LL2-24 — Ch24 Protecting early warners and late victims#

Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013), Part D “Costs, justice and innovation”. Report pages 581–606 (PDF pages 583–608). Chapter text runs pp. 581–603; references pp. 603–606.

Reading record. I read the full text extract in order, through the final marker (PDF 608 / p. 606). I checked the opening summary box and all three panels visually in the PDF (pp. 581, 584–585, 591–593, 602–603). The body-text pages would not render as images, so I cross-checked the figures on pp. 596 and 598 with a second, independent text extraction (pypdf); they match. Author details come from the report’s own Annex 1 biographies, the Acknowledgements and the Part D list of panels (p. 563). Items marked [external: verify] come from my background knowledge, not from the report. I include them only where they bear directly on the accuracy or completeness of what the chapter says, and the hindsight pass should confirm them.

Authors and standpoint#

Main text: Carl Cranor (sole author). Annex 1 describes him as distinguished professor of Philosophy and a faculty member of the Environmental Toxicology Graduate Program at the University of California, Riverside. His research is in legal and moral philosophy, and for more than 25 years he has worked on risk, science and law: regulating carcinogens and developmental toxicants, scientific evidence in legal decisions, acceptable risk, susceptible populations, and how to regulate new technologies. The chapter cites six of his own works (1993, 2006, 2007, 2008a, 2008b, 2011). Its central claims about US tort law, the Daubert line of cases, battery and trespass as causes of action, and protecting susceptible subpopulations rest largely on those works.

Genre and stance. This is not a case study. It is a normative legal-policy essay. It proposes two “supporting laws” for a precautionary regime (p. 582):

1. extending whistleblower protection to “early warning scientists and others”;
2. prompt, fair compensation that does not require proving negligence.

The stance is openly pro-precaution and sympathetic to claimants. Its most distinctive move is to recast compensation as a precautionary instrument, whose job is to “minimise any harm as quickly as is institutionally reasonable and to shorten its duration” (p. 587), to deter, and to generate information. The conclusion is modest in tone: the chapter “has explored some ideas for reform … in the hope that this will raise awareness” (p. 603).

Geographic frame. The chapter is US-centred. Its models are the US Whistleblower Protection Act, SEC bounties, US tort, workers’ compensation, the Vaccine Injury Compensation Program and the 9/11 fund. It adds the UK (PIDA, the radiation compensation scheme, asbestos case law), South Africa, New Zealand and (for the whistleblower award) Germany. EU law is almost absent, although this is an EEA report: REACH is mentioned twice in passing (pp. 596, 601) and the Environmental Liability Directive appears only through a citation (p. 600) (see Limitations).

Opening summary box (p. 581). It is unsigned and its authorship is not stated. It sets out the chapter’s aims but makes two claims the body never supports (see Limitations). Treat it as framing, not findings.

Panels (three; none dissenting):

• Panel 24.1, “Better scientific support for early warning scientists?”, by David Gee (pp. 584–585). Gee is at the EEA; Annex 1 calls him the “catalyst, EEA editor and a chapters author” of both Late Lessons volumes. His background is economics and politics, and since 1974 he has worked for trade unions and NGOs, including as a former Director of Friends of the Earth UK. His position is that early-warning scientists are not the same as whistleblowers. What they need is a supportive scientific culture, explicit policies from professional societies, and a European award. This is advocacy, and it complements the chapter.
• Panel 24.2, “Liability for asbestos-related illness: redefining the rules on ‘toxic torts’”, by Owen McIntyre (pp. 591–593). McIntyre is Senior Lecturer in Environmental Law at University College Cork (Annex 1) and sat on the Late Lessons advisory board (Acknowledgements). The panel analyses UK case law on causation and foreseeability in mesothelioma claims, sympathetically. Its base is McIntyre (2004), so in substance it is a 2004 analysis (see Limitations).
• Panel 24.3, “Precautionary assurance bonds for potentially serious environmental risks” (pp. 602–603). The Part D list of panels attributes it to Robert Costanza (p. 563). The panel itself opens with a third-person italic framing (“Robert Costanza has long advocated…”) and ends “Source: Costanza et al., 2010”. It is therefore a summary of a 2010 co-authored article in Solutions, written soon after the Deepwater Horizon spill. Costanza is an ecological economist; I found no Annex 1 biography for him. The panel strongly advocates internalising risk through bonds and reversing the burden of proof.

There is no industry, insurer, regulator or defence-side panel, and no dissent. All three panels push in the same direction as the main text.

Section-by-section notes#

Opening summary box (p. 581)#

• Many Late Lessons chapters show early-warning scientists “harassed for bringing inconvenient truths about impending harm” to public and regulators. There is “also some evidence that young scientists are being discouraged from entering controversial fields for fear of such harassment”. No evidence or citation for this appears anywhere in the chapter.
• Where warnings were ignored and harm followed, prompt and fair compensation has been hard to get. Minamata is the extreme case: “almost fifty years, between 1956 and 2004, before the victims attained equitable levels of compensation and legal recognition of responsibility” (p. 581).
• The box names “nanotechnology, genetically modified crops and mobile phone use” as emerging technologies that “currently can provide broad public benefits but potentially at a cost to small groups of victims”. The body discusses GM plants (pp. 586, 595) and mobile phones (p. 595) but never nanotechnology.
• The key normative claim: “widespread exposure and uncertain science could justify ‘no-fault’ administrative schemes … where the benefit of scientific doubt would be given to victims” (p. 581). The box also flags assurance bonds, and the UK asbestos case law as a pointer to “the future direction of long-tail liability in both environmental damage and personal injury”.

Introduction (p. 582)#

• The foundational claim is that precaution “requires, first and foremost, administrative laws aimed at detecting and addressing risks before they materialise into harms” (p. 582). Ex ante regulation comes first; liability and compensation are supporting tools. Compare p. 601, where bonds are said to resemble “a common rationale for administrative regulation of risks rather than post-market injury suits” (Cranor 1993). Cranor 1993 is cited only there, not on p. 582.
• Two supporting laws follow from the precautionary principle’s emphasis on acting on early warnings and minimising harm:
• whistleblower extensions, so that early warners “should feel free, and be free, to research and report early warnings without the threat of adverse actions”;
• compensation “without having to prove negligence by specific parties” for “foreseeable and surprise events” that cause harm “despite precautionary actions” (p. 582).

24.1 Encouraging and protecting early warning scientists and others (pp. 582–585)#

The problem (p. 582). - Employees in academia, business or government who learn of serious risks may face retaliation: firing, demotion, withheld raises. Private employees are most exposed; for government employees it is “perhaps … less likely”. - Footnote 1 cites the Union of Concerned Scientists (2012) on US government scientists restricted from speaking out, and a current administration initiative to “restore the integrity of science”. - Whistleblowers are often seen as “disloyal, malcontents, grumpy employees, even bitter individuals”. Cranor argues this negative view is exactly why protection is needed (p. 582).

Models surveyed (pp. 582–583; all from Kaplan 2001 unless noted). - US federal employees: the Civil Service Reform Act 1978, as amended by the Whistleblower Protection Act 1989. - It bans threatened or actual “personnel action” against someone who makes a protected disclosure. - A disclosure is protected if the employee “reasonably believes” it shows a legal violation, gross waste or mismanagement, abuse of authority, or “a significant and specific danger to public health or safety”. - It “need not prove ultimately accurate”; good faith and an objectively reasonable belief are enough. - An independent Office of Special Counsel enforces the Act. It can correct adverse actions, prosecute retaliation, and runs a “secure channel” for reports. - To make a prima facie case of retaliation, the employee need only show the disclosure was a “contributing factor”. - US securities, under post-crisis financial reform (Harvard Law Review 2011) [external: this is the Dodd-Frank Act]. - Bounties of 10–30 % of sanctions over USD 1 million, plus protection against retaliation. - Two tiers: stronger protection for reporting directly to the SEC than for reporting internally. - The two tiers “strongly encourage” reporting to the SEC rather than to the company. Cranor calls this double-edged: a “strength — boosting the hand of SEC enforcers, but discouraging local corrections for abuses — or a weakness, undermining internal compliance systems within companies” (p. 583). - The law’s anti-retaliation clause expressly covers “threatening, harassing” and other discrimination (quoting Harvard Law Review 2011). - UK Public Interest Disclosure Act (1998): - covers public and private workers, on a reasonable-belief standard; - covers crimes, miscarriages of justice, and health or safety “likely to be endangered”; - prefers disclosure to the employer or a prescribed agency, and going outside those channels is risky; - encourages employers to set internal procedures, “but employees are not restricted to these” (p. 583), which sits somewhat awkwardly with the preceding point; - has no independent investigating agency. - South Africa, Protected Disclosures Act (2000): modelled on the UK Act, but explicitly lists harassment, intimidation, and adverse or refused references as “occupational detriments”. There is no independent agency, so the whistleblower must go to a court or tribunal. - Emerging democracies have been “slow” to legislate, and free-speech guarantees widen protection.

What makes these laws work (p. 584). - Employees must know about the protections and the channels. - Protections must be strong enough to overcome reluctance to use them. - Organisations must become “receptive, rather than hostile, to employees who ‘rock the boat’”, and this must be communicated “from the top”.

Cranor’s recommended features (p. 584). - Protect public and private employees from adverse actions, “broadly construed”. - Trigger protection on a reasonable belief of either illegality or threats to the environment or public health. - Provide a secure reporting channel. - Make a prima facie case of retaliation easy. - Create an independent office to protect whistleblowers and prosecute retaliation. - Optionally, pay bounties as a share of fines.

