Uncertainties in tort liability for uncertainty

Transcription

1 Law, Probability and Risk (2002) 1, Uncertainties in tort liability for uncertainty Ariel Porat and Alex Stein Tort Liability Under Uncertainty Oxford University Press, Pp ISBN REVIEWED BY VERN R. WALKER Professor of Law, Hofstra University School of Law, New York, USA. [Received on 26 June 2002; revised on 12 July 2002; accepted on 12 July 2002] In Tort Liability under Uncertainty, Ariel Porat and Alex Stein build upon the proposition that tort law values accurate fact-finding as a means of avoiding unfairness to parties and reducing inefficiencies due to over-deterrence or under-deterrence. Courts should base their judgements on findings that are true, and order compensatory awards that are set neither too high nor too low. But certain types of case challenge this ideal by exhibiting considerable uncertainty in fact-finding. Such cases are notorious in tort law, and judges have addressed the problems with a variety of doctrinal theories (such as res ipsa loquitur) and a variety of procedural devices (such as shifting the burdens of production or persuasion). The aim of the authors is to place these solutions on a principled footing and to supplement them with a comprehensive treatment of the uncertainty problem (p. 11). The resulting book makes a very important contribution to our thinking about how to handle uncertainty in those cases. The critical first chapter sets up the main topic of the book, and also provides a favourable framework for the authors preferred solution. That framework presupposes a fairly simple notion of fact-finding uncertainty. In the second half of this review essay, I shall raise some questions about the wisdom of imposing this conceptual framework on the problem, and about their proposed solution to the cases. First, however, I shall describe the major contents of the book, assuming for the time being the usefulness of the authors framework. 1. Outline of contents Chapter I is a defence of the traditional default rule for decision-making in torts: using the preponderance standard of proof for fact-finding, coupled with a winner-takes-all rule for awarding damages. If the fact-finder assigns a probability greater than 0 5 toall of the factual issues essential to the plaintiff s cause of action, then the court awards the plaintiff full damages. 1 Porat and Stein call this decision rule the P > 0 5 rule (p. 18). Lawvrw@hofstra.edu 1 Iambeing intentionally vague at this point about whether the fact-finder assigns probabilities to entire causes of action or to individual factual elements of the cause of action. I discuss this question further in the second part of the paper. c Oxford University Press

2 176 V. R. WALKER They interpret this rule as requiring an award of full damages if, but only if, the plaintiff s case is more probable than not (p. 18), her evidence preponderates that of her opponent (p. 18), or the probability of her allegations exceeds 0 5 (p. 20). They defend this rule from the standpoints of both utility and fairness. After thus securing for the plaintiff the high-probability ground ( P > 0 5), the authors turn their attention in the remainder of the book to securing at least some damages for the plaintiff in a variety of low-probability cases, where the plaintiff would arguably lose under the P > 0 5 rule. They argue that in certain types of case, the P > 0 5 decision rule is neither fair nor efficient, and is therefore in need of some modification or exceptions (pp. 42 4). Chapter II categorizes many cases in which the P > 0 5 rule yields unsatisfactory outcomes. In those cases, uncertainty about critical facts is a dominant feature, but we have good evidence that the P > 0 5 rule biases outcomes both unfairly and inefficiently. Porat and Stein divide these cases into five categories (p. 57): (1) cases of wrongful damage where the wrongdoer is unidentifiable; (2) cases of wrongful damage where the injured party is unidentifiable; (3) cases of wrongful conduct that may or may not have resulted in damage; (4) cases of damage wrongfully inflicted by separate wrongdoers, where it cannot be determined which part of the damage was caused by each wrongdoer; (5) cases of damage originating both from a wrongdoer and a non-wrongful cause, in which it cannot be determined which part of the damage was caused by the wrongdoer. These categories capture the kinds of case that the authors address throughout the remainder of the book. At first glance, this list seems to be generated by attributing uncertainty to different factual elements of the plaintiff s prima facie case. The plaintiff must prove that a particular defendant engaged in wrongful conduct that was a legal cause of a compensable injury to the plaintiff. Upon reflection, however, we see that the list does not include uncertainty about the proposition that specific conduct is wrongful, or about the existence or extent of any injury. Rather, the uncertainty being examined here is about tracing a cause-infact relationship from specific conduct by a particular defendant to any specific injury of the plaintiff. As the authors