” responsible), so the claim of fear-driven over-regulation “do[es] not bear up to scrutiny”.

Authors’ lessons (pp. 34–35): 1. Be open about disagreement and do not imply consensus. 2. Be transparent about uncertainty and keep alternatives open. 3. Available alternatives minimise the impact of false positives, so fund alternatives assessment. 4. Take particular care with large-scale introductions, because of “unknown unknowns”. 5. Research should supplement regulation, not substitute for it. 6. Precautionary action, even when unnecessary, can spark innovation. 7. Build re-evaluation into decisions.

They conclude that “fear of false positives should not be a rationale for avoiding precautionary actions where warranted” and that false positives are “few and far between as compared to false negatives” (p. 35). They explain the scarcity partly by deliberate doubt-manufacturing and over-claiming by some industry groups and think tanks (pp. 33–34). They also point to institutional and legal biases towards inaction: agency blame-avoidance, judicial review, cost-benefit hurdles and “ossification” (p. 34). They call for “equal weight” to both error types (p. 34).

Main mechanisms (beyond the insights below): a warning that fits prevailing theory, and memories of recent failures, push towards over-reaction (pp. 27, 31). Decisions blamed on “fear” often had economic motives (p. 24). How benefits and risks are distributed shapes whether a technology is accepted (p. 32).

Transferable insights (technology-neutral): 1. Claims of false alarm deserve the same scrutiny as claims of harm (pp. 19–25). Moderate. 2. Definitions and evidential thresholds determine how many errors of each kind are found, and asymmetric bars make one error type rare (pp. 18–19, 33). Strong. 3. Absence of evidence of harm is not evidence of safety (p. 21). Strong as logic. 4. Judge decisions ex ante, not by outcome (pp. 31–32). Moderate. 5. Evidence lags a changing technology (p. 21). Moderate. 6. Suppressed dissent and pre-commitment degrade decisions (pp. 27–28, 34). Moderate to strong. 7. Scale amplifies unknowns, so precaution must also cover the intervention (pp. 28, 35). Moderate. 8. Alternatives and graduated, adaptive measures reduce the cost of error (pp. 25, 32, 35). Moderate for nitrites; suggestive otherwise. 9. Agencies face structural incentives towards inaction (p. 34). Moderate, drawn from the literature. 10. Organised doubt-manufacturing exists (pp. 33–34). Moderate as a general claim; suggestive as the explanation for the scarcity of false positives. 11. Mistaken precaution spurs innovation (pp. 32–33). Suggestive: anecdotal, with no counterfactual. 12. A “false alarm” verdict based on one harm endpoint can ignore other documented harms (“too narrow a definition of risk”) (p. 23). Moderate as logic. 13. Countervailing risks from one response (e.g. a ban) do not show the original concern was unfounded; they argue for a different response (nitrites, p. 25). Moderate, one case.

Main caveats: - The sample is the critics’ showcase cases, not all precautionary decisions. Lieberman and Kwon (ACSH) alone sourced 28 of the 88. So “few false positives” does not establish a low error rate for precaution. The chapter also counts no false negatives. - The evidential bar is asymmetric within the scheme. A false positive needs high confidence of no harm; “real risk” has no stated threshold (the authors defend the bar as matching what regulators demand before acting). Proportionality, the charge of “over-regulation of minor risks”, has no category of its own. - The regulation-only scope excludes harms caused by alarms acting through markets, liability or rhetoric (e.g. MMR). - Errors and questionable classifications: - The chapter wrongly links MMR to thimerosal (p. 22) [external, verify]. Its “10 of the 12 authors” retracted is imprecise (13 authors; 10 of 12 co-authors). - Figure 2.1 (33 “jury still out”, 10 unregulated) and Table 2.3 (32 and 11) disagree by one case (pp. 20, 35–36). - It cites 2001 data to a 1993 source (p. 24), and describes the saccharin label inconsistently (pp. 25, 32). - Tris, Dalkon Shield and BSE classifications look questionable. - Dated evidence (core analysis 2004–07). It omits the 2010 Wakefield retraction, IARC’s 2011 classification of radiofrequency EMF, and 2009 H1N1 [external, verify]. - Signs of advocacy. After conceding irradiation as a false positive, the authors argue it may “cover bad practices” (p. 32). They sidestep the fossil-fuel risk-risk argument on nuclear power. - Hindsight has cut both ways. Some “jury still out” cases later moved towards harm (e.g. BPA). Others moved towards no harm (e.g. GM foods, possibly mobile phones) [external, verify].