The Yale Law Journal


Mistakes, Misunderstandings, and Misalignments

20 Mar 2012


In his recent article in The Yale Law Journal, entitled Misalignments in Tort Law,1 Ariel Porat identifies and characterizes five “misalignments” in the tort of negligence.2 Misalignments are problematic insofar as they might lead potential injurers to make inappropriate investments in accident prevention. Porat suggests that in some, but not all, cases the inefficiencies that result from misalignments can be justified on other grounds: in particular, those of corrective or distributive justice.3 I leave it to others to determine the extent to which Porat’s article contributes to the economic analysis of law. My worry is that it fails to contribute to our understanding of either tort law in general or the tort of negligence in particular.

This Response proceeds as follows. First, I identify the notion of misalignment that Porat employs and display some problems internal to it. I then characterize the notion of alignment that is internal to the law of torts. With the proper notion of alignment in hand, I demonstrate that none of the cases Porat discusses are problematic.4 The Response closes by emphasizing the importance of characterizing the distinctive normative structure of tort law correctly before assessing its relative ability to achieve social policies.

I. Porat’s Notion of Misalignment

For all the work that Porat requires of the concept of misalignment, it is puzzling how little he offers in the way of an analysis of it. Instead of analyzing the notion of misalignment, he is satisfied to offer an illustration of the idea he has in mind.5 There are two problems with Porat’s approach. The first is that the illustration he offers raises more questions than it answers. The second is that theoretical work is always better served when the concepts asked to carry a heavy argumentative burden are carefully and precisely formulated—all the more so when the concepts are technical or otherwise unfamiliar, which is the case with the concept of misalignment in tort law. Let us begin, then, by filling in some of the gaps in Porat’s formulation by providing a basic analysis of the concept of misalignment. To do so, we must understand the more basic idea of alignment.

First, alignment is a relational concept that applies to objects or elements of a set. Second, objects are aligned with one another with regard to some property or other, not in the abstract. Third, alignment is a normatively neutral notion. Whether or not the elements of a set are aligned is one thing; whether the elements being aligned or not is desirable or valuable is another matter.

We can illustrate these ideas with a simple example. Suppose I line up the students in my torts class in some order that appears random in that the taller and shorter students are dispersed throughout the line. So the students in my class are not lined up according to relative height—e.g., shortest to tallest, front to back. They are not aligned according to height, though they may well be aligned with regard to some other property—for example, their exam scores.

They are misaligned only if the aim of lining them up is to do so in a way that reflects their relative height—shortest to tallest, front to back. This implies that while alignment as such lacks a normative valence, misalignment does not. The elements of a set are misaligned only if the way in which they are lined up is defective, inapt, or incorrect relative to an applicable standard of assessment.

With these preliminary remarks in hand, we can now ask the obvious questions about the tort of negligence. (1) What are the elements of the set in the tort of negligence whose alignment or misalignment is at issue? (2) What is the property in virtue of which they are to be aligned? (3) What is the standard by which we are to judge whether alignment according to that property is correct or incorrect, appropriate or not?

II. Alignment and Correctness

Porat offers the following illustration of what he calls the alignment principle:

  To illustrate the alignment principle, consider an injurer who creates a foreseeable risk of harm to his neighbor’s property in the amount of 10 but can reduce it to zero by taking precautions that would cost him 7. Under a rule of negligence, as interpreted by the courts, failing to take these precautions would amount to negligence since the costs of these measures are lower than the expected harm (7 < 10). Therefore, if the injurer’s failure to take precautions results in harm to his neighbor’s property, he will be liable for the ensuing harm. Here the alignment principle clearly applies: the expected harm (the risk) is taken into account by the court when it sets the standard of care, and if the injurer fails to meet that standard, he bears liability for the harm that materialized. Indeed, this is the logic of negligence law that is acknowledged by law and economics theorists but is also consistent with corrective justice: the negligent injurer faces a liability risk that is equal to the foreseeable risk he negligently created for others.  
  Assume now that the law imposes liability on the injurer in our example for only half of the harm that materialized, despite the lack of contributory negligence on his neighbor’s part. Under such an approach, a misalignment between the standard of care and the compensable harms would arise: on the one hand, the standard of care would be set according to the full expected harm of the wrongful behavior, yet on the other hand, liability would be imposed for only half of the harm done.6  

The point is straightforward. The ground of the tort of negligence is the injurer’s negligence. Injurer negligence is determined by employing the Learned Hand formula.7 According to the Hand formula, a defendant is negligent only if the cost to him of taking precautions is less than the harm to the victim discounted by the probability of the harm’s occurrence. In the event that the defendant is negligent, he will bear the full costs of the harm caused and not the value of the harm discounted by the probability of its occurrence. In this case, the remedy would be compensatory damages of 10. The harm to the plaintiff as represented in the Hand formula is 10 as well, and so in this sense, the Hand formula and the damage award are aligned. In contrast, should the plaintiff be awarded only 5 in damages, the elements (the standard of care and the remedy) are misaligned because the harm to the plaintiff is differentially represented in both: i.e., as 10 in the Hand formula, but only 5 in the remedy.

The illustration raises at least as many questions as it answers. Suppose the actual damages to the plaintiff are 10, but the jury miscalculates the harm as 5. But suppose as well that in awarding damages the court similarly misconstrues the damages and awards the plaintiff 5. In other words, if the jury and the court make the same mistake in both cases the elements are aligned, but incorrectly so. There is therefore an important difference between alignment and correctness. In tort law, departures from correctness are mistakes—sometimes harmless, sometimes excusable, but always less than what tort law calls for.