put it, they are investigating cases where the injury is either untraceable or indivisible (pp. 8, 58). The first three categories present the problem of traceability, while the last two add to this the problem of divisibility. The chapter introduces several model cases that are based on actual tort cases. Some examples will clarify the five categories, as well as show how the cases involve uncertainty about the cause-in-fact relationship. The first category consists of cases with multiple (possible) defendants and a single injury. The category includes cases where each defendant in the group has engaged in wrongful conduct toward the plaintiff, as well as cases where at least one defendant has been negligent, but perhaps not all. In each model case, the injury can be traced (P > 0 5) to a group of similarly placed defendants, but the injury cannot be traced to any particular defendant in that group (with P > 0 5). In the Hunters Case, two hunters negligently fire simultaneously in the plaintiff s direction, and one (but only one) of them seriously injures the eye of the plaintiff. 2 In the DES Case, 2 The Hunters Case is modelled on the famous case of Summers v. Tice, 199 P.2d 1 (Cal. 1948).

3 UNCERTAINTIES IN TORT LIABILITY FOR UNCERTAINTY 177 hundreds of manufacturers wrongfully market identical pharmaceutical products, but only one or a few sell the particular product that causes the injury of any specific plaintiff. 3 In the Two Cars Case, a pedestrian is killed in a road accident involving two drivers, one of whom was driving negligently, but it is not known which driver was negligent. In the Injured Shoulder Case, while the surgery patient is unconscious, one or more of the healthcare defendants injures the patient or fails to protect him from injury. 4 A final example is the Cargo Case, in which cargo is damaged either by the shipping company or by the port authority receiving and delivering the cargo. I have related numerous models from this first category of cases in order to show the wide range of traditional tort problems that the book addresses. Instead of leaving these and the following cases in their traditional doctrinal categories, such as products liability or medical malpractice, Porat and Stein reorganize them based on uncertainty configurations. The second major category of cases involves a single defendant and multiple potential plaintiffs. However, from the standpoint of any particular plaintiff and any particular lawsuit, the problem is still the inability to trace the injury to the defendant. The category is exemplified by the Radiation Case, in which the defendant s factory wrongfully emits carcinogenic radiation and increases the rate of cancer to 25% above the background rate. However, the defendant-caused cancer cases cannot be distinguished from the baseline cancer cases. The third major category includes single-defendant and single-plaintiff cases, but with uncertainty about whether the defendant caused any injury at all. This category includes the so-called lost chance cases for example, medical malpractice cases in which the defendant doctor s negligence deprives the decedent patient of a significant chance of survival, although the available statistics indicate that the plaintiff probably would have died even if the defendant had not been negligent. In each of these first three categories, the plaintiff or the plaintiff s decedent suffers an indivisible injury, and the litigation problem is causally tracing that specific injury back to the wrongful conduct of any particular defendant. The fourth and fifth categories add the complication of apportionment of injury: dividing the plaintiff s injury into identifiable portions, and then tracing each portion to a particular defendant s wrongful conduct. An example of the fourth category is the Dogs Case, in which two dogs, each owned by different wrongdoers, simultaneously attack the plaintiff. It is impossible to trace any particular part of the total injury to any particular dog or defendant. The fifth category covers extent-of-injury cases in which we know that the defendant did not cause all of the injury. An example is the Fire Case, in which the defendant fire department s negligent fighting of a fire causes additional (but unspecifiable) property damage, beyond the damage that a competently fought fire would have caused. In each of these last two categories, the plaintiff can prove that negligent conduct by the defendant or defendants caused some damage, but the plaintiff cannot trace (with P > 0 5) any specific part of the total injury to the negligent conduct of any particular defendant. After Chapter II organizes the problematic cases into categories, the next three chapters analyse some major judicial approaches to the problem of residual uncertainty, and discuss their advantages and disadvantages. I mention these chapters only briefly. Chapter III defends the res ipsa loquitur doctrine, and argues that the strong version of the 3 The DES Case is modelled on cases such as Sindell v. Abbott Laboratories, 607 P.2d 924 (Cal. 1980). 4 Modelled on Ybarra v. Spangard, 154 P.2d 687 (Cal. 1944).