Panel 24.1 (Gee), pp. 584–585. - Distinction and examples. - “An early warning scientist is not the same as a whistleblower who reports on wrongdoing”, but Late Lessons cases show early-warning scientists harassed like whistleblowers. - Named examples: Snow (cholera); Selikoff (asbestos); Henderson, Byers, Patterson and Needleman (leaded petrol); “Osakawa” (mercury); “Putzai and Chapella” (GMOs); Schneider (climate); several scientists in the French bees story; and others who wish to remain anonymous. Public servants who were silenced are added from outside the case studies (UCS 2012; Martin 1999, 2008). - Verified in the PDF: the misspellings are in the printed original. They should read Hosokawa, Pusztai and Chapela. The report’s own Minamata chapter spells Hosokawa correctly (and, p. 94 fn, records that he drew comfort from Ibsen’s play). A full-text search of this volume finds Pusztai and Chapela nowhere except this panel; volume 1 has no GMO case study [external: verify]. Martin 1999/2008 are missing from this chapter’s reference list. - Recognition comes late. Recognition that scientists were harassed rises “alongside acceptance that the early warning has been vindicated”, but “the luxury of such hindsight is of little use” to those harassed now (p. 584). - False alarms are an accepted cost. Supporting early warners may mean backing some false alarms. Gee calls this “an acceptable price” for defending the right to warn on “reasonably plausible evidence” (p. 584). - Criteria borrowed from whistleblowing law: - the scientist acted in good faith; - the belief was reasonable, though it need not prove “ultimately accurate”; - the harassment was serious. - What counts as serious harassment: personal attacks “(distinct from scientific criticism)”; being silenced or blocked from publishing; removal from work; loss of funding, contracts, facilities or staff; “unreasonable difficulties” in publishing; accusations of misconduct; being passed over for promotion; threats of legal action (p. 584). - Culture over law. - Better a scientific culture that “explicitly supported challenges to conventional scientific ideas and paradigms” (citing Sunstein 2005 and Mercer 2010 on the value of dissent). - It may be “asking too much” of individual scientists whose life’s work is challenged to respond well. Independent professional associations should instead adopt explicit policies that encourage early warners and defend them if harassed (p. 585). - A European award for a scientist who made a “reasonably credible challenge to conventional science” and was then harassed. - Precedent: rewards to whistleblowers, going back to an 18th-century US law against lead in alcohol production (see Ch. 3). - It would be judged by “an authoritative and independent scientific body of scientists, free from direct bias”, i.e. whose own work would not be challenged by crediting the warning. Gee labels this “intellectual bias” and points to the WTO Appellate Body (2008, the EC–Hormones continued-suspension dispute per the reference list), which held that scientists serving as expert witnesses who had co-authored the JECFA reports under criticism “cannot be considered to be independent and impartial” (p. 585). - Suggested name: “The Henrik Ibsen award”, after An Enemy of the People, the play that sustained the harassed Chisso company doctor. - Existing model: the German Whistleblower Award, given by VDW and IALANA.

Main text (p. 585): early warners “could receive greater support from scientific communities”.

24.2 Providing compensation in a precautionary world (pp. 585–603)#

Why compensation if we have precaution? (pp. 585–586). - “Flawless implementation of precautionary approaches is unattainable.” Compensation is needed for harms that occur despite precaution (p. 585). - Socially accepted risks: society uses lead or cadmium in batteries and electronics despite the hazards. Risk–risk trade-offs: some pesticides are used to mitigate worse dangers. “Even if such harms do not involve wrongful acts … justice arguably demands that society compensate those adversely affected” (pp. 585–586). - Less visible threats: - Developmental exposures cause subtle diseases that are hard to detect in human studies, with “extremely long causal tails” and some rare outcomes. - Paraquat or trichloroethylene (TCE) exposure in early or mid life “may hasten the early onset of Parkinson’s disease, as shown by animal and human studies” (Cranor 2011). - Long-delayed wildlife harms in the Arctic (Cone 2005). - GM plants may cause subtle changes that are not immediately perceptible (p. 586). - Worked example of risk displacement: proposals to use transgenic plants to extract organic mercury from soil and volatilise it as elemental mercury (NRC 2002). This might help locally, but at scale it would likely raise local atmospheric mercury, which is then deposited elsewhere and re-methylated into more toxic organic mercury, “exacerbating existing effects” (p. 586). - An obligation to compensate “can also help deter firms from undertaking harmful activities” (p. 586).

Three compensation rationales (p. 586). - Tort (Priest 2003): deals with “harms inflicted by some identifiable person who was in a position to have prevented the harm”. It shifts costs to responsible parties to create incentives to reduce harm, and awards full damages for deterrence. - Insurance: covers “losses that cannot realistically be prevented”, through private risk pools or government insurance for generalised societal risks such as unemployment or disaster. - Source note: Priest (2003), used for both the tort and insurance definitions, is titled “The Problematic Structure of the September 11th Victim Compensation Fund” (reference list). The chapter uses it only for definitions and does not engage its critique, although it later recommends 9/11-fund features (p. 601). - Reparations (Boxill 2011): presuppose wrongdoing. The best reparations include acknowledgement by the wrongdoer, but “institutionalising acceptance of wrongdoing presents difficulties”, and compensation is more likely if acknowledgement is not required. Footnote 2: Minamata victims “have sought such acknowledgement but have not secured it”. This sits oddly with the summary box’s claim that “legal recognition of responsibility” was attained by 2004 (p. 581). - The chapter admits that the institutions it surveys “do not always carefully distinguish these different dimensions of restitution”; the “central idea” is “some substantial degree of recompense” (p. 586). - The ideal: “ensure that the full social costs of a technology are incorporated into the costs of the activity (removing negative externalities)”, or at least that costs do not fall on “innocent bystanders or the environment” (p. 586). Not all approaches do this, because a technology’s creator may not have acted wrongfully (p. 587).

Five guidance points for compensation in a precautionary world (p. 587): 1. Successful precaution should reduce the need for compensation. 2. Compensation should minimise harm “as quickly as is institutionally reasonable” and shorten its duration. 3. Compensation is needed for long-tailed, less visible, low-probability and subtle consequences. They are “long-tailed in two senses”: highly unlikely, and possibly years in the future. 4. Compensation should be combined with adequate deterrence of negligence or recklessness. 5. Single-payer health systems reduce the medical component, but income loss, suffering, bereavement and other non-medical losses still need compensating.

The models to be assessed are: tort; workers’ compensation; tailored schemes (the US VICP and the UK radiation scheme); New Zealand’s no-fault system; and “flexible assurance bonds” (p. 587). The 9/11 fund is added later.

24.2.1 Tort or personal injury law (pp. 587–590)#

Elements (p. 587). The plaintiff must show breach of a duty (usually negligence), a compensable injury, and both cause-in-fact and proximate cause, each “by the preponderance of the evidence”. US strict liability applies to products, ultra-hazardous activities and trespass.

History (pp. 587–588). - 19th century. Tort was “cramped”. Liability required “moral fault”, and defences shielded railroads and factories. Courts feared that liability would “too greatly burden enterprises” and distrusted sympathetic juries. “Both continue in current debates” (Friedman 1985). - Liberalisation, until “about 1980”. Strict liability for products and ultra-hazardous activities; workers’ compensation in the early 20th century; cause-in-fact rules recognising “the multifactorial nature of causation” (Anderson 1920, Summers v Tice 1948, Sindell 1980); and joint and several liability. - Rollback. Firms, “with support from some legal scholars”, then pushed to roll these doctrines back. Other scholars (Abel 1988) argued that tort had never compensated or deterred well. - Late 1980s to early 1990s. Caps on pain-and-suffering damages, limits on joint liability, and “considerable pressure” for high standards of scientific evidence (p. 588).

Expert evidence (p. 588). - Causal framing: the chapter says the Supreme Court intervened “as a consequence” of the defence pressure for high evidentiary standards (p. 588). This is Cranor’s reading of why Daubert came about, offered without separate evidence. - Daubert (1993) “seemed to liberalise” admissibility. As lower courts applied it, and as Joiner (1997) and Kumho Tire (1999) extended it, it “substantially burdened experts, especially those for plaintiffs”. “Some state and federal circuits are especially onerous”, with Havner (Supreme Court of Texas, 1997) cited as the example. - Key claim: the rules “impartially apply” in principle, but “in reality they asymmetrically hamper plaintiffs, who bear the burden of proof”. - Consequences: - lawyers must invest more up front, so they take only safer cases, which “likely reduces somewhat any deterrent effect”; - courts erected “unscientific barriers” (Cranor 2007, 2008b). - Counter-trend: Milward v Acuity (First Circuit, 2011), covering about one-twelfth of the US. - Scientists may use in court “the same kinds of arguments that they would use in the lab”. - Human epidemiology gets no priority as evidence of cancer causation.

Overall assessment (pp. 588–589). - Barriers for plaintiffs: - they bear the burden of proof; - scientific evidence raises the hurdle; - some federal courts require epidemiology showing a doubling of relative risk, “a further barrier because of the insensitivity of epidemiological research”; - damages caps can be low; - cases take a long time. - Milward as a “typical” case (typical, the chapter says, in involving a single plaintiff and many defendants): filed 2007; experts excluded by the trial judge in 2010; reinstated 2011, after which many of the 22 defendants settled; in 2012 one defendant was still seeking a jury trial. - Class actions are slower. But mature mass torts such as asbestos settle quickly through a “settlement matrix” that grades plaintiffs by exposure, outcome and confounders such as smoking. This makes such settlements resemble compensation schemes. - Low claiming rate. Saks: in clear cases of medical malpractice, “4 % or fewer of injured parties even approach a lawyer”. The chapter generalises from this single malpractice figure to injured parties “rarely” suing where harm is clear (p. 589). - Verdict: US tort law is “a poor legal model for providing rapid and adequate compensation”. It “persists but its achievements fall far short of its goals” (p. 589).

Battery and trespass (pp. 589–590). This is Cranor’s own proposal, from Legally Poisoned (2011), reviewed by Lyndon (2012). - Battery requires a voluntary act intended to make contact (not to cause harm) and an offensive contact. No harm needs to be shown. - “Intentional invasions of one’s body by potentially harmful chemical substances without consent”, which a person would reasonably find offensive, could qualify. - The remedy is nominal damages plus compensation for mental disturbance. - Trespass covers “molecules, particles, or toxic substances” entering property without permission. - Claimed advantages: many more people could sue; it “might better facilitate safety testing and better deter invasions”; cases “would likely be much quicker”; awards would be smaller per person but potentially “substantial” in total. The claims are hedged throughout (“could”, “potentially”, “might”). - The only illustrations are historical (battery’s role in keeping “the King’s peace”; trespass to the person by blasting). The chapter cites no successful precedent for diffuse chemical exposure.