The key point is that whereas tort law aspires to correctness, correctness may or may not matter from the point of view of efficiency. To see this, note that in our example if the costs of precaution are 3, then the rational defendant will be induced to take efficient precautions quite apart from whether the damages are represented as 5 or 10 (or anything greater than 3 for that matter). But if the costs of precaution are 6, representing damages as 5 will not do—yet any representation of the damages greater than 6 would! This suggests two problems. The first is that efficiency does not depend only on alignment of the damages in both the Hand formula and the remedy, but also on the costs of precaution (and the accuracy with which it is represented as well). The second problem, perhaps more worrying than the first, is that the correctness of the representation of the harm—crucial to a well-functioning tort law—is of only contingent and secondary importance to efficiency. If correctness matters only in the way that I worry it might to alignment, and thus to efficiency, then I worry that alignment may be of more interest to efficiency than either alignment or efficiency are to tort law.

III. Porat’s Examples of Misalignment

Here are summary descriptions of four cases that demonstrate what Porat describes as misalignments:

  a) One day an injurer drives in a poor neighborhood; the next day in a rich neighborhood. Typically, the risk of danger (assuming the driver is equally careless) on both days is the same. However, in the event he causes harm both days he will pay more, ceteris paribus, to the rich victim than he will to the poor one because of the relative disparity in lost income between the two victims. He will pay more to the rich victim than he will to the poor even though he drove as carelessly with regard to the interests of the one as the other—even if he caused the same physical damage to each. So there is a misalignment between damages as they appear in the Hand formula and as they appear in the damages remedy. On the other hand, were we to realign the elements by having the Hand formula reflect the differential damages in both cases, we would be in an awkward position: we would require potential wrongdoers to take greater precautions to guard against injuring the rich than they would be required to take to protect the same interests of the poor.8  
  b) In a particular hospital, a unit treats very ill patients whose average chances of recovery are 30%. The doctors are sometimes negligent toward these patients, many of whom do not recover in the end. Under the preponderance-of-the-evidence rule, no patient can recover since it is only 30% likely that the doctor’s negligence is the cause of her harm, and recovery would require a greater-than-50% chance of causation. That is so whether the doctor in any particular case was negligent and the cause of the patient’s demise. By contrast, if the average chance of recovery is 70%, then victims will recover every time the defendant is negligent, even though in almost a third of the cases it is likely that the defendant’s negligence is not the reason the plaintiff fails to recover.9  
  c) Imagine the following case. The harm to the victim is 8, and the cost to the defendant of guarding against it is 10. Thus, the injurer is not negligent for failing to guard against the harm to the victim. But suppose now that by not guarding against the harm, the defendant runs the risk of imposing a loss of 7 upon himself. Now the total loss is 15, and the cost of prevention is 10; thus, it would be inefficient not to guard against it. By failing to include the cost to the defendant as well as the cost to the victim, the law leads to too little investment in precautions. The harm the defendant should be guarding against is 15, not 8.10  
  d) A driver rushes a wounded person to the hospital in his car and on the way hits a pedestrian. Assume that, had he driven his car a bit more slowly, the driver would have avoided the accident. The court determining whether the driver was negligent would take into account both risks increased (to pedestrians) and risks decreased (to the wounded person) and decide whether, given both sets of risks, the driver behaved reasonably. A negative answer would yield liability for the entire harm, without giving due allowance for the decreased risks to the wounded person. Thus, if fast driving increased risks to pedestrians by 500 but at the same time decreased risks to the wounded person by 400, the driver would be considered negligent and found liable for the entire harm. This liability for the entire harm creates a misalignment: again, the offsetting risks are taken into account only at the stage when the court sets the standard of care and disregarded when it awards damages. In contrast, the alignment principle would dictate liability for only one-fifth of the harm done.11  

At first blush, these cases seem to have very little in common. Porat agrees. But he claims that, upon closer inspection, there is a common thread:

  [I]n all [these cases], there is an inconsistency in how the standard of care is set versus how damages are awarded. It is as though one theory of tort law applies to the former and another to the latter. Generally speaking, in all of the cases except the injurer’s self-risk cases, the courts set the standard of care efficiently, but liability is not efficiently imposed. When setting the standard of care, the court takes into account all foreseeable risks, but at the imposition of liability stage, it ignores some of those risks or else assigns them incorrect values. Thus, in the lost income case, the standard of care is set uniformly, regardless of the potential victim’s lost income, but damages vary according to the particular victim’s lost income. In the proving causation case, the standard of care is set according to the expected harm, but when there is a downward or upward bias in the probability of causation, damages (in the extreme case) are set at zero or at the amount of the full harm, respectively. With offsetting risks, both risks increased and risks decreased are considered when courts set the standard of care, but damages are awarded for risks increased without due allowance for risks decreased. . . . The only instance where the misalignment goes in the opposite direction is the injurer’s self-risk case: while the negligent injurer bears all risks created by his negligence, the standard of care is set only according to the risks he imposed on others.12  

Porat’s idea must be this: The elements of the tort of negligence that are to be aligned are the Hand formula’s representation of the harm and the damage award. The property in virtue of which they are to be aligned is expected damages. The standard for determining whether the elements are correctly aligned is efficiency.13

Aligning the Hand formula with the damage award matters in tort law to the extent that efficiency matters. But tort law has its own standard of alignment that is independent of efficiency. Indeed, it is independent of any particular theory of tort law or of any particular aim or goal of tort law. It is a standard of alignment internal to tort law itself. The elements of tort law are aligned in that, if satisfied, they provide grounds or reasons for imposing the remedy. They are aligned insofar as they are grounds for the remedy and not otherwise. Let me explain.