4 178 V. R. WALKER presumption is preferable to the weak version. Under the strong version, once the plaintiff establishes the prima facie conditions, the burden of persuasion shifts to the defendant, who must be found negligent unless he proves (P > 0 5) that he was not negligent (pp ). Therefore, under the limited factual circumstances that allow the plaintiff to establish the conditions for res ipsa loquitur, the fact-finder should allocate the costs of residual uncertainty to the defendant. In Chapter IV, Porat and Stein discuss judicial approaches that modify or relax the injury requirements in certain types of case. They argue that there are good policy reasons for not recognizing bare risk as a compensable injury in torts (pp ). Once the plaintiff actually suffers a qualifying type of injury, however, such as physical injury to person or property damage, the case might fall into one of the categories for which the authors later offer a solution. For example, in the Radiation Case above, a person merely placed at risk by the defendant s wrongful emissions would have no compensable claim (although there might be compensation for actual psychological injury, such as reasonable fear of cancer). Once that person develops the relevant kind of cancer, however, the case falls into the second major category of uncertainty cases. Porat and Stein argue that applying the P > 0 5 rule in such cases would be both unfair and inefficient, so some alternative solution should be found. Past judicial attempts to develop a lost chance damage doctrine are (they argue) both conceptually and functionally problematic (p. 129). They propose instead that the defendant should compensate the plaintiff for the evidential damage the defendant causes, in keeping with their evidential damage doctrine (which I explain below). Chapter V discusses the judicial doctrine of collective liability, which imposes liability on every member of an identified group of defendants, so that each is required to pay for all or some of the plaintiff s injury. The authors argue that judges should recognize and employ collective liability under certain circumstances (p. 131). Those limitations on justifiable use, however, would not eliminate the need for the authors evidential damage doctrine. The two final chapters, VI and VII, discuss the authors proposed solution to most of these model cases. They describe their solution as liability for uncertainty, and call it the evidential damage doctrine (p. 160). They argue that from the standpoint of corrective justice, victims have an entitlement to ascertain the causes of their respective damage (pp. 167, 174), and a person sustaining any damage that might have been inflicted wrongfully is entitled to information that identifies the causal forces that produced her or his damage vis-à-vis the person who wrongfully deprived her or him of that information (p. 176). This entitlement derives from a person s autonomy in enforcing his or her legal rights. The law of torts should hold the defendant responsible for evidential damage when the defendant s wrongful actions have impaired the plaintiff s ability or reduced his chances to establish the facts necessary for prevailing in a direct-damage lawsuit for an injury normally compensable in torts (pp ). As they put it: Evidential damage must thus be perceived as an indirect or adjective damage: its existence will always depend on the actual (or sufficiently probable) occurrence of a directly actionable damage (p. 161). What should the remedy be if the defendant is found liable for causing evidential

5 UNCERTAINTIES IN TORT LIABILITY FOR UNCERTAINTY 179 damage? Merely shifting the burden of persuasion on the cause-in-fact issue will often be inadequate, even in cases where the defendant who is liable for the evidentiary damage is the same defendant who acted wrongfully and might have caused the underlying direct damage (e.g., the death or the cancer). The reason is that, given the limited evidence available in the case, the probabilities might be against the plaintiff, and the defendant will be able to prove this (P > 0 5). The problem is that additional evidence would have been available, except that some wrongful act of the defendant caused it to be unavailable. Porat and Stein argue that in cases where the burden-shift remedy is inadequate, the defendant should compensate the plaintiff for the loss of the information necessary for ascertaining the cause of her or his direct damage (p. 167). Damages should equal the diminution in the expected value of the plaintiff s direct-damage lawsuit that is due to the evidentiary damage (p. 168). We would calculate the expected value by multiplying the amount of the direct damage by the probability of the allegation that the