FACING UP TO RISK - Oxford Academic

FACING UP TO RISK

Barbara H. Fried*

A B S T R A C T

By banishing decision-making in the face of uncertainty (risk) to the margins of tort

theory, nonconsequentialist legal philosophers have obscured the quotidian, unavoid-

able, and ubiquitous tradeoffs we face in almost every arena of life. This article explores

the historical antecedents of the marginalization of risk in contemporary moral philoso-

phy, and details how legal philosophers have deflected the challenge that risk poses to

any nonconsequentialist theory of tort liability.

In 1972, a new Ford Pinto stalled in the middle of a highway, causing the car

behind to rear-end it. The force of the collision pushed the Pinto’s rear-

mounted gas tank forward and into the differential housing, puncturing it.

Fuel spraying from the punctured tank ignited inside the car, engulfing the

car in flames. One of the two occupants died from the resulting burns. The

other, a 13-year-old boy, was severely disfigured for life, requiring multiple skin

grafts over the next decade.

In the lawsuit the heirs and survivors subsequently filed against Ford for

defective product design, it emerged in pretrial discovery that before the

Pinto ever went into production, Ford knew about the danger that a rear-end

collision would explode the gas tank. Indeed, its own crash tests revealed the

engine was vulnerable to puncture and explosion at crash speeds as low as 20

MPH. Internal memos from Ford engineers to top management estimated that

at a cost of $5–$11 per car, the Pinto could be redesigned to decrease the

* Stanford Law School, Stanford, CA 94305, USA, Email: [email protected]

Barbara Fried, William W. and Gertrude H. Saunders Professor of Law, Stanford University. Earlier

versions of this article have been presented at a number of workshops and symposia, including UC

Irvine Legal Philosophy Colloquium, University of Arizona Law School Faculty Workshop, Berkeley

Political Philosophy Workshop, Stanford Law School Faculty Workshop, Yale Legal Theory

Workshop, University of Amsterdam Faculty Workshop, American Studies Workshop at

Princeton University, Legal Theory Workshop at University of Toronto Law School, Queens

University Law School, the Center for Ethics and Public Affairs at Tulane University, and the

2015 Bell Distinguished Lecture in Law, College of Wooster. I am grateful to the organizers and

participants for their very helpful comments.

� The Author(s) 2019. Published by Oxford University Press on behalf of The John M. Olin Center for Law, Economics andBusiness at Harvard Law School.This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License(http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in anymedium, provided the original work is properly cited. For commercial re-use, please contact [email protected]:10.1093/jla/laz003

Dow

nloaded from https://academ

ic.oup.com/jla/article/doi/10.1093/jla/laz003/5549883 by guest on 12 July 2022

probability of explosions significantly, saving an expected 180 lives. Ford man-

agement, estimating that the redesign would cost the company a total of $137

million but save it only $49.5 million in projected damage awards from the

incremental deaths, injuries, and property damage caused by the design defect,

declined to make the change.

The jury found against Ford, awarding the plaintiffs roughly $3 million in

compensatory damages, and the boy an additional $125 million in punitive

damages, an amount subsequently reduced on appeal. (Grimshaw v. Ford Motor

Co., 119 Cal. App. 3d 757, 808 (1981)).

In the years since, Ford’s conduct in the Pinto case has come to stand for all

that is wrong with cost/benefit analysis, the decision procedure that Ford em-

ployed in deciding not to redesign the Pinto. Taking the generally received view

of Ford’s behavior as accurate, it is easy to understand why Ford was widely

regarded at the time as the poster child for bad corporate actors.1

But the moral that many have drawn from the Ford Pinto case goes well

beyond the particular facts of the case. Put simply, it is that trading off lives for

any amount of money or other lesser goods is wrong. In one form or another

that sentiment underlies much of the vast literature on harm to others in Anglo-

American moral and legal philosophy over the past half-century. It also, not

coincidentally, drives the intuitive moral responses of many if not most

Americans, at least when the tradeoff is made explicit, as it was in the Pinto

case. Its intuitive appeal is obvious. But in a world of epistemically indetermin-

ate consequences—which is to say, the world in which we live—it cannot

supply a general guide to action.

I explore here why that is so, why many moral and legal philosophers along

with a large portion of the general public continue to believe otherwise, and the

social costs to all of us when that belief asserts itself in law, politics, and policy-

making.