IV. Tort Law’s Notion of Alignment

Every first-year torts student should be able to identify the four basic elements of the tort of negligence: duty, breach, harm, and causation.14 In the tort of negligence, the duty is to take appropriately into account (and, in doing so, to show adequate regard for) the interests of others in the security of their possessions and person.15 Breach consists of failing to take appropriately into account the interests of those to whom one owes the duty of care.

It is a commonplace that negligence is behavior rather than a state of mind. This means that one’s negligence consists in a failure on a particular occasion to act as would a person who takes the interests of others appropriately into account. Negligence does not require a disposition to act in a particular way or to have or hold others in a particular regard—as unworthy or less equal or what have you.

We turn to the harm requirement. In spite of what economists of law would have us believe, tort law speaks infrequently, if at all, of costs. Instead it invokes the concepts of wrong and of harm.16 The fact that someone has been harmed or wronged gives others reasons to respond: to come to the victim’s aid, to feel sympathy, empathy, and the like. That one is burdened by various costs—as such—provides no similar grounds for action or emotion. Harms and wrongs have a normative valence that costs do not. The concept of cost may be all that the accountant or the economist needs, but it is not the concept the law of torts relies on as a basis of repair or recourse.

We come finally to the causation requirement. The connection between the breach (the wrong) and the harm must exhibit the right causal structure: the causal structure required for responsibility. These are the elements that, in the standard case, the plaintiff must prove in order to establish the prima facie case.

With these relatively familiar points in place, we can now specify the concept of alignment that is internal to the law of torts: the elements and the damage remedy align insofar as the elements of the tort are the grounds of the remedy. If the conditions set out in the four elements are satisfied, then the remedy is grounded, which is to say that it is called for, warranted, or justified. The notion of alignment inherent in torts implicates the concept of justification. The elements of the tort must align with one another and the remedy in the sense that, taken together, the satisfied elements warrant, justify, or provide grounds for the remedy.

Suppose the plaintiff brings an action in negligence and successfully argues that each of the elements is satisfied. If what I have argued to this point is correct, it follows (ceteris paribus) that the remedy is warranted or justified. But what is the remedy in tort? It is not, as Porat implies,17 the imposition of damages on the defendant. In fact, the remedy in tort is the conferral of a power on the plaintiff—a power to impose a liability, not a liability as such.

A power provides its possessor with an option. He has the authority to impose the liability, but he is not required to do so. As Hohfeld pointed out nearly a century ago, liabilities are normative correlates of powers.18 If B has a power with regard to A, then A is vulnerable to B’s exercise of that power. A is liable in that sense. From the normative point of view, the successful tort suit renders the defendant vulnerable to the plaintiff’s power to impose an evil upon him. The successful plaintiff has the option, but not the duty, to impose an evil in the form of a liability upon the defendant.

This means that the elements of the tort align insofar as, taken together, they warrant the conferral of a power to impose an evil (a liability) on the defendant. Our question is: What kinds of considerations are reasons for conferring powers to impose liabilities? Yet even that question is too broad. The better formulation is: What kinds of considerations are adequate or appropriate to confer a power on the plaintiff to impose a liability or evil against the defendant—a power that is enforceable by the coercive authority of the state? To answer this question, we need to consider the relationship between the defendant and the plaintiff both prior to the breach and after the injury.

The defendant owes the plaintiff a duty of care. The duty of care gives rise to two related but nevertheless distinct demands. Because the defendant owes the plaintiff a duty of care, the plaintiff is entitled to demand that the defendant take the interests she has in her person and property appropriately into account. In addition, the plaintiff is entitled to demand that the defendant not harm her through the defendant’s carelessness. It is the latter demand that figures prominently in tort law and forms the basis of the plaintiff’s action.

In the tort suit the plaintiff contends that the injurer failed to live up to his responsibility not to cause harm to the plaintiff as a result of a failure adequately to display proper regard for the plaintiff’s interests in her person and property. If the plaintiff succeeds in her suit against the defendant, then the plaintiff is in the normative situation of having the right to demand an appropriate response from the defendant. In a loose sense we might say that the power to demand care ex ante becomes the power to demand “repair” or “recourse” ex post. The more precise way to put the point is that the failure to respond adequately to the duty not to harm as a result of taking inadequate care provides grounds upon which the power to demand repair or recourse rests. The point of the lawsuit, of course, is to determine if the power to demand repair or recourse is warranted. And the key part of that case—from the normative point of view—is the element of duty, not the element of breach. After all, there is no breach without a duty.