direct wrongdoer (or the originator of the relevant risk of damage) actually inflicted that damage (p. 168). For example, in the Hunters Case, two hunters negligently fire simultaneously in the plaintiff s direction, with one hunter (and only one) causing a serious eye injury. One of the two defendants causes the physical injury, while the other causes evidential damage by firing simultaneously although we cannot establish which defendant played which role. The authors believe that each hunter should be responsible for the entire cost of the physical injury (pp. 162, 167, 174 5, 186): the defendant that caused the physical injury should pay for it, and the other defendant caused the value of the plaintiff s case for physical injury to fall from full damages to zero. We cannot prove which cause of action (for physical injury or for evidential damage) applies to which defendant, but this no longer matters, since the remedy has the same value in either case. In cases involving more than two defendants (such as the DES Case), the evidential damage caused by any particular defendant would be measured by the monetary value of the physical injury multiplied by the probability that that defendant actually inflicted the physical injury (thus approximating the market-share doctrinal solution to the DES cases). (See pp , 202.) The Radiation Case provides yet another scenario. The cancer that the defendant wrongfully causes to some people is indistinguishable from the cancer in the baseline cases. If the defendant has also wrongfully impaired the victim s ability to establish which group of cancer cases she is in, then the remedy should be equal to the value of the lost information (pp ). If d equals the monetary value of the victim s physical injury, and 0 2isthe probability that the plaintiff is a defendant-caused case instead of a baseline case (because 20 cases per 100 are defendant-caused), then 0 2d equals the expected value of the victim s lawsuit against the defendant absent the evidential damage (p. 177). With the evidential damage taken into account, the value of the suit (under the P > 0 5 rule) is zero. According to the authors, 0 2d is therefore the value of the lost information, and the value of the evidential damage inflicted. 2. Critique of the conceptual framework In this second half of the review essay, I raise some questions about the workability of the authors proposed cause of action for evidentiary damage. I also make some general comments on the conceptual framework of the analysis. Before doing so, however, I wish to state clearly that this book performs a very important service in attempting to rationalize a

6 180 V. R. WALKER variety of difficult cases. The survey above suggests the range of cases covered by this synthesis. The distinctive feature of the synthesis is organizing these cases around the problem of dealing fairly and efficiently with residual uncertainty. The result is not only a substantive contribution to tort law, but also a pattern of analysis useful in other areas of law that conduct fact-finding under uncertainty. So although I am raising some questions about the particular proposals of Porat and Stein, this is not a criticism of the scope and strategy of their enterprise. With respect to their proposed solution to the problem, I believe that they will have a difficult time convincing any court to adopt their doctrine of evidential damage. While it may be true that there are serious failings in past attempts by judges to address these problematic cases, the courts will nevertheless be wary of adopting this new, sweeping doctrine. Liability for evidential damage raises an indefinite number of issues that would trigger extensive appellate litigation. I shall mention only a few of them. I treat the evidential damage doctrine as establishing a distinct cause of action, since it recognizes a distinct type of legal injury, can involve a distinct wrongful act, and is applicable when the evidential-damage wrongdoer is a different person from the direct-damage wrongdoer (pp ). First, with respect to the claim that there is a legally protected interest in information, what is the nature of that entitlement? Is there merely a duty to preserve evidential information of value to another, or is there sometimes a duty to generate it? Does creating a risk of physical harm to another automatically generate such duties? Moreover, we can expect plaintiffs to ask the courts to recognize special relationships that trigger an affirmative duty to protect or generate information by analogy to the traditionally recognized special relationships that create a duty to protect another from physical harm. (See p. 170.) For example, could an academic researcher ever have a duty to add important variables to a