Discovery (pp. 589–590). - Interrogatories, documents and depositions can reveal “hidden data about adverse health effects, decisions made by responsible people that contributed to harm”. - Once publicised, as happened with asbestos, lead and vinyl chloride (Brodeur; Markowitz and Rosner 2002), it can alert public health officials and expose “scientists who have acted without integrity”.

Fit with precaution (p. 590). - Tort that requires proof of harm is “a poor compensatory model” that “would greatly slow efforts to reduce harm and clean up environmental contamination”. - But tort “is not set in stone”. UK developments: - in mesothelioma cases, plaintiffs can recover without ruling out other causes, and each liable defendant pays in full. The chapter says this also applies to “dermatitis caused by coal dust” [external: verify; McGhee concerned brick dust]; - asbestosis damages are apportioned by exposure time; - where a defendant “should have reasonably foreseen a risk of pulmonary injury, not necessarily mesothelioma”, it has been held liable for mesothelioma in nearby residents (the panel, p. 593, gives the fuller test). By extension, firms may be liable for “other types of harm to local residents of a type that emerged after the initial exposure” (p. 590).

Workers’ compensation: origins (p. 590). - Under tort, employees had to prove causation and negligence, and employers “had an incentive to delay” because workers had few resources. - Suits were slow and unpredictable, workers were afraid to sue, and witnesses were scarce (Franklin 1979). - The no-fault bargain: employees “exchange their common law damage actions for smaller but more reliable recoveries”.

Panel 24.2 (McIntyre): UK asbestos case law (pp. 591–593)#

Frame (p. 591). Two issues have been “traditionally … very onerous” for toxic-tort plaintiffs: 1. proving causation; 2. showing that harm appearing long after exposure, when science was less developed, was foreseeable.

“Considerations of justice and injustice played a major role” in how the courts resolved both.

Why foreseeability matters here: the House of Lords in Cambridge Water v Eastern Counties Leather (1994) placed “emphasis … on the requirement of foreseeability in environmental claims” (p. 591). The Court of Appeal’s 1996 ruling reached liability “by employing a broad concept of injury” for foreseeability (p. 591).

Causation (pp. 591–592). - The House of Lords in Fairchild (2002) dropped the “but for” test where medical science cannot say which of several employers caused the mesothelioma. McIntyre says this “effectively creates joint and several liability” (p. 591). [External: verify; see Limitations on Barker and the Compensation Act 2006.] - The alternative, Graham v ReChem (Scotland). Farmers sued an incinerator operator over harm to their cattle. The case ran 896 hours over 198 days, with 80 lay witnesses and 21 expert witnesses. Costs were about GBP 4.5 million for the defence and GBP 1.5 million for legal aid. It was lost on causation because other explanations were possible (p. 591). - Indivisible versus divisible disease. Mesothelioma is “indivisible”, unlike asbestosis, which is cumulative or “divisible”. Indivisibility is what makes causation impossible to prove against successive employers. - The case sequence: - Fairchild at first instance (Feb 2001) refused recovery. - Five months later, Matthews (July 2001) awarded full damages against two of 15 employers. The facts were “very similar”, and the same principal expert gave “substantially similar evidence” (fn 9). - Matthews relied on McGhee (1972) and Bryce (1987): the claimant’s exposure was “sufficient greatly to increase his risk” (pp. 591–592). - Reasoning in the House of Lords (p. 592): - Lord Reid (in McGhee): legal causation follows “the practical way in which the ordinary man’s mind works”. There is no substantial difference between materially increasing the risk and materially contributing to the injury. - Lord Hoffmann: imposing causal requirements that exclude liability would be “morally wrong”. He endorsed the Californian Rutherford test, “substantial factor contributing to the … risk”. - Lord Bingham: the injustice to a duty-breaking employer “is heavily outweighed by the injustice of denying redress to a victim”.

Foreseeability of a “surprise” disease (pp. 591–593). - Latency (p. 592). Mesothelioma can follow very brief exposure. Average latency is 34 years (Munich Re). Latency under ten years is so rare that exposures in the ten years before diagnosis may be excluded as causes (Miller 2002). - Margereson/Hancock v JW Roberts (Court of Appeal, 1996) (p. 593). - The plaintiffs had been exposed as children in Armley, Leeds. - They won although mesothelioma was not “a concept known to medical science” at the time, and although the defendant argued it could not have foreseen the disease. - The defendant never disputed that its dust control was “woefully inadequate”. - Liability arises where a risk of “some pulmonary injury, not necessarily mesothelioma” was foreseeable, and the damage occurred when the defendant “was on actual or constructive notice as to the potential pulmonary damage that exposure to asbestos could cause” (p. 593). The notice limb is a real limit on the rule. - The plaintiffs also pleaded strict liability (Rylands v Fletcher) and nuisance, but only negligence was considered. - The duty extends beyond “the factory wall” to the “extramural neighbour”. - The panel uses bold type to flag the “strong implications for the precautionary principle” (p. 591).

Hindsight (p. 593). Hale LJ, in Shell Tankers v Jeromson (2001): courts must not use later developments “to inflate the knowledge which should have been available earlier”, nor use them “to the contrary effect”. “The fact that other and graver risks emerged later does not detract from the power of what was already known”.

Forecast (p. 593). - Whether this reasoning extends beyond personal injury, to environmental damage generally, “remains to be seen”. Footnote 16 notes that after Page v Smith (1996) it suffices in personal injury that “any personal injury to a ‘primary’ victim is foreseeable”. - If foreseeability attached to classes of damage, the foreseeability test, seen as one reason tort failed “to compensate for historic pollution”, “would effectively be relaxed”. - Several Late Lessons cases show that “much harm arises after the first wave of harm”: PCBs (printed “PCS”), mercury, CFCs, benzene, radiation.

24.2.2 Alternatives to the tort system (pp. 590, 594–601)#

Workers’ compensation (pp. 590, 594–596). Its origins are summarised under 24.2.1 above (p. 590).

Design (p. 594). - Rationale: “The cost of the product should bear the blood of the workingman”. - Funding: employer-funded and often rated on each employer’s safety record. Tax funding “would likely disconnect compensation from current modest incentives” for safety. - Discovery, where allowed, slows the process. - Coverage: originally injuries only, later some diseases, some through separate statutes (e.g. black lung). Domestic, agricultural, casual and some small-business workers are excluded. - Eligibility: injuries must fit prescribed categories and be caused by work (“not weekend soccer”). - Benefits: wage replacement with a cap (e.g. two-thirds), scheduled payments for specific losses, death benefits. Medical and rehabilitative care is “the most effective single part”. - Workers’ compensation is the exclusive remedy, with tort as a fallback. - US variants: a federal scheme for non-military federal employees; the Black-Lung Benefits Act; and laws for railroad and maritime workers, “most of” which pay “only if employers were negligent” (Cornell University Law School 2010) (p. 594). US workers’ compensation is a state matter (p. 590).

Analogues for a precautionary world (pp. 594–595). - The idea is a generic strict-liability scheme for injury, disease or death from environmental hazards. It would also cover environmental damage “caused by products whose causal consequences were missed by prior review”. - For people it could compensate and rehabilitate quickly and so shorten harm. Whether analogues “for environmental damages” can be devised is, Cranor concedes, a “more difficult issue” that “would need to be addressed” (p. 595). - The limitation: workers’ compensation works for injuries that are “immediately cognisable and causally traceable”. Disease categories have to be built as the science develops. Coke oven emissions and lung cancer: “very easy”. Bisphenol A and metabolic syndrome, breast cancer or reproductive effects: “extremely difficult”.

Proposals (p. 595). - Presumptive categories. Build them from what is known, from background information, and from analogies with the same effect caused by other sources. - Compensation tables prepared ahead of proof. Evidence “may or may not be sufficient” to link electromagnetic waves to brain cancer, but the costs of brain cancer are known. So “if it turns out that cell phones do contribute”, compensation tables “could begin to be developed even before the causal evidence is fully sufficient to support a case for compensation”. Read closely, this is about preparing the cost side in advance, conditional on causation later being shown; it is not an explicit proposal to pay before causation is established. The presumptive categories built from plausibility and analogy (previous point) go further. - Learning over time. Decision-makers “learn as they go”. - Faster tables for known hazards. Tables could come faster for “known hazardous but socially important” chemicals. - Monitoring of workers and of communities downwind or downstream. Lead is the example: “highly toxic … with no known safe level of exposure”, yet likely to remain in batteries (Wigle and Lanphear 2005). - Environmental harms. Categories of “plausible or even remote harms” compensable without fault, with monitoring for long-tailed risks. - Analogies for GM organisms, e.g. a GM weed killer, or GM pesticidal plants with wild relatives.

Decline in practice (pp. 595–596). - “Compensation rarely appears to be adequate, some employees engage in fraud, and companies resist”. The public sees the system as “substantially flawed”. - For long-tailed risks, firms may not survive until the harm appears.

US Vaccine Injury Compensation Program (VICP) (pp. 596–597). - Origin. Manufacturers argued they “could no longer manufacture vaccines” on thin margins if the few people injured could sue. Congress created a streamlined scheme, and “companies were immunised from suits” [external: verify; claimants can go to court after the programme, within limits]. - Design. - “On-table” injuries (typical reactions) receive “fairly automatic compensation with minimal evidentiary showings”. - “Off-table” claims go to the Court of Federal Claims. Tort’s preponderance standard applies, but Daubert does not, because the magistrate acts as both judge and jury. - The chapter’s “magistrates” [external: officially special masters] “appear to have a much more sophisticated grasp of scientific evidence” than general federal district judges (Stevens 2001). - Awards: past and future non-reimbursable medical, custodial and rehabilitation costs; up to USD 250 000 for “pain and suffering, lost earnings and/or reasonable lawyer expenses”; death benefits up to USD 250 000 plus reasonable legal fees. - Funding: a USD 0.75 excise tax per vaccine dose, ultimately paid by patients, insurers or government. - Performance claim. Over “the 12 years of its existence” the programme has been less adversarial, cheaper and quicker than tort: “More than 1 500 people have been paid in excess of USD 1.18 billion”, or “about USD 78 000 per plaintiff” (DoJ 2010). - Arithmetic error: USD 1.18 billion ÷ 1 500 is about USD 787 000, not USD 78 000. - “12 years” also implies data from around 2000 [external: VICP began in 1988]. - Cranor’s proposed analogue (pp. 596–597). He calls it “superior” to tort. - Pre-market review (e.g. REACH) and post-market standards cannot protect everyone, because susceptibility varies with “life-stage, genetic heterogeneity, variation in detoxifying enzymes, age, pre-existing illnesses”. Provision is needed for “citizens who are harmed because health standards failed to protect them”. - It would need: 1. tables of expected injury or environmental damage, hard to build for subtle diseases but something that “should be addressed”; 2. proof of exposure, which unlike for vaccines “would likely be subject to numerous disputes”; 3. explicit provision, with fewer procedural hurdles, for long-tailed and subtle effects. - Weakness: the preponderance standard still applies to off-table claims.