If we put these points together, we can identify three significant ways in which actual tort law differs from Porat’s characterization of it. First, tort law has its own standard of alignment. That standard specifies a relationship between the elements themselves and a relationship between those elements taken as a whole and the remedy: namely, do the elements of a tort, taken together, provide warrant for the remedy? The alignment between the harm to the victim represented in the Hand formula and the damage award may be important for efficiency, but that is not the notion of alignment internal to tort law. Second, the remedy in tort is a power to impose an evil upon the defendant; it is not a liability imposed upon the defendant as such.19 Finally, the most normatively important feature of the tort of negligence is the duty to take care, not the Hand formula as a way of determining breach of that duty.20

I have long argued that the relationship between the elements of a tort and the remedy represents a scheme of practical reasoning that looks something like this:

  Question: Does the defendant owe a duty of care to the plaintiff?  
  Answer: Yes, the defendant owes such a duty.21  
  Question: Did the defendant breach the duty to the plaintiff?  
  Answer: Yes, the defendant breached the relevant duty.  
  Question: Was the plaintiff harmed in a way the law seeks to prevent?  
  Answer: Yes, the plaintiff suffered a harm of the appropriate sort.  
  Question: Was the harm the result of the breach?  
  Answer: Yes, the harm was the result of the breach.  
  Question: What ought to be done given these answers?  
  Conclusion: The plaintiff is entitled to a power to impose a liability upon the defendant.  

The elements of the tort of negligence are aligned (if they are) by this pattern of practical inference. If there is a notion of alignment appropriate to tort law, it is this one. There may be other potential alignments that bear on inducing injurers to take efficient precautions, but they are not alignments in tort law; nor does their failure to obtain constitute a misalignment in tort law. Having the damages remedy line up with the damages as represented in the Hand formula may be desirable from the goal of inducing efficient investments in accident avoidance, but that is no reason for thinking that their failure to line up thus is a failing of tort law. In fact, without more, the notion of alignment that is the focus of Porat’s interest may well be of no interest to tort law at all. Certainly Porat has given us no reason for thinking that it is. Tort law, as we have seen, has its own notion of alignment intrinsic to it, represented by the scheme of practical reason that partially constitutes tort law and is itself partially constituted by tort law.

V. Reasons of the Right Kind

As Porat sees it, the failure of the Hand formula and the damage remedy to align is a problem because it threatens to induce inefficient investments in precautions.22 He then allows that other principles, in particular those of corrective or distributive justice, might explain or justify these departures from efficiency.23 But one need not invoke a substantive principle of corrective or distributive justice in order to explain Porat’s misalignments—and that is because they are not misalignments at all.

Remember Porat’s claim that the extent of misalignments in tort law is such that it makes one think that tort law is governed by two quite distinct theories—one that applies to setting the standard for negligence, and the other governing the damage award.24 But this is a mistake that Porat makes because his notion of misalignment is orthogonal to the notion of alignment embodied in the normative structure of tort law.

All we need to do is remember that the elements of the tort are supposed to provide grounds for the practical inference, which is that the plaintiff is entitled to a power to impose a liability upon the defendant. The Hand formula and the damage remedy are not governed by two theories. They are parts of the same pattern of inference: the pattern of inference that is at the center of tort law. Let us see what the consequences of this realization are for Porat’s examples.

A. Rich and Poor Victims

We can start with the example of careless driving in rich and poor neighborhoods.25 The example invites two questions. First, why is it that the precautions the driver is required to take do not vary depending on whether he is in the rich or the poor neighborhood? The second is that, given that the precautions the defendant takes are invariant between the rich and poor neighborhoods, why is it that the liability he bears varies in ways that reflect the relative wealth of the victims?

The prospect that a driver might be required to exercise levels of care that reflect the relative wealth of his potential victims is so unappealing that one might think that requiring him to take the same precautions in both cases needs no justification.26 But it does require justification, especially if considerations of efficiency, for example, might call for differential investments in precautions that reflect the wealth of potential victims.

The justification of the current practice is obvious once one recalls that the elements of the tort are supposed to provide reasons of the right kind for the practical conclusion: the conferral of a power to impose a liability. One’s relative wealth is not the kind of fact about one that can ground a power. The fact that you have wronged me is a reason of the right kind for conferring a power on me to impose an evil upon you. However, the fact that I am richer than someone else (who is also a potential victim of your actions) is not a reason of the right kind for conferring a power on me to impose a liability upon you.

The plaintiff’s power to impose a liability gives him a kind of practical authority in a limited domain over the defendant. To have an authority is to have a normative power to affect what those over whom one has such a power have reason to do. If, in exercising a morally legitimate power, X imposes a liability on Y, then Y has a moral duty enforceable by law to make appropriate repair. Your promising to do something for me might ground such an authority. Your wronging me might confer a power on me as well. My being wealthier than you or others does not; it is what we in philosophy call a “reason of the wrong kind.” This is not a matter of corrective or distributive justice. One needs no theory of either sort to know that one’s relative financial standing is not a ground of practical authority—the normative power to alter what others have reason to do.

But if wealth is not a ground for conferring a power to impose a liability, then why should a richer plaintiff receive more in compensation than a poorer one? The nature of the evil one is entitled to impose depends on a range of factors that are connected ultimately to what we take the underlying wrong of torts to be. Without going into very much detail here, my view is something like this.

Someone who commits a tort against me robs me of the control I have over some of the resources otherwise at my disposal. If you negligently destroy my car or my house, then you have deprived me of those resources. Those resources figure in my projects and plans. Decisions about what to do with those resources are mine (subject to constraints), not yours. When you negligently injure me, you interfere with my autonomy in the sense of the control I have over resources at my disposal. The remedy I am due should respect this fact by returning to me my lost resources, or their monetary equivalent. I am entitled to demand that you replace what I have lost in a way that respects my control over those resources. The damage remedy provides me with the resources to reconfigure my plans, projects, and life in light of a new set of resources. So there are very good reasons for thinking that fair compensation for different victims will differ. The grounds of the judgment are the same; the content of the liability the victim is entitled to impose need not be.27 There is no misalignment. Nor is the fact that there is no misalignment a matter of corrective or distributive justice.