governmentally funded study, or to maintain data files once they exist, or to produce unpublished data or analyses upon demand? There may also be no way to anticipate the implications for other areas of law if we recognize an entitlement to information. With regard to the factual element of breach of duty, what are the criteria and tests of wrongful conduct when the issue is evidential damage? What would a reasonable person standard amount to when applied to negligent impairment of the plaintiff s evidential case? It may be too much to expect judges or juries to have settled conceptions of what a person of ordinary prudence would do to preserve someone else s ability to prove a tort case. For example, would it constitute constructive notice of the risk of causing evidential damage that a reasonable person would know that the evidence might be of some value to some plaintiff? What is the minimum preventive action that the defendant would be required to take to avoid liability for evidential damage, short of not putting the plaintiff at risk at all? A cause of action for intentional infliction of evidential damage raises additional questions, as does the question of joining a claim of negligent infliction of evidential damage to a claim of battery. Such questions are pressing even in the easiest of cases, where the same person who creates the risk of physical harm also allegedly causes the evidential damage. In the Hunters Case, even if firing a gun is negligent because it puts the plaintiff at risk for physical injury, is firing that gun also negligent, ipso facto, with respect to evidential damage? The simultaneity of the firing seems crucial to making the act negligent with

7 UNCERTAINTIES IN TORT LIABILITY FOR UNCERTAINTY 181 respect to evidential damage, but not with respect to physical injury. Firing 10 seconds apart would still be negligent with respect to physical injury, but probably not with respect to evidential damage. Would a plaintiff claiming evidential damage have to prove actual or constructive notice of the simultaneity? Is this simply the risk one accepts by going hunting with someone who has a similar reaction time and shooting speed? In the Radiation Case, even assuming that the emission of carcinogenic compounds into the air breaches a duty of care because it creates an unreasonable risk of cancer, is such conduct ipso facto a wrongful act with respect to evidential damage? Does the conduct become wrongful with respect to evidential damage simply because the kind of cancer it can cause is indistinguishable from baseline cancer cases? 5 The authors consider evidential damage to be foreseeable in such a case (p. 173), but there is more to wrongfulness than foreseeability. Porat and Stein provide occasional examples of conduct that they consider wrongful with respect to evidential damage, but appellate litigation would still have to provide definitive answers if a cause of action were adopted. With respect to the elements of causation and injury, even more questions arise. Is the notion of impairing the plaintiff s ability to prove her case sufficiently operational? In evidence spoliation cases, we can limit the claim to instances where someone destroys physical evidence. But the authors vague references to taking away information (e.g., pp ) invite confusion. What information do the hunters take away in the Hunters Case, or does the factory owner destroy in the Radiation Case? The essence of the causal claim in these cases seems to be that but for the simultaneous firing by one defendant, the plaintiff would be able to prove whether the other defendant s shot hit the plaintiff or not; and that but for the emissions, the plaintiff would know that her cancer is a baseline case. Sorting out causal connections to evidential damage is puzzling enough, but talk about taking away information does not help, and probably adds to the confusion. Even on the question of the appropriate measure of damages for evidential harm, I foresee an explosion of appellate litigation. Is the value of lost information about who caused a particular injury simply a function of the probability that some particular party caused the injury? What would a rational person be willing to pay for information that might help to prove a case against some particular defendant? If the compensation were really for an impaired ability to prove the case, and not simply for the increased risk of the underlying, physical injury, it seems that we would need fact-finding about the expected value of the various litigation outcomes, with and without the missing information. For example, would not the ability of a direct-injury defendant to satisfy a judgement be relevant to the plaintiff s willingness to pay for information about that defendant? If