September 11 Victim Compensation Fund (pp. 597–598). - Structure. An administrative alternative to tort; suing remained possible but was restricted. - A Special Master set the rules, decided eligibility and set amounts, covering economic and non-economic losses. - Neither liability nor punitive damages were considered, and there was no aggregate cap. - Awards were reduced by payments from other sources, and “strict time limits” applied. Congress wanted awards “individualised” but “not overly disparate”. - Only those physically harmed or killed at the sites and in the immediate aftermath were eligible. - Appeals were internal and final. - Why it moved fast. It passed “in less than one week”, and there was “no debate about the need or justification for compensation”. This is “likely to be quite different” for environmental harms (p. 597). - Lesson drawn. An agency can administer compensation, with courts reviewing only whether it had “substantial evidence”. That standard is deferential “but … not so strict as to force frequent second-guessing” (p. 598). - [External: verify. The fund’s reopening under the Zadroga Act (passed December 2010, signed January 2011) for latent illness among responders and others exposed is not mentioned.]

UK Compensation Scheme for Radiation-Linked Diseases (CSRLD) (pp. 598–599). - Strict liability did not end slow litigation. The Nuclear Installations Act 1965 imposed “strict or absolute liability”, yet litigation stayed “complex, contentious and slow”. The chapter does not say why; that causation disputes were the reason is a reasonable inference from how the scheme works (probability-of-causation tables), not a statement in the text. - Origins. In 1982 BNFL and the trade unions agreed a voluntary scheme, “quicker, more generous”, later extended to other employers and from deaths to illness. Its aims included reducing “expensive litigation for BNFL” as well as for claimants, so the operator also had an interest in it. It was possible “because the causes of radiation-induced cancers were well understood as a result of past experience”. - Eligibility: work for a signatory employer, a recorded radiation dose, membership of a signatory union (usually), and a disease typical of radiation (most cancers qualify). - Probability-of-causation sliding scale, described as “generous toward claimants”:

Tort would require at least 50 %. “Most compensation … is for probabilities of causation below 50 %.” - Results: 106 people received GBP 5.3 million, about GBP 50 000 each. - Verdict. The scheme “achieved its goals”, a claim sourced to the scheme’s own website (p. 598). - Limits (p. 599). - The UK’s National Health Service makes comparison with the US hard. - The scheme “could not be instituted quickly because creating the compensation tables depends so critically on a history of previous diseases”. That is a limitation for “new and subtle diseases, more likely to be typical of contemporary technological risks”. - Incentives to control exposure “appear to be modest”, and how well it deters “is difficult to judge”. Cranor asks whether payouts may sometimes be much smaller than prevention costs; “if so”, payouts alone “would lack deterrence value”. He adds that “other protective mechanisms, such as regulatory rules, inspections” could supply deterrence (p. 599).

New Zealand’s Accident Compensation Corporation (ACC) (pp. 599–600). - Scope. In 1972 New Zealand abolished “almost all” of its tort system. Coverage grew from work and motor accidents to “virtually all accidental injuries” (Schuck 2008), including medical treatment injury, the self-employed, children and the elderly. - Benefits: medical care and rehabilitation; up to 80 % of wages; compensation for lost earning capacity or bodily function; possibly lump sums; payments to survivors. - Reputation: a “mainstay of their social policy” and “one of the simplest in the world for patients to navigate”. - Retrenchment. The cost of medical injuries led to narrower cover, shorter claim periods and no lump sums for pain and suffering. The chapter says fault was reintroduced for medical injury (p. 599). [External: verify. The 2005 treatment-injury reform, discussed in the chapter’s own source (Bismark and Paterson 2006), is usually described as removing fault.] - Funding: from separate accounts by accident type and category of victim (p. 599). The chapter does not link NZ’s weak deterrence to how it is funded. - Exclusions. Illness not caused by an accident, and wilful self-inflicted injury, are excluded, “creating tensions within the system” (Henderson 1981). People without earnings (children, the elderly) get no earnings compensation. - Assessment. - “In large part” “consistent with a precautionary view”, because it is simple and fast; all accidents “(with minor exceptions)” appear eligible. - “A serious shortcoming appears to be the ineligibility of illnesses, at least with respect to the medical compensation system”. The “accident” category “might suggest” a view of illness as “‘natural’ … not induced by human activities”, so the system “might underemphasise” illness from toxic exposures (pp. 599–600). The claims are hedged. [External: verify; some work-related gradual-process disease is covered.] - The scheme “appears to lack deterrents” (p. 600).

Precautionary assurance bonds (pp. 600–601). - The idea. Liability regimes “apply after the damage is done” (EC 2008–2010, the only trace in the text of the EU Environmental Liability Directive material). - Bonds (Costanza and Perrings 1990) make firms “pay in advance for the costs they might inflict on society if they adopted the most harmful method of disposal”. This reverses “the usual presumption of ‘innocence’ over ‘guilt’”. Bottle deposits are the everyday analogy. - Private insurance cannot do this, because future costs cannot be calculated. The choice is therefore between loading costs on the public or on victims later, or on the firm now (p. 600). - Mechanism (p. 600). - An agency sets the bond at its best estimate of worst-case repair costs. - The bond is refundable if the firm shows damages were lower, with the burden of proof on the firm. - It is revised as estimates change, which gives firms “a strong economic incentive to … research the future environmental costs of their activities”. - Bonds are not a substitute for redress: the scheme “would need to be supported by an appropriately quick and adequate compensation system” (p. 600). - Claimed virtues (pp. 600–601): - it tackles “ignorance and uncertainty”; - it funds repair and compensation; - it encourages research before activities start and “help[s] correct underinvestment” in risk research; - it is proportionate to the problem “insofar as this could be determined”; - “no bond, no market”; - it guards against “economically marginal firms” that would go bankrupt rather than pay, since funds are held in trust; - it needs no mandated testing (unlike REACH) but rewards testing; - it may surface long-tailed harms earlier and encourage firms “to reveal risks earlier”; - it deters through money at stake rather than penalties. - Acknowledged downside (p. 601). Bonds may “overly burden new technologies, possibly discouraging investment”. The proposed remedy is to scale bonds to risk and size, growing with the activity. That requires monitoring, which “Governmental agencies may or may not be up to”.

Panel 24.3 (Costanza): Deepwater Horizon and assurance bonds (pp. 602–603)#

• Damage in 2010, written in the present tense:
• at least 20 categories of ecosystem services affected;
• the USD 2.5 billion-a-year Louisiana commercial fishery “almost completely shut down”;
• large tourism losses;
• damage to non-market services (carbon sequestration, hurricane protection by wetlands, cultural values).
• Valuation (Batker et al. 2010): Mississippi River Delta ecosystem services are worth USD 12–47 billion a year, and the asset USD 330 billion to 1.3 trillion, “far more than the total market value of BP (USD 189 billion) before the spill” (p. 602).
• Lessons:
• Natural capital is “far too valuable” to put at such high risk from private interests.
• “We need better (not necessarily more) regulation”.
• The current approach puts “liability [on] the private interests, but with the burden of proof on the public”. The public must prove damages after the fact through lengthy litigation, and total liability is often capped (oil spills, nuclear accidents). “This gives private interests strong incentives to take large risks with public assets” (p. 602).
• Solutions: value natural capital in corporate and public accounts; reverse the burden of proof “from public to private interest”; make firms internalise and monetise risks.
• The bond: large enough to cover worst-case damages and refundable with interest. “The critical feature is that the risk to the public asset is apparent to the private interests in financial terms before the fact” (p. 602).
• Counterfactual:
• What happened was “pretty close to the ‘worst-case’”.
• Estimated potential damages: USD 34–670 billion. A review panel might settle on USD 50 billion, “several orders of magnitude riskier than the USD 50 million liability limit previously in force”. [External: verify. The Oil Pollution Act 1990 cap for offshore facilities is usually cited as USD 75 million plus removal costs.]
• The firm would then either not drill or cut risk, for example with “the acoustic blowout preventer costing a mere USD 500 000 which they failed to install”. [External: verify. Whether that device would have prevented the blowout is contested.]
• “Very strong economic incentives … to find creative ways to reduce the risks rather than … cutting corners” (pp. 602–603).
• Conclusion: “Our entire society is taking far too many risks with public assets”; shifting the financial burden to those who benefit would set the right incentives (p. 603).