Before doing so, a few clarifications on terminology and the scope of the

argument.

The term “consequentialism” refers broadly to a system of moral principles

that judges the goodness or badness of an act by its expected consequences. I use

1 It was alleged, among other damning facts, that Ford’s cost/benefit analysis was faulty; that the Pinto

was significantly less safe than most other cars on the road, likely making its behavior legally neg-

ligent even without a faulty cost/benefit analysis; that the crash engineers at Ford strongly urged

redesign, but were overruled by top management, which was notoriously indifferent to safety con-

cerns; that Ford knew that federal safety regulations then pending would require it to protect the

integrity of the gas tank in crashes up to 30 MPH, but chose to wait until the regulations were

adopted to comply with the standards; and that Ford withheld information about the risks for fear of

bad publicity. In the years since the case was decided, some have questioned the accuracy of the

received account. The best known of these revisionist accounts is Schwartz (1991).

176 ~ Fried: Facing Up To Risk

Dow



the word to refer to a particular form of consequentialism: one that measures

the expected consequences of a given act by the aggregate (expected) costs and

benefits to everyone potentially affected by it. The measure of costs and benefits

generally employed is a welfarist one: decreases and increases in individual

welfare. In common parlance in law and public policy, the welfarist version

of consequentialism is generally referred to as “cost/benefit analysis.” I adopt

that meaning here, but nothing turns on the choice. The critical feature of

consequentialism for current purposes is that it aggregates all expected conse-

quences of an act to determine the optimal course of action, however those

consequences are measured.

“Nonconsequentialism,” as the name suggests, is defined in opposition to

consequentialism. In legal and moral philosophy, “nonconsequentialism” typ-

ically refers to a system of moral principles that measures the goodness or

badness of an act in significant part by whether it respects others’ individual

rights (fulfils one’s duties to others). Critically, for present purposes, “respect”

is generally taken to mean that no individual claim of right can be traded off

against the claims of others—that is to say, none can be aggregated. Law is

speaking the language of nonconsequentialism whenever individual rights are

taken to operate as a side constraint on promoting aggregate social welfare or

other values, or individual wrongs are regarded as deserving of punishment on

moral grounds (e.g., retributive justice in criminal law and corrective justice in

tort law).2 An absolutist reading of the language of the First Amendment

(“Congress shall make no law . . . abridging freedom of speech . . .”), or of the

Second Amendment (“the right of the people to keep and bear Arms shall not be

infringed.”) construes each provision as imposing an absolute side constraint on

any regulatory measure (e.g., campaign finance laws, banning private owner-

ship of military-grade high capacity assault rifles) that trades off those rights

against the “public interest,” however defined and however weighty it may be.

Most nonconsequentialists hold mixed views, permitting aggregation, for

example, when the number of individuals whose interests would otherwise be

ignored becomes large enough (“threshold deontology”), or when the govern-

ment rather than private parties is the one making the tradeoff (so-called “gov-

ernment-house utilitarianism”). I use “nonaggregation” here to refer to any

decision rule that evaluates our conduct at least in part without regard to the

numbers of people it puts at risk or the probability that they will be harmed. To

the extent that nonconsequentialists hold mixed views, for current purposes I

2 The term “side constraint” is commonly used in philosophy for any value that trumps (takes lexical

priority over) competing values.

2018: Volume 10 ~ Journal of Legal Analysis ~ 177

Dow



am interested only in the portion that is ostensibly governed by nonaggregative

rules. The rest has already been ceded to aggregation.

Finally, in pressing the futility of nonconsequentialists’ opposition to aggre-

gation in this context, I do not mean to deny that nonaggregative principles can

and do play a significant role in society. Whenever we refuse to give weight to

certain kinds of preferences—of rapists to rape, of vigilantes to take justice into

their own hands, of racists, homophobes, misogynists to refuse to hire qualified

members of disfavored groups, of state actors to censor unwelcome speech, of

the public to curtail a disfavored individual’s right to a fair trial—our motiv-

ation for doing so is typically nonaggregative, and the decision rule that results

is at least partially nonaggregative. That is to say, rights can and frequently do

operate as partial side constraints on permissible tradeoffs.3

But the voluminous and still growing philosophical literature on harm to

others has been concerned principally not with conduct that is deemed per se

impermissible (rape, murder, intentional torts, abrogation of due process

rights) but rather with conduct that is, in Johann Frick’s words, “intuitively

permissible” but potentially harmful to others (2015, pp. 205–211). One of the

principal sources of confusion in both the legal and philosophical literatures

over the past half-century has been the failure to keep the two categories dis-

tinct. My argument here is limited to the latter.