evidential damage is to be a distinct cause of action, we have to think through its factual elements in a systematic fashion. It is one thing to allocate to a defendant a burden of production or persuasion on a traditional factual issue, and to give as one policy reason for doing so the fact that the defendant is responsible for the uncertainty about that factual issue. It is quite another 5 Ihave argued elsewhere that there is a baseline risk paradox : in scenarios such as the Radiation Case, unless additional information allows us to eliminate the baseline cases by identifying a sub-group of cases that are all caused by the defendant s wrongful act, then assigning the plaintiff to the defendant-caused group or to the baseline group must be epistemically arbitrary, and can be justified only by non-epistemic policies. See WALKER, V. R The Concept of Baseline Risk in Tort Litigation. Kentucky Law Review, 80,

8 182 V. R. WALKER matter, however, to create a new cause of action for the uncertainty, with new factual elements whose meanings and applications would have to be litigated. Moreover, courts will inevitably worry about use patterns for the new cause of action. If the proposed cause of action were established, adding a claim of evidential damage might become de rigueur for plaintiffs, defendants might counterclaim against plaintiffs for evidential damage (as when a plaintiff who alleges injury from a defective product throws the product away), and disappointed losers in underlying suits would be on the hunt for hapless evidence damagers. Even in cases claiming spoliation or destruction of physical evidence, where the courts face questions that are far easier than those raised above, courts have been quite reluctant to establish a cause of action for damages. 6 It is therefore inadequate to respond to the above questions by saying we should leave it to the courts to work out answers on a case-by-case basis. The courts will be leery of adopting any part of the proposed solution, until they are satisfied not only that a solution is needed, but also that this particular solution will solve more problems than it will create. I shall conclude this essay with a few remarks about the framework that the authors use to set up the general problem and their proposed solution. They begin by positing the basic decision rule of torts as a combination of two rules a fact-finding rule and a damageaward rule. The first rule directs the fact-finder to find proposition p to be true for purposes of the litigation whenever the fact-finder assigns to p a probability greater than 0 5. The second rule awards full damages on the basis of the findings of fact about liability. This is the rule that they defend in Chapter I and take as the default rule throughout the book. Despite their giving this rule a single name, it is obviously composed of two distinct parts. The fact-finding rule could be modified without modifying the damage-award rule, and damage awards could be modified without changing the fact-finding rule. I shall confine my remaining remarks to the fact-finding part of the rule, which directs that findings be made on an all-or-nothing, true-or-false basis, even though evidentiary support for those findings may be only probabilistic and may have degrees of strength. My first comment is that we should be cautious about assuming that a fact-finder can always assign probabilities to proposed findings of fact in tort cases, at least if we require those assignments to be meaningful, reasonably accurate, and warranted by the evidence in the record. This is especially true if we equate probabilities with mathematical probabilities, as in P(proposition p) >0 5. There are of course times when doing so is possible for example, when the probability estimates the relative frequency of outcomes from a causal system that we understand reasonably well, such as gaming mechanisms and some other physical systems. We know what such a probability means, and sampling data can provide both warrant and a determination of accuracy. But most propositions at issue in torts are more problematic. Probability values interpreted as relative frequencies are often unwarranted by the available evidence, and perhaps even meaningless when the proposition is about unique events. Probability values interpreted as subjective degrees of confidence are often unwarranted as predictors of events, and subjective beliefs about propositions are not always admissible. Moreover, there is no validated theory to warrant assigning probability values to propositions if the probabilities represent degrees of evidentiary support. In many of these areas, we are making some headway, but we are very far from 6 See, e.g., DOBBS, D.B.2000 The Law of Torts 451 (West Group).