B. Offsetting Risks

Turn now to the offsetting risks case.28 The offset in risks reduced is reflected in the Hand formula but not in the damage award. In order to be aligned, those risks should be reflected in both the Hand formula and the damage award, or in neither. Porat rejects the second possibility and thus endorses the first.29 Suppose a dangerous driver increases the risk of damage to the victim in the amount of 500, but in doing so he reduces the danger to others by 400. Had he been able, cost-effectively, to eliminate the risk to the potential victim entirely then he would be negligent in any case. Suppose as a result of his negligence, the driver injures someone to the tune of 500. Under current tort law, the victim will be entitled to demand 500 in damages from him. And that is so even if the defendant reduced risk elsewhere by 400.

In that case, the defendant has actually only increased total risk by 100, not 500. Were the damage remedy and the Hand formula properly aligned, the defendant would be liable for one-fifth of the damages, not all of the damage: 100 instead of 500. In requiring the defendant to pay the victim 500, tort law creates a misalignment. The Hand formula and the damage remedy do not properly match up.

But it is just a mistake to think that the difference between the Hand formula and the damage award represents a misalignment. The question is whether the injurer has wronged his victim, and the example indicates that he has. To be sure, in wronging the victim, he has benefited others. But the point of the elements of the tort is to establish whether there are reasons to confer upon the victim a power to impose an evil upon the wrongdoer. The compensation the victim can rightly demand depends on what the injurer has done to him and not on the overall quality of the injurer’s conduct.

The defendant can wrong the victim even in cases in which his conduct produces more benefits than harms. This is true in cases like Vincent v. Lake Erie,30 Joel Feinberg’s famous “backpacker” example,31 and in the famous Trolley Problem32 as well. The ground of repair is the wrong done, and the scope of appropriate recourse is fixed by the harm done to the victim, not in the difference between the harm he suffers and the benefits others have accrued at his expense. On the way of thinking that Porat endorses, wronging me in ways that benefit others in excess of the harm that results to me should lead to no liability at all. This is old news to act-utilitarians, but that does not change the fact that in creating benefits to others, one has done me a wrong—and that wrong is the basis of my claim to repair. The loss to me helps to fix the content of the repair owed to me.

Perhaps the injurer should seek the aid of those he has benefited to discharge his obligations of repair. But there is nothing in the arrangement of elements in the tort of negligence that suggests that the victim’s demands for recourse are to be mitigated or otherwise muted by the benefits that the injurer’s mischief has bestowed upon others. Again, this is not a matter of corrective justice or of distributive justice. It is a consequence of the elements of the tort aligning with one another so as to render the law’s response in the form of a remedy justified or warranted.

Perhaps we can deepen this point by considering the Trolley Problem a bit further. The Trolley Problem raises a number of philosophically interesting questions. By and large, the central question that has preoccupied philosophers has been whether the bystander or switchman is morally permitted to switch the train so that it endangers the one to save the five.33 To be sure, we want to know whether what the bystander does is morally permissible, and if so, why. But that is hardly the only interesting question to which the problem gives rise. Here are some others. If the one has a reasonable belief that the bystander will redirect the trolley, can the one shoot the switchman in self-defense? Can the one shoot the five in self-defense and thus remove the reason the switchman has for putting the one in danger? If the one is not killed but maimed instead, does he have a claim to repair? If so, against whom is his claim—the bystander, or the five? And why?34

Suppose the bystander miscalculates the risks. He plausibly thinks that by redirecting the trolley much more will be gained than lost, but as luck would have it, he is wrong; the five would have each been injured—but only slightly. So in diverting the trolley he is making a mistake. He wrongs the one, but in doing so he completely eliminates whatever danger the five would otherwise have faced. When the one is injured or killed as a result of the bad decision of the injurer, should his recovery be set to reflect the benefits conferred on the five? Perhaps the bystander has grounds to enlist the aid of the beneficiaries of his decision in making good on his debt to the one he has wronged. Does the bystander have a right to demand their contributions to ease his burden? These are difficult issues, but they do not raise questions about the relationship of the wrongdoer and the wronged. They do not raise concerns about what the wrongdoer owes the victim,anymore than they raise concerns about what the victim is entitled to demand of the wrongdoer.

Nor is it helpful to suggest, as Porat does, that the best explanation of the fact that the wrongdoer’s debt to the victim is not reduced by the benefits his mischief bestows upon others is a matter of corrective justice.35 That way of thinking misunderstands the place of corrective justice in explaining tort law. The right way to think about the issue is much more straightforward. Just like ordinary morality, tort law recognizes that what a wrongdoer owes those he injures depends on the wrongs he has done them—the harms he has caused them—and not in the first instance on the harms and wrongs he has done them minus whatever good he has done thereby.

It is important to distinguish between the deontic and nondeontic areas of the normative landscape. Sometimes we assess behavior as careless, inattentive; but we recognize that even careless and inattentive behavior may provide benefits—and not merely to those who benefit by saving the costs of greater attentiveness or care. Other times, we characterize our conduct in terms of duties and rights, powers and liberties: claims we have against others, authority we have over them, and demands that we can stand or call upon. Roughly, the latter is part of the deontic area of the normative landscape; the former is not, or at least need not be. The benefits my wrongdoing confers on others may appropriately figure in an overall assessment of what I have done—whether and to what extent I should be punished for my mischief, for example. But those benefits do not figure in what I owe you as a matter of the duty of care, and thus what I owe you in the event that I have failed to provide you with the care I owe.