24.2.3 Desirable features of compensation systems (pp. 601, 603)#

• Quick identification and minimisation of disease and damage from poorly understood technologies would “promote precautionary approaches better than slower systems” (p. 601).
• The different systems’ rationales pull against each other, and fit will vary by country (e.g. universal health care, existing schemes like New Zealand’s).
• The best systems have “nearly automatic provisions for many classes of injuries (ideally all)”. Tables for poorly understood technologies are hard, “but decision-makers must do the best they can” (p. 601).
• Some problems would be eased by single-payer universal health care, “such as many countries in Europe have”, and by a New Zealand-style no-fault accident scheme “with extensions to diseases and the environment” (p. 601), in effect a proposal to extend that model.
• Recommendation: an administrative arrangement with some features of the 9/11 fund (administrator discretion, speed), with “constraints on decision-maker discretion to ensure justice between applicants” (p. 601).
• The CSRLD’s sub-50 % compensation is “notable” but hard to replicate because it “rests on a long, well understood history of radiation-caused diseases with prior victims” (p. 601).
• Funding (p. 603): a fee on covered activities, as in workers’ compensation and VICP, is attractive to budget-constrained governments. Contributions tied to a firm’s safety record give “some modest deterrence feedback”. Workers’ compensation does this explicitly; VICP has no “deterrence surcharge”; for CSRLD it is “not clear” (p. 603).
• Health care does not exhaust needs: lost earning capacity, long-term care and families must be covered.
• Bonds, or insurance where it exists, paid up front fund “anticipatory research” and compensation “so that society does not have to pay”. “When companies put up their own money, this provides incentives for better research on risks … and deters carelessness or recklessness” (p. 603).

24.3 Conclusion (p. 603)#

• Early warners “have sometimes been discouraged in the past or actually lost positions or suffered various kinds of losses”. They “often bring forth useful and timely knowledge and therefore need to be encouraged and not harmed”.
• Laws should discourage retaliation, and “justice requires rectification” when it happens.
• Compensation has often been hard to get “if warnings are not heeded and damage results, or if damage results even when there were not warnings”.
• The chapter offers “plausible improvements” and aims to raise awareness.

References (pp. 603–606)#

• Cited in the text but missing from the list: Martin 1999/2008 (Panel 24.1); Munich Re (fn 12, though the footnote itself names the Employers Liability Handbook). “Buckley, supra, n. 1” (Panel 24.2, fn 12) points to a footnote that does not exist in this chapter, a leftover from McIntyre’s 2004 article.
• Listed but never cited:
• Ad-hoc Industry Natural Resource Damage Group (2010) on insurance and financial security under the EU Environmental Liability Directive;
• CEA 2007, 2008, 2009 (European insurers on the insurability of environmental liability and the ELD);
• EBTP 2009; FERMA 2010;
• EU Habitats Directive (1992) and Birds Directive (2009), which define protected-species and habitat damage under the ELD;
• Guth 2009; Gillette and Krier 1999; Honoré 1995; Leigh and Wakeford 2001 (radiation litigation); Berger 1997 (eliminating general causation); Prosser and Keeton 1984.
• Inference: a section on the EU Environmental Liability Directive, financial security and insurability (and perhaps on radiation litigation) was probably drafted and then cut. This matters: the European regulatory material a European reader would most need is what is missing.
• Minor citation errors: “Kumho Tire v. Firestone” (should be v. Carmichael, as the text has it); “Costanza and Perrins” in Panel 24.3; “Wooley” in Panel 24.2 against “Woolley” in the list; Brodeur dated 1983 in the text but 1985 in the list, and listed as “Brodeur, D.” [external: Paul Brodeur]; PIDA cited as House of Commons 1998 in the text and 2009 in the list; Saks dated 2000 but in volume 140 of the Pennsylvania Law Review [external: that volume is 1992]; “Friedman, F.” (Lawrence Friedman).

Case timeline#

Not a case study. What follows is a chronology of the embedded cases and legal and institutional developments, as the chapter gives them, with the lags it implies.

Early warnings and responses, as far as the chapter documents them. The chapter itself documents no warning-to-action sequence. Its evidence for harassment of early warners is the list of names in Panel 24.1, which refers the reader back to other chapters. For the lag between warning and compensation, the only concrete figures are Minamata (1956–2004), the asbestos latency and litigation sequence, and Milward.

The authors’ own lessons and conclusions#

Lessons Cranor derives from the evidence he presents#

1. US tort law is a poor vehicle for rapid, adequate compensation of those harmed by others’ products or actions (p. 589). The reasons: burden of proof, hurdles for scientific evidence, doubling-of-risk rules, damages caps, delay, very low claiming rates (4 % or fewer, in clear medical malpractice), and a post-Daubert admissibility regime that asymmetrically burdens plaintiffs (pp. 588–589).
2. Tort’s emphasis on proven harm makes it ill-suited to a precautionary system; it would “greatly slow efforts to reduce harm and clean up” (p. 590).
3. Tort can change. The UK asbestos cases show judges relaxing causation and foreseeability on grounds of justice (p. 590; Panel 24.2).
4. Workers’ compensation works for obvious, traceable injuries but not for subtle, latent disease. Its “apparent attractiveness has been reduced in practice”: inadequate awards, fraud, employer resistance (pp. 595–596).
5. Compensation tables depend on a history of prior victims. This is why the CSRLD works for radiation, and why the model struggles for new technologies (pp. 599, 601).
6. Compensation does not reliably deter. The VICP’s per-dose excise tax carries no “deterrence surcharge” (pp. 596, 603). New Zealand’s scheme “appears to lack deterrents” (p. 600; no reason is given). The CSRLD’s incentives are “modest” and “difficult to judge”, and payouts smaller than prevention costs “would lack deterrence value” (posed as a question, p. 599). General taxation would “disconnect” workers’ compensation from safety incentives (p. 594).
7. Whistleblower laws work only if employees know about them, trust them, and the organisation’s culture changes from the top (p. 584, from Kaplan).
8. The two-tier design of reporting channels involves trade-offs between external enforcement and internal correction (p. 583).

Recommendations and advocacy#

• Extend whistleblower law to early-warning scientists and others, public and private. The listed features are: a reasonable-belief standard covering threats to health and environment; a secure channel; an easy prima facie case; an independent protective and prosecuting office; optional bounties (p. 584).
• Scientific-community support (Gee): a culture supporting challenges to paradigms; explicit policies from professional associations; a European “Henrik Ibsen award” judged by an independent body “free from direct bias” (what the panel then calls “intellectual bias”); accept some false alarms as the price (pp. 584–585).
• Five guidance principles for compensation (p. 587).
• Presumptive compensation categories built from existing knowledge and analogy, with compensation tables prepared before the causal evidence is “fully sufficient” (the mobile-phone example); monitoring of workers and nearby communities; categories of “plausible or even remote” environmental harm compensable without fault (p. 595).
• A VICP-style scheme for regulated substances, to catch people failed by health standards, particularly susceptible subgroups; on-table and off-table tracks (pp. 596–597).
• An administrative compensation body with 9/11-fund-like discretion, bounded by rules; judicial review on a “substantial evidence” standard (pp. 597–598, 601).
• Extend New Zealand-style no-fault cover to diseases and the environment (p. 601).
• Consider CSRLD-style probabilistic compensation below 50 % (p. 601), in line with the summary box’s “benefit of scientific doubt … given to victims” (p. 581).
• Fee-based funding, experience-rated on safety records (p. 603).
• Battery and trespass claims for chemical body-burden invasions without consent (pp. 589–590).
• Assurance bonds sized to worst-case costs, with the burden of proof on the firm and scaled with growth; “no bond, no market” (pp. 600–601, 603).
• Costanza (Panel 24.3): “better (not necessarily more) regulation”; full valuation of natural capital in accounts; reverse the burden of proof from public to private (p. 602).

Mechanisms and dynamics#

1. Proof rules decide who bears the cost of scientific uncertainty. This is the chapter’s deepest thread. - In tort the claimant bears the burden of proof on every element (p. 587). - Gatekeeping rules that look neutral “asymmetrically hamper” the party that carries the burden (p. 588). - Demanding human epidemiology, or a doubled relative risk, turns the insensitivity of epidemiology into a legal barrier (p. 588). - Where science cannot settle causation, which is typical for rare, subtle, latent harms (p. 586), the loss stays with the victim by default.

Every alternative the chapter favours shifts that default: - presumptive compensation categories and on-table analogues (pp. 595–597); - partial awards below the 50 % probability threshold (p. 598); - relaxed causation rules for indivisible disease (pp. 591–592); - foreseeability of a class of harm rather than the specific harm (p. 593); - reversing the burden of proof, as bonds do (pp. 600, 602).

Lord Bingham’s weighing of the two injustices (p. 592) states the normative logic outright.

2. Time, latency and the vanishing defendant. Harms take a long time to show and a long time to redress: - “extremely long causal tails” (p. 586); - mesothelioma’s average latency of 34 years (p. 592); - litigation lasting years (Milward, p. 588; Graham v ReChem, p. 591); - about 48 years to compensation at Minamata (p. 581).

Cranor stresses two consequences. Harm lasts longer when redress is slow (p. 587). And the firm may be gone by the time harm appears (pp. 596, 601). Bonds collect resources “when firms are optimistic about their activities” and, if held in trust, keep them available “even if the firms are no longer in business” (p. 601).

3. The “first wave” problem. - With asbestos, the first recognised harm (lung disease) was followed by a graver “surprise” disease that was unknown to medicine while people were being exposed (pp. 591–593). - McIntyre generalises that “much harm arises after the first wave” (p. 593). - The legal response, making foreseeability of a class of harm sufficient, turns a scientific regularity into a rule of responsibility. Being on “actual or constructive notice” that something causes one kind of harm puts you on notice for the class of harm (p. 593).

4. Hindsight runs both ways. - Hale LJ (p. 593) warns against inflating what could have been known in the past, and equally against discounting what was known because graver risks emerged later. - A related point about contingency, not hindsight as such: Fairchild and Matthews reached opposite first-instance results on causation five months apart, on “very similar facts” and “substantially similar evidence” from the same principal expert (pp. 591–592, fn 9). Judgments about causation are more contingent than they look.

5. How warnings get suppressed. The chapter and Panel 24.1 identify distinct channels: - employer retaliation (p. 582); - stigmatising labels such as “disloyal, malcontents” (p. 582); - cultures hostile to those who “rock the boat”, whose change “must be communicated from the top” (p. 584, quoting Kaplan); - peer and professional sanctions outside the employment relationship: attacks in the literature, publication barriers, loss of funding and staff, misconduct charges, legal threats (p. 584); - the intellectual bias of established scientists whose life’s work is challenged (p. 585); - recognition that arrives only once a warning is vindicated (p. 584).

