LL2-A2 — Digest: Annex 2, overview of Late lessons from early warnings vol. 1 (2001)#

Late lessons from early warnings: science, precaution, innovation, EEA Report No 1/2013, pp. 701–716 (PDF 703–718). Full notes: notes/LL2-A2.md.

What it is#

An unsigned, 16-page recap of the fourteen vol. 1 case studies. It has three parts:

• an introduction (p. 701)
• Table A2.1, which summarises the lag between warning and action for each case (p. 702). Its source is Gee (2009), a paper on EMF.
• a short summary and chronology for each case (pp. 703–716)

The parts date from different years, and the annex doesn’t say so. The chronologies stop in 2001 and Table A2.1 is from 2009. The summaries were at least edited for 2013: one cites the 2013 volume’s Chapter 3 (p. 712). There are no panels. Some authors were actors in their own cases: Infante’s 1977 benzene study (p. 705) and Farman’s 1985 ozone-hole paper (p. 708).

Core argument#

The premise is that innovation is outpacing our ability to predict its consequences, while human interventions grow in scale (p. 701). The stated aim is lessons “particularly in terms of providing information and identifying early warnings”. The cases were chosen “where sufficient is now known”, and judged by “the spirit of the times” rather than hindsight (p. 701).

Table A2.1 reports the “years of substantial inaction” between first warning and effective action (p. 702). These are selected rows, the shortest and longest. Across the 13 cases other than fisheries, the median is about 40 years.

Key evidence#

• Asbestos (p. 706). An inspector warned in 1898. By 1930, 66 % of long-term workers had asbestosis. Mesothelioma was identified from 1959. Bans came in 1998–99, and some 400,000 European deaths are expected “over the next three decades”, with no base year given.
• Benzene (p. 705). Exposure limits fell from 100 ppm to 1 ppm. The 1977–87 legal delay is said to have caused more than 200 deaths, and disease was later found even at 1 ppm.
• DES (p. 709). It was prescribed in the US for 18 years after a 1953 trial showed no benefit. The US then withdrew it for use in pregnancy within 7 months of the April 1971 link to cancers in patients’ daughters. Use in pregnancy continued elsewhere until 1985.
• MTBE (p. 712). This substitute for lead belongs to the ether family, whose low biodegradability in water was documented in 1954, and in a textbook, with taste and odour problems, by 1960. An air-quality law increased its use, and one city lost 71 % of its local water supply.
• Ozone (p. 708). The US acted on a “reasonable expectation” of harm in 1977, and the Montreal Protocol followed in 1987. But the replacement chemicals were still ozone-depleting or greenhouse gases.
• Hormones (p. 715). The EU banned them against its own expert committee’s advice, citing the narrow scope of the committee’s assessment. It disbanded the committee and left its findings unpublished. It then faced trade sanctions of about EUR 160 million a year. A 1999 EU scientific committee later found no safe threshold could be defined for six promoters.
• BSE (p. 716). Almost half of the abattoirs checked failed to comply with the offal ban.

The authors’ own lessons#

• Safety margins were underestimated (p. 703).
• Missing units of measurement and weak analytical methods held protection back (pp. 704, 714).
• Industry attacked the science (p. 705).
• Each restriction on PCBs required a “high level of scientific proof” (p. 707).
• Politics failed to push radical alternatives to CFCs (p. 708).
• The idea of a protective placental barrier had to be abandoned (p. 709).
• Independent research lagged behind industry research (p. 710).
• In the 1980s, proof of harm came to be demanded before funding a clean-up (p. 713).
• The lack of precautionary measures and of public information “contributed to” prolonging the BSE crisis and to the “collapse of public trust” (p. 716). The case title names “reassurances” as the mechanism.

Much of the advocacy works through framing: titles, the “inaction” label, entries like “At last” (p. 716).

Main mechanisms#

• The standard of proof sets the pace. Low bars led to fast action; high bars led to delay; courts pushed both ways (pp. 705–716).
• Rules often failed to reduce risk because they were narrow, poorly implemented or relaxed, or because growth elsewhere swamped them.
• Harm migrated to new people and places.
• Knowledge existed but went unused.
• Substitutes brought regrettable new problems.
• Detection limits hid harm for years.
• Single dramatic harms and clear economic losses often triggered action, though not always (the 1952 smog, p. 711).
• Harm that crossed borders required shared science and institutions.
• Latency and persistence locked in harm even after action.

Transferable insights (strength)#

1. Delays of decades between warning and action are common, but their length depends on how “warning” and “action” are defined (p. 702). Moderate.
2. Separate delays in knowing from delays in responding (pp. 705, 709, 712, 714). Strong.
3. The first harm recognised is often not the largest, and harms spread (pp. 704, 706, 707, 709, 711, 714). Strong.
4. Risk estimates tend to rise over time, and “safe” thresholds recede (pp. 704, 705, 715). Moderate.
5. The standard of proof chosen governs timing (pp. 705–715). Moderate–strong.
6. Proving a cause doesn’t guarantee a remedy when the fix is costly (p. 713). Suggestive–moderate.
7. Adopting a rule is not the same as reducing risk (pp. 706, 708, 710, 713, 715, 716). Strong.
8. Substitutes can carry hazards that could have been known in advance (pp. 708, 712). Moderate.
9. Scrutinise claimed benefits, not just risks (pp. 704, 709). Moderate.
10. Knowledge of a hazard often sits unused inside organisations (pp. 705, 707, 712). Moderate.
11. Warnings often come from outside expert circles (pp. 703, 704, 706, 710, 713). Moderate.
12. Monitoring is a precondition for any warning (pp. 704, 707, 713, 714). Moderate–strong.
13. Precaution can be claimed without being practised (pp. 702, 703, 713). Moderate.
14. Vivid harms to organised groups speed up action (pp. 708, 709, 714, 716). But the 1977 US aerosol ban preceded observed harm, and the 1952 London smog drew only “modest remedies” (pp. 708, 711). Suggestive–moderate.
15. Action works, but recovery is slow (pp. 702, 703, 706, 708, 713). Strong.
16. Acting against current expert advice can carry real economic costs, and possibly costs to legitimacy (p. 715). Moderate for cost; suggestive for legitimacy; one case.
17. Reassurance erodes trust (p. 716). Asserted.

Main caveats#

• Selection on outcome. The cases were chosen because harm happened, so there are no false alarms to compare against.
• An inconsistent headline measure. Table A2.1’s Great Lakes figure doesn’t follow from its own entries, and it’s ambiguous which action the PCB figure counts to. Start and end rules vary by row, and the lags for recent cases are truncated.
• Hindsight in the dating. Some “first warnings” only look like warnings with hindsight.
• Benefits and costs are not quantified.
• Frozen at 2001. The chronologies stop in 2001 without saying so.
• Asserted claims. Several claims come without supporting evidence.
• Probable factual slips, e.g. the Beaver Report date and “France 1970” for imposex.
• Hormones. This case is the internal counterpoint to the precautionary framing.