Under an appropriately broad definition of harm, the latter category includes

almost every noncriminal activity we engage in: driving, riding a bike, reroofing

our house, practicing the piano in an apartment building without soundproof-

ing, living a life that generates a non-zero carbon footprint. It accounts for the

vast majority of risk-regulation decisions routinely made by the government:

setting standards governing pollution, occupational safety and health, product

safety, nuclear energy, etc.; allocating budgets among competing needs and

competing claimants; and articulating standards of due care in common law

torts. And, importantly, it accounts for virtually all of the tragic tradeoffs that

the nonconsequentialist literature is preoccupied with. This is hardly surprising.

If the individual prerogative at stake on one side is the right to punch people

you don’t like in the face, and on the other the right not to be punched just

because someone doesn’t like you, to nonconsequentialists this does not count

as a tragic choice—indeed, morally speaking, it is no choice at all.

3 I am setting to the side some deep questions about whether the scope of the rights we protect against

utilitarian objections is itself determined by some version of aggregation (e.g., “reasonableness” in

tort law). For present purposes, I am happy to stipulate that some fundamental norms in society can

be explained and defended only in nonconsequentialist terms. In addition, every comprehensive

moral theory, including welfarism, is vulnerable to the critique that it smuggles in other norms at

critical junctures. For a helpful discussion of this point, see Kelman (2005).


Dow



Finally, while my own views are strongly consequentialist, I don’t press my

underlying normative disagreements with nonconsequentialists here. My argu-

ment, rather, is that in a world of indeterminate consequences, we cannot

logically resolve the vast majority of interpersonal conflicts in civil society with-

out resort to aggregation.

1 . T H E M O R A L T E R R A I N O F I N T E R P E R S O N A L T R A D E O F F S

To elucidate how the nonconsequentialist literature has sidelined the problem

of indeterminate consequences (“uncertainty” or “risk”4), I borrow Philip

Tetlock’s taxonomy of tradeoffs (2000, pp. 853–870).5

Tetlock divides peoples’ interests into sacred and nonsacred values, which

correspond (roughly) to fundamental versus more trivial interests. He identifies

three categories of tradeoffs, based on the interests implicated on each side:

routine tradeoffs, taboo tradeoffs, and tragic tradeoffs.

Routine tradeoffs are tradeoffs between two nonsacred values. Crucially, non-

consequentialist philosophers typically take routine tradeoffs to include not just

those in which the underlying interests at stake on each side are nonfunda-

mental (e.g., allocating tickets to an oversubscribed sporting event) but also

those in which sacred values are implicated on one or both sides, but the

probability of harm to those values is very low (e.g., flight paths that cross

over a residential neighborhood, thereby putting the population at risk of

death in the highly improbable event of a crash). At least when judging actions

from an ex ante perspective (that is, before we know whether the plane actually

kills anyone), most nonconsequentialists have ceded routine tradeoffs to aggre-

gation, either because they have concluded that the choice among the interests

at stake does not pose a moral dilemma and hence can be handled by any

noninvidious decision rule, including aggregation, or because they have reluc-

tantly concluded that, in the case of decision-making under uncertainty, we

have no choice but to aggregate.

At the other extreme are taboo tradeoffs, which are tradeoffs between sacred

and nonsacred values. Taboo tradeoffs describe the realm in which interper-

sonal tradeoffs are forbidden on nonconsequentialist grounds, because the

sacred value(s) on one side always trump the nonsacred value(s) on the

other, irrespective of the numbers involved. The common view among

4 Risk and uncertainty have different technical meanings. Roughly, risk refers to gambles with known

odds, and uncertainty to gambles with unknown odds. For present purposes, that distinction is

irrelevant, and for ease of exposition I use the terms interchangeably here.

5 I am grateful to Rob MacCoun for pointing me to Tetlock’s formulation.


Dow



contemporary nonconsequentialists that more serious harms trump less serious

harms, irrespective of the numbers of affected individuals, rules out one type of

tradeoff as taboo (Scanlon 1998). Thus, if we must choose between saving

10,000 from paralysis and one person from death, we must save the one.