Law and economics scholars focus on behavior as such—as good or bad according to a standard of efficiency—and see the law as responsive to defective behavior. In these scholars’ view, the law responds by providing incentives to induce better behavior insofar as doing so is feasible and desirable. It is no surprise, therefore, that economists of law in general see taxes, fines, subsidies, punishments, and tort and contractual liabilities as fungible instruments: different ways of inducing desirable behavior. The overriding concern is whether tort liability is being used well as a tool to achieve those goals. Porat’s point is that when there are misalignments, the tool is not being effectively used and adjustment is required. In the rare case, the misalignment can be justified, however, on other grounds—perhaps corrective or distributive justice.

But it should be plain by now that this picture is deeply mistaken. Tort law has an essential deontic structure reflected in its basic norms and its remedy and in the scheme of practical inference that connects the two. Corrective justice and alternative “moral” accounts of tort law like recourse theory are designed to explain that essential deontic structure. Corrective justice and recourse are not alternative “goals” of tort law such as optimal cost reduction, to be balanced against other goals or to override them when tort law adopts rules or procedures that make achieving efficiency difficult or unlikely.

Here is the other thing that proponents of law and economics like Porat miss. It does not follow from the fact that tort law has an essential deontic structure that the only explanation of it can be provided by deontic norms, e.g., norms of justice or right. There is no reason in principle why the deontic structure of tort law cannot be explained in nondeontic terms. In principle, different theories of tort law—from efficiency to corrective justice to recourse theory to what have you—can be offered to explain the pattern of reasoning that embodies the deontic normative structure of tort law. But one cannot hope to explain tort law without appreciating that it has a distinctive normative structure and that this structure tracks the distinctive place that tort law occupies in our normative landscape.36 For tort law is not a taxing system; it is not a sanctioning system; it is not a pricing system; it is a torts system. Only when we identify the distinctive normative structure of that system can we appropriately explore the ends, goals, and social projects that we can pursue through it.

C. Injurer’s Self-Risk

Let us turn now to the alleged misalignment that results from the failure of the law to include in the Hand formula the risks that the injurer imposes on himself in the course of imposing risks on others.37 In Porat’s example, the victim suffers an alleged loss of 8 that would cost the defendant 10 to prevent. In that event the Hand formula would not require the defendant to take precautions. But suppose that the defendant’s activity imposes costs of 7 on himself. In that case the total costs of the conduct are 15, whereas the costs of precautions are 10, and in that case the Hand formula would require that the injurer take precautions. He would be negligent in the event he does not. However, because the law does not include the costs of the defendant’s endangering himself, the defendant does not take the precautions that efficiency requires of him. Porat sees this as a misalignment, but one that goes in the opposite direction from that in his other examples. The damage remedy correctly represents the relevant risks, but the Hand formula does not.38

It is worth noting that the law in fact takes into account the risks the victim imposes on himself in determining plaintiff negligence as a partial or complete defense. And it would seem reasonable that the law ought to consider the risks the injurer imposes on himself to determine whether he is negligent. After all, the real cost to the injurer of taking precautions is 3, not 10. It is not so much that the law fails (if it does) to take the costs of the harm to the injurer into account; it simply systematically misrepresents the costs to him of taking precautions. Once it represents those costs accurately, it is clear that the injurer in Porat’s example would be negligent. The cost of his taking precautions is not a forgone benefit of 10, but a forgone benefit of 3.

On this point, Porat and I have no disagreement worth arguing about. Still, unlike Porat, I do not think that there is a misalignment here. The problem is not systemic or structural. Rather, it is that the standard formulation of the Hand formula is inadequately nuanced in these cases.

We do disagree about how to represent the risks the potential injurer imposes on himself. For Porat, it is a matter of indifference whether the risks the defendant imposes on himself are represented on the precautions or the damages side of the ledger in the Hand formula.39 If self-risk is represented on the damages side, the injurer is negligent because 10 < 8 + 7; if the injurer’s risk to himself is represented on the precautions side of the ledger, the injurer is negligent because 10 – 7 < 8.40

To be sure, we get the same result in both cases, but it is not a matter of indifference how we represent self-risk in the Hand formula. The risks the injurer imposes on himself can only be represented on the precautions side of the ledger. That is because the question in tort is whether the injurer has taken the plaintiff’s interests appropriately into account—not whether he has been sensitive to the overall riskiness of what he has done. The latter bears on the overall assessment of his conduct, not on the question of whether he has attended appropriately to the plaintiff’s interests in the security of his person and property. To be sure, the danger the injurer imposes on himself may contribute causally to the danger he imposes on the victim. But they are risks he imposes on himself, and as such should be reflected in the actual costs to him of taking precautions. The risks the defendant imposes on himself are not damages to the victim. The risks the defendant imposes on himself are not themselves grounds for the victim’s right to redress. I do not disagree with Porat insofar as we both agree that those risks should count. We differ, however, in the manner in which they should count. They bear on the costs of precautions to the defendant; they are not part of the harm to the victim; and it is the failure adequately to respond to the victim’s interests that is the source of the right to demand repair. This is not just my point. It is my point because it is tort law’s point.