Whistleblower law is aimed mainly at the first; Kaplan’s conditions (awareness, effective protection, cultural change led from the top) address the second and third (p. 584). Gee’s turn to culture and awards (p. 585) implicitly concedes that the rest lie beyond the law’s reach. (This grouping is an interpretation; the chapter does not set out the channels as a typology.)

6. The design of reporting channels steers knowledge. - Bounties and two-tier protection send reports to external enforcers, possibly at the expense of internal correction (p. 583). - The UK prefers internal or prescribed channels and penalises going outside them (p. 583). - Channel design determines whether knowledge of a hazard produces enforcement or quiet correction inside the organisation.

7. Whether compensation deters depends on who pays and how. - Employer levies rated on safety records give “modest” safety incentives (p. 594); general taxation would “disconnect” them (p. 594). - The VICP’s per-dose levy falls on patients and insurers and carries no “deterrence surcharge” (pp. 596, 603). - New Zealand’s scheme “appears to lack deterrents” (p. 600). - Payouts smaller than the cost of prevention “would lack deterrence value”, though other mechanisms (regulation, inspections) may compensate (p. 599). - Tort’s own deterrent effect is “likely” reduced “somewhat” by the up-front cost of expert admissibility fights (p. 588). The low claiming rate (4 % or fewer in clear malpractice, p. 589) points the same way, although the chapter does not itself link it to deterrence. - Liability caps give “strong incentives to take large risks with public assets” (p. 602). - Bonds make risk “apparent … in financial terms before the fact” (p. 602) and reward firms that research and reduce their risks (pp. 600–601). The chapter says they “help correct underinvestment in research” on adverse effects (p. 600); reading this as countering an incentive to stay ignorant is an interpretation.

8. Who gains and who bears the harm. - Broad benefits, with costs falling on “small groups of victims” (p. 581). - Risks society knowingly accepts, such as lead and cadmium in batteries or pesticide risk–risk trade-offs, where “justice arguably demands” compensation even without wrongdoing (pp. 585–586). - Susceptible subgroups left unprotected by standards set for the population as a whole (p. 596). - Private interests putting public, largely non-market assets at risk, while the public carries the burden of proof and liability caps limit recovery (p. 602). - Institutions move costs around: single-payer health systems absorb medical costs (pp. 587, 598–599), and VICP levies fall on patients and insurers (p. 596).

9. Compensation as an industrial bargain and a lever on innovation. - The VICP was a deal to keep a socially valuable product in production. Manufacturers threatened to leave the market; Congress traded immunity from suit for a no-fault fund (p. 596). That also dilutes deterrence (p. 603). - Bonds might “overly burden new technologies”. The answer offered is to scale bonds to risk and size, which depends on the state’s capacity to monitor (p. 601). - Only partly considered: the scaling remedy explicitly allows “lesser bonds” for “low probability, relatively contained risks when enterprises are small” (p. 601). What is not discussed is the competitive advantage bonds give well-capitalised incumbents. The chapter’s approval of excluding “economically marginal firms” (“no bond, no market”, p. 601) points that way.

10. Doctrine under organised pressure, and the competence of institutions. - Fears of “burden[ing] enterprises” and distrust of juries have persisted since the 19th century (p. 587). - Plaintiff-friendly doctrine grew until about 1980, followed by a defence-led rollback “with support from some legal scholars” (p. 588). The chapter reads this as interest-driven. It acknowledges the aim of “high standards for scientific evidence” and that experts need “good scientific foundations” (p. 588), but does not engage on its merits the argument that the Daubert rules screen out unreliable science. - Administrative alternatives stand or fall on competence. Compare special masters with a “more sophisticated grasp of scientific evidence” (p. 596) against agencies that “may or may not be up to the task” of monitoring bonds (p. 601).

11. How categories and framing allocate cost. - New Zealand’s “accident” category, which Cranor suggests “might” trace to a view of illness as “‘natural’”, excludes non-accidental illness and so “might underemphasise” toxic illness (p. 600). - Whether a disease counts as “indivisible” or “divisible” decides whether each defendant pays in full or in proportion (pp. 590–591). A biological model is converted directly into a cost-allocation rule. - “The factory wall” (p. 593) shows how legal duties once followed property lines rather than exposure pathways.

12. Moral salience governs how fast redress comes. The 9/11 fund passed in under a week with “no debate” (p. 597). Cranor contrasts this with contested environmental harms. The binding constraint is often social agreement that there are victims at all, not just scientific proof.

13. Remediation can displace harm. Using transgenic plants to volatilise mercury (p. 586) would clean up one site but lead to deposition and re-methylation elsewhere. Fixes need a systems view.

14. Discovery and monitoring force information into the open. - Litigation discovery pries loose what producers knew internally (p. 590), so curbing litigation also curbs public learning. - A scheme that must pay for harm acquires an interest in detecting it early. Cranor ties compensation to surveillance of workers and of communities downwind or downstream (p. 595), an underdeveloped link between redress and early warning.

Transferable insights (technology-neutral)#

1. Rules about the burden and standard of proof decide who absorbs scientific uncertainty. Where harms are rare, subtle or latent, a proof-of-harm default leaves the loss with those harmed. - Evidence: pp. 586–589, 591. - Strength: strong as a structural point (it follows from how the rules work, illustrated by Graham v ReChem and Fairchild at first instance). Moderate on how large the practical effect is, since the chapter gives no outcome data.

2. Evidentiary rules that are formally neutral can operate asymmetrically against whichever party carries the burden, especially when the demanded evidence type (e.g. large human studies) is insensitive to the harm in question. - Evidence: p. 588. - Strength: moderate. Well argued and grounded in Cranor’s extensive prior work, but contested by those who see such rules as filtering unreliable testimony; the chapter does not present that view on its merits.

3. The first recognised harm of a hazard is often not the only or the worst. Later “surprise” harms of a different kind follow, so responsibility can reasonably attach to foreseeing a class of harm rather than the specific outcome, provided the actor was on actual or constructive notice of the known harm. - Evidence: pp. 590–593. - Strength: strong for asbestos (court findings). Moderate as a generalisation: McIntyre asserts the pattern for PCBs, mercury, CFCs, benzene and radiation (p. 593), but the evidence lies in other chapters.

4. Hindsight distorts in both directions. Assessments of past decisions must neither inflate what was knowable nor discount what was known because graver risks emerged later. - Evidence: p. 593 (Hale LJ). - Strength: strong as a methodological principle (judicial reasoning), though not an empirical finding.

5. Warnings are suppressed through several distinct channels: employer retaliation, stigmatising framing of warners, hostile organisational culture, peer sanctions, and the intellectual investment of incumbent experts. Legal protection reaches only some of them. - Evidence: pp. 582, 584–585. - Strength: moderate. The typology is well reasoned, and the harassment list rests on named cases in other chapters. There is no systematic data, and the case selection favours vindicated warners.

6. People whose own prior work is under challenge are structurally conflicted as judges of the challenge. Independent review of dissenting evidence needs reviewers not invested in the challenged conclusions. - Evidence: p. 585 (WTO Appellate Body 2008). - Strength: moderate. Grounded in a legal ruling and intuitively robust. One example, and the same logic applies to the warners themselves, which the panel does not note.

7. Protection for those who raise warnings should turn on good faith and reasonable belief, not on being proved right later. Supporting warners means accepting some false alarms. - Evidence: pp. 583–584. - Strength: asserted (normative). It rests on established whistleblower-law precedent, but the costs of false alarms are never weighed.

8. Recognition of those who raised early warnings tends to arrive only after vindication, which is too late to protect them. Protection mechanisms have to act before the science settles. - Evidence: p. 584. - Strength: suggestive. Plausible and consistent with the named cases, but survivorship bias is strong: false alarms are not examined.

9. How reporting channels are designed decides where knowledge of hazards flows (to external enforcers or internal correction), with trade-offs between enforcement strength and organisational self-correction. - Evidence: p. 583. - Strength: moderate. A clearly reasoned trade-off with no outcome evidence.

10. Long latency combined with firms that may not last means ex post liability often cannot be collected. Ex ante financial security, secured while actors are optimistic and solvent, addresses defendants who cannot pay.

    • Evidence: pp. 596, 600–601.
    • Strength: moderate. Sound incentive logic. The chapter itself offers no empirical test of bonds for novel hazards.

11. Capped liability plus an after-the-fact burden of proof on the public encourages actors to take risks with shared, largely non-market assets that are far larger than is socially optimal.

    • Evidence: p. 602.
    • Strength: moderate. A standard economic argument, illustrated by one case. The specific counterfactual (a cheap device would have averted the harm) is contested.

12. Fast, categorical (“table-based”) redress depends on an accumulated history of harm. The schemes that work best exist for old, well-characterised hazards, which is the opposite of the situation with novel ones.

    • Evidence: pp. 595, 599, 601.
    • Strength: strong as a structural dilemma. The chapter states it candidly, and it follows from how the CSRLD and VICP are built.

13. Compensation deters only if its cost falls on the risk-creator and is linked to that actor’s own safety record. Pooled levies, general taxation and payouts below prevention costs decouple redress from prevention.

    • Evidence: pp. 594, 596, 599–600, 603.
    • Strength: strong conceptually; moderate empirically (the chapter itself calls the incentives “modest”).

14. Removing the fault requirement does not by itself remove slow, contested litigation. Where causation is uncertain, strict liability can still leave claimants facing long disputes.

    • Evidence: p. 598 (1965 Act: strict liability, yet litigation “complex, contentious and slow”). Graham v ReChem (p. 591) shows causation as a bottleneck, but in a negligence and nuisance case, not a no-fault one.
    • Strength: suggestive. One example, stated in a sentence. The chapter does not say that causation was why the 1965 Act litigation was slow; that is an inference from how the CSRLD then worked.

15. Probabilistic or proportional compensation can replace all-or-nothing thresholds, giving claimants partial benefit of the scientific doubt.

    • Evidence: pp. 581, 598.
    • Strength: moderate. A single working scheme whose success is self-reported (CSRLD 2010). Transferability is limited by the need for dose records and good dose–response models (p. 601).

16. How fast redress comes depends on social consensus that there are victims. Harms with obvious, agreed victims get fast schemes; diffuse, contested harms do not.