9 UNCERTAINTIES IN TORT LIABILITY FOR UNCERTAINTY 183 having any general assurance of meaning, accuracy, and warrant for assigning probability values to fact-finding propositions. The preponderance standard does not itself presuppose that such assignments are always possible. The preponderance standard of proof provides important guidance to factfinders, such as telling them that they need not be sure that p is true in order to make a finding that p is true. But it can perform this function without requiring fact-finders to assign mathematical probabilities to all (or even most) propositions at issue in law. 7 Porat and Stein are appropriately skeptical of allowing probabilities that reflect only the subjective experience-based beliefs of the fact-finder (p. 51, note 81), but they are also too quick to justify probability assignments using the principle of indifference (pp. 46, 178). But without recourse to such default justifications, there is no reason to think that we can routinely make meaningful, accurate, and warranted probability assignments to propositions. Secondly, the authors seem ambivalent about whether a fact-finder assigns probabilities to entire causes of action or to the factual elements of a cause of action. On the one hand, they sometimes talk about the plaintiff s case, allegations, or factual account as having a single probability (pp. 18, 20). Their argument in Chapter I certainly presumes that we can assign a single probability to the plaintiff s entire case. On the other hand, they also talk about assigning probabilities to the factual element of causation, as opposed to the elements of negligent conduct or damage (e.g., pp , 84). And they argue that the factfinder must establish separately each proposition that is elemental to the cause of action, without applying the multiplication rule for conjunctions (pp , note 83). We are left to wonder how a single probability for the case relates to the several probabilities assigned to its elemental propositions. Many readers will find unsatisfying the authors assertions that the probability debate poses a false dichotomy and that the conjunction paradox is unreal (pp ). I shall not criticize this excellent book for not containing a more comprehensive theory of probability. Providing a theory of probability would have diverted attention from the main topic. Nevertheless, many readers will miss having a more complete account, and answers to such problems will eventually come due. Thirdly, perhaps because the authors represent the uncertainty about the truth of a proposition by assigning a single probability to the proposition (with P < 1) (pp ), they conduct their analysis using a fairly simple concept of uncertainty. This is true despite their distinction between the dimensions of probability and weight (pp ). They do not analyse uncertainty into different kinds (such as measurement uncertainty, sampling uncertainty, or modelling uncertainty), nor do they distinguish different sources of uncertainty, nor do they explore different ways to characterize different degrees of uncertainty. 8 As a result, they primarily address the question of where to allocate the residual risk of error when the single probability assigned to the plaintiff s case is not greater than 0 5. This conception of uncertainty places limitations on their analysis. For example, their analysis misses opportunities to devise decision rules that are based 7 Ihave argued elsewhere that the preponderance-of-the-evidence rule for fact-finders can implement major legal policies without adopting a precise mid-point on a cardinal scale (such as 0 5 on a scale from zero to one) as a threshold decision point for purposes of legal fact-finding. WALKER, V. R Preponderance, Probability and Warranted Factfinding, Brooklyn Law Review, 62, See, e.g., WALKER, V.R.2001 Theories of Uncertainty: Explaining the Possible Sources of Error in Inferences, Cardozo Law Review, 22, 1523.

10 184 V. R. WALKER on different types and degrees of residual uncertainty. A more complex conception of uncertainty could lead to more refined ways of dealing with the cases they address. 9 In sum, Porat and Stein have made a valuable contribution to tort law, and to the general legal problem of how to deal with fact-finding uncertainty. Future writers on these subjects will have to address their categories and model cases, their treatment of past doctrinal and procedural solutions, as well as their own proposed doctrine of evidential damage. Questions remain to be answered, but those questions stand as a tribute to the boldness of the scheme presented. Acknowledgement The author is grateful for research support for this article from a grant from Hofstra University. 9 For adifferent approach to uncertainty in the lost chance cases, see WALKER, V.R.1994 Direct Inference in the Lost Chance Cases: Factfinding Constraints Under Minimal Fairness to Parties, Hofstra Law Review, 23, Review of Porat & Stein, Tort Liability under Uncertainty June 2002 Vern R. Walker. P. 21.