D. Proving Causation

Let us turn finally to the hospital case.41 According to Porat, the culprit is the preponderance-of-the-evidence rule (PER) to the extent that it figures in determining whether the causation requirement has been satisfied.42 To appreciate Porat’s worry, assume that a certain percentage of the doctors treating their seriously ill patients are negligent in the care they provide. On average, though, in only 30% of the cases is the negligence of the doctor the reason a patient fails to recover. Most of the time the illness runs its course no matter the quality of care the doctor provides, and those stricken by it fail to recover. Because in each case the likelihood that the patient’s failure to recover is due to the doctor’s negligence is so low, no patient can establish that the negligence is the cause of his ultimate demise.43 No patient victimized by negligence can expect to recover, and no negligent doctor is threatened by liability for his wrongdoing. This is so even though some nontrivial number of times the failure of patients to recover will be the result of doctor negligence.44

The opposite obtains in the event that it is 70% likely that patients will recover, including those cases in which doctors are negligent.45 Since the probability of recovery is so high in these cases, any time a patient who has received negligent care fails to recover from his illness, it is more likely than not that the doctor will be found liable.46 For in such cases, it will be very likely that the patient’s failure to recover (which is unusual) would result from doctor negligence. Thus, negligent doctors will always be liable even though, in some nontrivial percentage of the cases in which they are negligent, their negligence is not the reason the patient fails to recover.

In one scenario, the doctor faces no real threat of liability, whereas in the other he faces a threat of being held liable all the time. In neither case do his prospects align with the expected likelihood of liability as expressed in the Hand formula. Thus we have a misalignment in Porat’s sense of the term.47 The question is whether we have a misalignment from the point of view of tort law.

The hospital case differs from the other cases that we have considered insofar as it involves the likelihood of recovery of patients—taken as a whole—suffering from an illness, and the likelihood of doctors—again, taken as a whole—being responsible for the failure of patients to do so. This fact allows Porat to aggregate the cases and discuss them in terms of the likelihood on average that the failure to recover is owed to a doctor’s negligence.48 Indeed, I believe that Porat’s argument gains whatever traction it has precisely because the case involves this sort of aggregation.49

Porat suggests that whenever there is a downward bias in the probability of causation—that is, when the likelihood of being able to prove causation is under 50%, which is the case in his example—the way to align the elements is to abandon the preponderance-of-the-evidence rule in favor of a probabilistic recovery rule (PRR):

  Under this rule, liability is imposed and damages are awarded in the amount of the harm done multiplied by the probability that it was caused by the defendant’s wrongdoing. The PRR could restore alignment but only if applied both when probability of causation is less than 50% and when it is greater. In practice, however, the courts apply the rule almost only to cases with a less-than-50% probability of recovery.50  

If I understand him correctly, Porat’s view is that the PRR would be preferable to the PER if courts would apply it in cases both in which the probability is greater than 50% and in which the probability is less than 50%. In other words, were the PRR applied all the time it would be preferable to the PER. It may not be preferable otherwise.

Let us see what it would mean to apply the PRR in all cases and then see whether it would be preferable to the PER under those conditions. Suppose that the damages to the plaintiff in every case are 1000. Now imagine the following cases. In one, the probability that the defendant caused the harm is 10%; in the next, it is 20%; the next 30%; then 40%, 50%, 60%, 70%, 80%; and finally 90%. If we apply the PRR, then the first defendant will pay 100 in damages, the next 200, the next 300, and so on. In the case in which it is 90% likely that the defendant caused the plaintiff’s injury, he will be saddled with only 900 of the 1000 damages that he very likely caused. That is how the PRR works, and now the question is whether there is anything to be said for it. 51

The suggestion that we should replace the PER with the PRR can make sense only if these rules are alternative approaches to the same problem. We know that the PER is an approach to an epistemic problem, namely: the degree of confidence in the truth of an assertion that is necessary in order rationally to accept it as true for the purposes of tort law. The PER is thus a rule governing belief acceptance. It holds that in the context of a tort action, one should accept a proposition as true (that is, believe it) only if the evidence renders it more likely than not to be true. In contrast, the PRR is not a rule governing rational acceptance. In fact, it has nothing at all to do with belief. It is a rule for imposing liability. In the PER, probabilities figure into determining the rationality of accepting a belief as true; in the PRR, probabilities figure into determining what one’s liabilities should be. So, taken literally, the PRR is not an alternative to the PER, and thus the idea that there are cases in which it would be better to replace the PER with the PRR involves a category mistake. It is thus incomprehensible and not merely indefensible.

This implies that if we are to make sense of Porat’s proposal we would have to reinterpret the PRR as an epistemic principle and not a liability principle: that is, we would have to treat it as a rule about rational acceptance of propositions as true (for the purposes of tort law). The problem is that so understood the PRR would be extremely implausible and there would be no reason to prefer it to the PER, as the following argument clearly demonstrates.

If it is 10% likely that I caused your harm of 1000, applying the PER would mean that without more no one would have adequate reasons for accepting as true the proposition that I caused your harm. Now if we treat the PRR as an epistemic rule, then one should take the fact that it is 10% likely that I caused you harm of 1000 as equivalent to the claim that everyone has adequate reason for believing that I caused 10% of your damage—or 100. Likewise, if it is 20% likely that I caused your harm of 1000, the PER implies that no one has adequate grounds for accepting as true that I caused your harm. But interpreted as an epistemic rule for accepting beliefs, the PRR says that everyone has reason enough to accept as true the claim that I caused 20% of your damage—or 200.