    • Evidence: p. 597.
    • Strength: suggestive. One contrast, drawn by the author.

17. Liability shields paired with no-fault compensation can keep socially valuable but risky products on the market, at the cost of weakened deterrence.

    • Evidence: pp. 596, 603.
    • Strength: suggestive. One case, told as the manufacturers argued it and as Congress responded; no evaluation of effects on supply, innovation or safety.

18. Litigation discovery acts as a public information-forcing mechanism, revealing what producers knew and decided. Restricting litigation also restricts this channel of knowledge.

    • Evidence: pp. 589–590.
    • Strength: moderate in this chapter. The historical record is strong (asbestos, lead, vinyl chloride), but here it is only cited, not demonstrated.

19. Standards set for the average person leave residual harm concentrated in susceptible subgroups, so some form of redress is needed for those the standards fail.

    • Evidence: p. 596.
    • Strength: moderate. Biologically well grounded, stated briefly.

20. Where society knowingly accepts a risk for broad benefit, harm falls on a few. Fairness arguably requires compensating them whether or not anyone did wrong.

    • Evidence: pp. 581, 585–586.
    • Strength: asserted (normative).

21. Up-front risk pricing that scales with worst-case estimates creates an incentive to research, and reveal, one’s own risks, reversing the incentive to stay ignorant. It may also burden new or small entrants, and depends on regulators’ ability to estimate and monitor.

    • Evidence: pp. 600–601, 603.
    • Strength: suggestive. Theoretical. Downsides are acknowledged only partly, and estimating the worst case under deep ignorance is not addressed.

22. A redress system that must pay for harm acquires an interest in detecting harm early, so compensation can be linked to surveillance of exposed workers and communities.

    • Evidence: p. 595.
    • Strength: suggestive. A proposal, not an observed effect.

23. Remedial interventions can displace harm across space and chemical form (local benefit, distant cost), so fixes need a systems-level assessment.

    • Evidence: p. 586.
    • Strength: suggestive. A hypothetical based on a National Research Council assessment.

24. Legal categories built on biological or conceptual models (e.g. “accident” versus “illness”, “divisible” versus “indivisible” disease) decide who is compensated and who pays.

    • Evidence: pp. 590–591, 600.
    • Strength: moderate. Clear doctrinal illustrations.

25. Legal doctrine on liability swings with organised pressure: periods of easing burdens on claimants are followed by producer-led rollback supported by allied scholarship.

    • Evidence: pp. 587–588.
    • Strength: moderate as history. The framing is one-sided: the rollback is attributed to interests, not also to concerns about reliability.

26. Rights-based claims that do not require proof of harm (unconsented invasion of the body or of property) could in principle move accountability earlier in the causal chain, before harm can be shown, trading smaller individual awards for many more claimants.

    • Evidence: pp. 589–590.
    • Strength: suggestive. The author’s own legal proposal (Cranor 2011, reviewed by Lyndon 2012), hedged in the text and with no cited successful application to diffuse exposures.

Limitations, contestation and bias check#

Genre and evidence base. - This is a normative proposal chapter, not an empirical study. - The evidence is mainly secondary legal literature (Kaplan 2001 for nearly all whistleblower material, with Harvard Law Review 2011 for the SEC provisions; Franklin 1979 for workers’ compensation; Schuck 2008 and Bismark and Paterson 2006 for New Zealand; Feinberg 2004 for 9/11) and Cranor’s own work (six citations). - There are almost no data on outcomes, for example: - whether whistleblower laws actually protect warners; - what administrative schemes cost compared with tort; - error rates in presumptive compensation; - fraud and administrative expense; - effects on innovation. - The CSRLD’s success claim comes from the scheme’s own website (p. 598).

Unsupported or inconsistent claims. - “Some evidence that young scientists are being discouraged from entering controversial fields” (p. 581): no evidence is cited anywhere in the chapter. - Nanotechnology is named in the summary box (p. 581) but never discussed. - Minamata: the box says “legal recognition of responsibility” was attained by 2004 (p. 581); footnote 2 says acknowledgement was sought “but … not secured” (p. 586). The two may refer to different things (legal liability versus acknowledgement), but the tension is not explained. - VICP arithmetic is wrong by a factor of ten. USD 1.18 billion ÷ 1 500 is about USD 787 000 per claimant, not USD 78 000 (p. 596). “During the 12 years of its existence” does not fit a 2010 source. - “Dermatitis caused by coal dust” (p. 590). [External: verify. McGhee concerned brick dust.] - Panel 24.1 misspells three names (“Osakawa”, “Putzai”, “Chapella”). Two of its GMO examples (Pusztai, Chapela) appear nowhere else in this volume (full-text search) and, as far as can be checked, in no volume 1 case study, contrary to the panel’s framing of them as Late Lessons case-study examples. Both are scientifically contested cases [external: Pusztai’s GM potato feeding study was criticised by the Royal Society; Nature later said Quist and Chapela’s 2001 maize paper should not have been published, though later studies reported transgenes in Mexican maize]. Listing them next to vindicated warners (Snow, Selikoff, Needleman, Patterson, Hosokawa) blurs the panel’s own criterion that recognition comes with vindication (p. 584). - Martin 1999/2008 are cited but missing from the reference list, and “Buckley, supra, n. 1” refers to a non-existent footnote. Both are signs of limited editing.

Dated material in a 2013 publication. The chapter seems to have been drafted around 2010–2012 and not updated. - Panel 24.2 is essentially McIntyre (2004) and calls Fairchild (2002) “recent”. [External: verify. It omits Barker v Corus (2006), which made Fairchild liability proportionate; the Compensation Act 2006 s.3, which restored joint and several liability for mesothelioma only; and Sienkiewicz (2011). So the claim that Fairchild “effectively creates joint and several liability” (p. 591) needs qualifying.] - Panel 24.3 describes Deepwater Horizon damage in the present tense as of 2010 (e.g. the fishery “almost completely shut down”). - The whistleblower material is from 2001. [External: verify. It omits the US Whistleblower Protection Enhancement Act of November 2012, and the widely reported weakness of WPA enforcement before it, which cuts against presenting the WPA as a model.] - The New Zealand account may be out of date on fault in medical injury. [External: verify. The 2005 treatment-injury reform. ACC also covers some work-related gradual-process disease, which undercuts the “illness excluded” critique for occupational exposures.] - The 9/11 section omits the fund’s reopening under the Zadroga Act (passed December 2010, signed January 2011) for latent illness among responders and others exposed. [External: verify.] That is the example most relevant to toxic long-tail compensation.

Missing EU dimension. For an EEA report, EU law is strikingly absent. The whistleblower survey covers the US, UK and South Africa (plus Germany’s award). There is no discussion of: - the EU Environmental Liability Directive (only cited, p. 600); - financial-security requirements under it; - the Product Liability Directive. [External: its “development risks” defence, which protects producers when the state of knowledge could not reveal a defect, runs directly against the Margereson reasoning and is central to “surprise” harms.]

The uncited references (CEA, FERMA, EBTP, the Ad-hoc Industry NRD group, the Habitats and Birds Directives) strongly suggest that an ELD section was cut.

Advocacy and one-sidedness. - No counter-voice. There is no industry, insurer, defence-bar or regulator panel. Even a critical source the chapter does cite, Priest (2003), whose title calls the 9/11 fund’s structure “problematic”, is used only for definitions, while the chapter recommends 9/11-fund features (p. 601). - The rollback of plaintiff-friendly doctrine is attributed to firms and allied scholars (p. 588), but the argument that Daubert filters unreliable testimony is not engaged. - Battery and trespass for chemical body burdens are presented as promising, though hedged (“could”, “potentially”, “might”, p. 589). They are a novel theory from the author’s own book, without cited successful applications. [External: US courts have generally been reluctant to award damages for exposure without present injury.] - Costanza’s panel treats the USD 500 000 acoustic device as decisive, which is contested. Its USD 50 billion bond is a hypothetical chosen after the fact from a USD 34–670 billion range: an exercise that knows the outcome.

What the chapter gets right about balance. Within its pro-compensation frame, Cranor is candid about the weaknesses of each model: - the SEC two-tier trade-off (p. 583); - false alarms (Gee, p. 584); - workers’ compensation’s decline (pp. 595–596); - exposure disputes (p. 597); - the preponderance standard for off-table claims (p. 597); - the CSRLD’s dependence on history, and its uncertain deterrent effect (pp. 599, 601); - the difficulty of devising workers’-compensation analogues for environmental damage (p. 595); - New Zealand’s lack of deterrence and illness coverage (p. 600); - bonds burdening new technology and depending on state capacity (p. 601); - the need to limit administrator discretion (p. 601).

He does not claim these systems are proven for novel hazards.

Hindsight and case selection. - The harassment narrative draws on warners later vindicated. Warners whose alarms proved false, and whose critics were right, are not examined, even though Gee concedes such cases exist (p. 584). - The criterion “reasonably plausible evidence” (p. 584) or “reasonably credible challenge” (p. 585) needs exactly the peer judgement the proposal is trying to guard against. The WTO “intellectual bias” point (p. 585) applies symmetrically to warners invested in their own findings, which the panel does not note. - There is a basic mismatch. Whistleblower law protects employees against employers, yet much of the harassment Gee lists (literature attacks, publication barriers, funding loss, legal threats) can come from outside the employment relationship. Other items on his list (removal from work, being passed over for promotion, loss of facilities or staff) are employer-type actions, so the list is mixed. Gee sees the distinction; the chapter’s legal proposal does not close the gap.

How the precautionary framing shapes the conclusions. - Compensation is evaluated mainly by speed, generosity and consistency with precaution. Accuracy (paying for harms not caused), cost and legitimacy with those who fund it get little attention; fraud and efficient use of funds are mentioned only in passing (pp. 596, 601). - “Benefit of scientific doubt … given to victims” (p. 581), sub-50 % probabilistic awards (pp. 598, 601) and presumptive categories built from plausibility and analogy (p. 595) move the cost of error onto producers and levy-payers. That is a defensible value choice, but the chapter presents it as a technical improvement.