The problem is that if it is 10% likely that I caused your harm, one thing we can be pretty sure of is that it is not more likely than not that I caused any of your harm, let alone a particular fraction of it. Now consider two cases. In one it is 10% likely that I caused your harm; in the other it is 20% likely that I did so. On the PPR we would all have good reason for thinking that I caused 10% of your damage in the first case and equally good reason for thinking that I caused 20% of the damage in the second: reason enough for holding me liable for 10% of the damage in the one case and 20% in the latter. But this cannot possibly be right, since in both cases no one has very persuasive reasons for thinking that I caused you any damage at all.

So far we have identified two possibilities. In one, the PER and the PRR are not alternative rules applying to the same problem, so the claim that there are circumstances under which we should adopt the second over the first involves a category mistake. In the other, we reinterpret the PRR as an epistemic rule so that the two are genuine competitors. The problem here is that the PRR is wildly implausible as an epistemic rule, and it is hard to imagine any circumstances under which we should replace the PER with it.

There is one last alternative interpretation of what Porat might have in mind, and that is that there are cases in which we should simply abandon the PER in favor of the PRR. That is, if we worry about evidence as a basis of liability, we will end up with too little liability for deterrence. So instead of worrying about matters of rational belief acceptance (the domain of the PER), we should instead focus entirely on liability (the domain of the PRR). The obvious problem here is that the proposal would have us assigning liabilities without any of the grounds that tort law requires for doing so. If it is 1% likely that I caused you harm, that hardly provides any reason for thinking that I caused any harm to you, let alone 1% of it. And that, of course, is the problem with the PRR and why courts should be (and are) reluctant to apply it.52 Porat’s argument is simply hopeless.

Anyone who is serious about the PRR should think about how it would apply in the criminal law context. In that context, it is going to be very hard to establish guilt beyond a reasonable doubt, and there are bound to be crimes that are inherently more difficult to prove than the average. In those cases, why not just discount the punishment by the probability that the defendant did the crime that warrants the penalty, and simply impose on him the suitable discounted prison term? Hopefully, the folly of this conclusion makes us realize that a rule that may well be appropriate for assigning taxes, licensing fees, or prices need not be suitable for assigning liability in tort law or in criminal law.

Not only does Porat confuse the deontic with the nondeontic aspects of the normative landscape; he conflates the epistemic with the normative. It is no wonder he finds so many problems to worry about where few if any are actually to be found.

To be sure, the rules of evidence—whatever they may be—will set a barrier below which one would not have adequate evidence for accepting a proposition as true. That in turn will mean that certain genuine wrongs will not be righted and that certain harms will go unrepaired. We can improve upon that situation in a number of ways. The problem may be so serious that we would do well to replace tort law with some other system altogether—something akin, perhaps, to the New Zealand53 system or perhaps with the “at fault” pool proposal that Guido Calabresi and I each have at one time or another offered as promising alternatives to the tort system.54 But as long as we are committed to the values embodied in the law of torts, this, as they say, is a price we will have to bear.


Ariel Porat’s paper is a subtle and sophisticated intervention in the law and economics of negligence law. It builds on concepts and ways of thinking that are familiar to that tradition and extends familiar arguments to draw unfamiliar conclusions. The problem is that the overall framework within which he is working is seriously misguided. He is right that there is a notion of alignment in torts, but he does not characterize that notion correctly. The tort of negligence is characterized by a set of conditions or elements that must each be satisfied if the remedy is to be warranted. These elements must align in the sense that they, taken together, if satisfied, explain why the remedy is called for.

Next, like his fellow travelers in the law and economics literature, Porat mischaracterizes the remedy. The remedy is not the imposition of costs on the defendant, but the conferral of a power on the plaintiff to impose a liability on the defendant. This means, of course, that the elements of the tort of negligence align only if and to the extent that they explain why the conferral of a power held by a private person (the plaintiff)—enforceable by the coercive power of the state—is justified.

And finally, again like his fellow travelers in the law and economics camp, Porat misunderstands the place in the normative landscape that tort law occupies. Tort law’s concern is not with bad behavior as such—understood as conduct that is too risky or not worth its costs. The normative landscape of tort law has a deontic structure. Its norms impose relational duties to take adequate regard for the interests of others; its remedies confer powers to impose liabilities. It reflects that part of our normative landscape that has to do with what we owe one another and what we have a right to demand of others in how they regard our interests in person and property.

Only once we understand what tort law is can we have a real debate about what normative theory makes the best sense of it. It is well known that I do not believe that economic analysis makes the best sense of tort law and that I think that corrective justice does. I may well be wrong. But nothing in this Response hinges on my views about corrective justice. Everything I have argued for can be summed up in one sentence. Our primary responsibility as tort theorists is to properly characterize the normative structure of the institution of tort law before we can adequately attend to what tort law is well or poorly suited to accomplish.

Jules Coleman is the Wesley Newcomb Hohfeld Professor of Jurisprudence at Yale Law School, Professor of Philosophy at Yale University, and Visiting Distinguished Scholar-in-Residence at the University of Miami School of Law.

Preferred citation: Jules L. Coleman, Mistakes, Misunderstandings, and Misalignments, 121 Yale L.J. Online 541 (2012),