Possible undisclosed interest [external: verify]. My recollection is that Cranor was involved as a plaintiffs’ methodology expert in Milward, the case the chapter holds up as a positive turn (p. 588). If so, the chapter does not disclose it. This needs checking before it is relied on.

Notable quotes#

1. “Implementing a precautionary approach to managing new technologies requires, first and foremost, administrative laws aimed at detecting and addressing risks before they materialise into harms.” (p. 582)
2. Early warners “should feel free, and be free, to research and report early warnings without the threat of adverse actions that would discourage them.” (p. 582)
3. A protected disclosure “need not prove ultimately accurate in order to be protected” (quoting Kaplan 2001, p. 583)
4. “The luxury of such hindsight is of little use to harassed early warning scientists” (Gee, Panel 24.1, p. 584)
5. “In reality, of course, flawless implementation of precautionary approaches is unattainable.” (p. 585)
6. “Although in principle these rules impartially apply to both plaintiffs and defendants, in reality they asymmetrically hamper plaintiffs, who bear the burden of proof” (p. 588)
7. “Tort law persists but its achievements fall far short of its goals.” (p. 589)
8. “The fact that other and graver risks emerged later does not detract from the power of what was already known” (Hale LJ, quoted in Panel 24.2, p. 593)
9. “If companies cannot afford the upfront bonds, they would not be permitted to engage in the activity; in short, no bond, no market.” (p. 601)
10. “The critical feature is that the risk to the public asset is apparent to the private interests in financial terms before the fact” (Panel 24.3, p. 602)

Open questions#

1. The mismatch between warners and whistleblower law. Much of the harassment in Panel 24.1 can come from peers, funders or litigants, not employers. What mechanism, beyond awards and professional-society policies, would reach it? Did any professional bodies adopt Gee’s proposed policies?
2. Telling harassment from legitimate criticism. Who decides what is “reasonably plausible” or “reasonably credible” (pp. 584–585), and how is that kept free of the intellectual bias the panel identifies? Does the same test apply to warners’ own investment in their findings?
3. “Young scientists discouraged” (p. 581). Is there any evidence for this, then or since?
4. Compensation tables before causation is known. How would presumptive tables for novel hazards be built without a history of victims (pp. 595, 599)? What are the costs, and the effects on legitimacy, of paying for harms later shown not to have been caused? Is “learning as they go” workable in both directions, including withdrawing categories?
5. Bond sizing under deep uncertainty. When worst cases cannot be bounded, how are bonds sized? How is capture of the bond-setting agency prevented, and how are new entrants protected from incumbents who can afford large bonds? Is there any real-world evidence on financial-security requirements (e.g. under the ELD) that the cut section may have covered?
6. The deterrence gap in no-fault schemes. Levies with no link to safety records (the VICP) and New Zealand-style systems weaken deterrence, on the chapter’s account (pp. 600, 603). Can experience rating work where harms are latent and the firm may no longer exist?
7. Updating the legal record. How did the UK Fairchild line develop after 2004, and did relaxed foreseeability ever extend to environmental damage, a possibility McIntyre left open (“remains to be seen”, p. 593)? What happened to Milward after 2012, and to admissibility standards in US courts generally?
8. Missing EU law. Why is EU law (the ELD and the Product Liability Directive’s development-risks defence) missing, and what did the uncited references support? The development-risks defence seems the most important European counterpoint to Margereson.
9. Disclosure. Did Cranor have an undisclosed role in Milward?
10. Cross-chapter synthesis. This chapter assumes that the barriers to heeding warnings are partly legal and incentive-based. How does it fit with Ch. 23 (costs of inaction) and Ch. 25 (why business did not act with precaution)? To be addressed in the synthesis, not here.

Audit log#

Independent audit against the full text extract (PDF 583–608), the Part D panel list (p. 563), Annex 1 biographies, the Acknowledgements and a full-text search of the PDF. Quotes spot-checked throughout were verbatim; the VICP arithmetic error, citation errors, author details and Panel 24.3 attribution were confirmed. Changes made:

1. Geographic frame: added Germany (award) and specified that REACH is mentioned twice (pp. 596, 601) and the ELD only through a citation (p. 600).
2. Introduction: corrected “Cranor 1993 cited again on p. 601”; it is cited only on p. 601, which the note now quotes.
3. SEC two-tier system: replaced a paraphrase that blended the strength and weakness clauses with the source’s own wording; added the anti-retaliation clause’s coverage of harassment.
4. UK PIDA: added that employees “are not restricted” to internal procedures (p. 583).
5. Panel 24.1 names: qualified “Pusztai and Chapela appear in no case study in either volume” (confirmed for this volume by full-text search; volume 1 not re-checked under the strand guard); added the Minamata chapter’s own note on Hosokawa and Ibsen.
6. Panel 24.1 award: corrected the jury criterion to the source’s “free from direct bias” (the “intellectual bias” label belongs to the WTO discussion); added the EC–Hormones dispute context from the reference list; fixed the same misquote in Recommendations.
7. Compensation rationales: added that Priest (2003), the source for the definitions, is a critique of the 9/11 fund’s structure that the chapter does not engage; added the chapter’s admission that institutions blur the dimensions of restitution.
8. Expert evidence: added the chapter’s causal claim that Daubert came “as a consequence” of defence pressure; corrected the Havner line (“some state and federal circuits are especially onerous”, Havner as the example).
9. Milward: clarified “typical” (single plaintiff, many defendants) and the 2012 status (one defendant still seeking a jury trial).
10. Saks 4 %: noted that the chapter generalises from one medical-malpractice figure; qualified the figure in Lessons and Mechanisms.
11. Battery and trespass: restored the source’s hedging (“might”, “would likely”) and noted that the only illustrations are historical; softened “presented with some confidence” in Limitations.
12. Main-text foreseeability (p. 590): removed “some” from a quote where the main text says “a risk of pulmonary injury”; pointed to the panel’s fuller test.
13. Panel 24.2: added Cambridge Water context and the “broad concept of injury” (p. 591); added the notice limb of the Margereson test (“actual or constructive notice”); added the Rylands and nuisance pleadings; added fn 16 (Page v Smith).
14. Workers’ compensation: added the US variants (federal, Black Lung, railroad and maritime, mostly negligence-based) and state jurisdiction; added Cranor’s concession that environmental-damage analogues are a “more difficult issue”.
15. Mobile-phone passage: renamed “Compensation ahead of proof” to “tables prepared ahead of proof”. The source conditions this on causation later being shown, and the note now separates it from presumptive categories; fixed matching wording in Recommendations and Limitations.
16. VICP: quoted “fairly automatic compensation with minimal evidentiary showings” exactly; used the chapter’s term “magistrates” (special masters flagged as external); completed the award categories (the USD 250 000 cap also covers lost earnings and lawyer expenses).
17. 9/11 fund: added the aim of “individualised” but “not overly disparate” awards; corrected the external note on the reopening (Zadroga Act, December 2010/January 2011), in two places.
18. CSRLD: retitled “Strict liability did not fix causation” (the chapter gives no reason the litigation was slow); added BNFL’s own interest in cutting its litigation costs; restored the hedged, question-form deterrence discussion and the reference to “other protective mechanisms”.
19. New Zealand: added funding by separate accounts, the self-inflicted-injury exclusion and Henderson (1981); restored the hedges (“appears”, “at least with respect to the medical compensation system”, “might underemphasise”) in the section notes and Mechanism 11.
20. Bonds: added “insofar as this could be determined” and the requirement for a supporting compensation system (p. 600).
21. 24.2.3: added the proposal to extend New Zealand-style no-fault cover to diseases and the environment (p. 601), and put it in Recommendations.
22. Lesson 6 (deterrence): removed the unsupported claim that New Zealand and the VICP are “funded by general levies” and restated each case as the chapter gives it.
23. Mechanism 2: removed the gloss “firms are most willing to commit resources before anyone knows the hazard”; replaced it with the source’s wording.
24. Mechanisms 3–5: added the notice requirement; re-labelled the Fairchild/Matthews point as contingency of causation findings, not hindsight; corrected “can only be changed from the top” to “must be communicated from the top”; marked the suppression-channel typology as interpretation.
25. Mechanism 7: qualified the tort-deterrence claim (“likely … somewhat”) and flagged the interpretive “incentive to stay ignorant”.
26. Mechanism 9: corrected “Not considered: bonds favour incumbents”; the chapter does propose lower bonds for small enterprises, so the point is now framed as incumbency advantage only partly addressed.
27. Mechanism 10: acknowledged that the chapter recognises the aim of high evidentiary standards while still not engaging the Daubert rationale on its merits.
28. Insight 3: added the notice condition. Insight 14: rewrote it and downgraded it from moderate to suggestive (Graham v ReChem was not a no-fault case, and the chapter does not attribute the slow litigation to causation). Insight 17: downgraded from moderate to suggestive (one case, not evaluated).
29. Added Insight 26 (rights-based claims without proof of harm; suggestive), a distinctive argument of the chapter that was missing.
30. Limitations: Kaplan covers “nearly all” (not all) of the whistleblower material; added the Priest (2003) point under “No counter-voice”; added two candour items (CSRLD deterrence, environmental analogues); qualified the “mismatch” point because Gee’s harassment list is mixed (some items are employer actions); softened “almost entirely” on evaluation criteria and replaced the cell-phone example with probabilistic awards and presumptive categories as the cost-shifting mechanisms.
31. Open questions: “most harassment” changed to “much”; McIntyre “anticipated” changed to “left open”; deterrence question reworded.
32. References: added “Wooley/Woolley” and “Brodeur, D.” (external: Paul Brodeur) to minor citation errors; noted that the Munich Re footnote names its source; marked the 2010 date for the SEC bounties in the timeline as external (the chapter gives no year).
33. Digest: synchronised with the above: the Margereson notice limb; Deepwater Horizon framed as the panel’s argument, with its hindsight caveat; insight 3 notice condition; insight 10 downgraded to suggestive; insight 15 (battery and trespass) added; the Priest caveat; the Pusztai/Chapela qualification; “most” harassment changed to “much”; the cost-shifting mechanisms corrected; the vaccine liability-shield point marked as one unevaluated case; deterrence feedback noted as “modest”.