Libertarian property rights as absolute side constraints rule out another. It

doesn’t matter how much the character of the neighborhood will be compro-

mised and neighbors’ home values depressed if we allow someone to put up a

commercial building on a quiet, low-rise residential street: a man’s home is his

castle, even if the man is an absentee commercial landlord and the castle in

question is a 50-story sliver building.

Taboo tradeoffs of all sorts are the principal redoubt of nonconsequentialism.

As a result, nonaggregative principles have a significant role to play in regulat-

ing potentially harmful conduct only if the domain of taboo tradeoffs is itself

significant. The nonconsequentialist literature generally presumes that it is.

In between routine tradeoffs and taboo tradeoffs lie tragic tradeoffs, which

are tradeoffs between two sacred values (a life versus a life, a life versus ten lives,

a life versus a lifetime of extreme pain, etc.). May we kill one to save a hundred?

May we privilege the survival of family members or friends over strangers? May

we bomb enemy targets to protect innocent civilians, knowing that some

number of other innocent civilians will be collateral damage? May we (must

we?) prefer known victims to unknown ones? How do we allocate scarce organ

donations among those who will die without a transplant? In T.E. Hill’s words,

cases like these are “notoriously hard” to resolve by appeal to the principle that

“one must not trade or sacrifice the dignity of anyone for ‘more dignity’ for

others,” because both of our available choices will result in serious harm or

death to one of the two interested parties (1992, p. 293). In such a situation, we

cannot “fully respect the dignity of all.”

For a variety of reasons, tragic choices of this sort infrequently arise in law or

popular thought.6 In moral philosophy, however, an enormous amount of

high-octane intellectual talent over the past sixty years has been devoted to

figuring out whether these hard cases can in fact be resolved by nonaggregative

principles, or instead have to be ceded to some form of aggregation. My concern

here is not with the answer, but with the negative pregnant that justifies the

6 The chief reason is that the choices we face in law and life rarely read as the paradigmatic ‘tragic

choice’ in the philosophical literature: do we save innocent A or do we save innocent B? In law, most

cases are resolved in accordance with prospective standards that have invisibly embedded within

them whatever tradeoff society has decided to make—e.g., what is required for a self-defense claim

in a murder case, burdens of proof, which embody a particular tradeoff between false positives and

false negatives. When they do present themselves as the paradigmatic tragic choice—e.g., the organ

transplant case—people typically focus on the plight of whatever person has been made salient by

the media, and ignore the implicit tradeoff of others whose lives will be sacrificed if we save the one.


Dow



attention that has been lavished on them: once nonconsequentialists figure out

what to do with “tragic tradeoffs,” the other two categories will take care of

themselves. (In the case of routine tradeoffs, we are permitted to aggregate; in

the case of taboo tradeoffs, we are not.)

This view of the moral terrain, I suggest, gets things exactly backwards. The

high-stakes tragic choices that populate the philosophical literature—do we

save one from death or 100,000 from a broken leg—may be difficult to resolve

without resorting to aggregation. But the tragic choices we actually face in life,

in which the outcomes of available choices are all indeterminate, are impossible.

As a result, far from offering a comprehensive alternative to aggregation, non-

aggregative principles can operate only at the margins of the real-world tragic

tradeoffs we face, relegating the philosophical literature on tragic choices to a

luxury indulgence of our nonaggregative instincts.

Thus, nonaggregative principles in this domain are viable in the minimal

sense of operationalizable only as long as they are confined to a very small and

rarified set of tragic choices. And they will seem to present a viable, systematic

alternative to aggregation only as long as attention is confined to that set. In my

view, that blinkered perspective describes most of the vast nonconsequentialist

literature in law and moral philosophy over the past sixty years on our duty not

to harm (or duty to rescue) others.

How then could it be that most nonconsequentialists (Kantian, contractual-

ist, and libertarian) have assumed otherwise?7 The answer, I suggest, lies in two

features of nonconsequentialist thought that have led philosophers and lay-

people alike to overestimate vastly the potential domain of nonaggregative

principles.

The first, relevant to the philosophical literature, is the disproportionate

amount of attention given to simple trolley-type tragic choices, in which the

consequences of each choice are stipulated to be known with certainty. While

such choices are notoriously hard—indeed, impossible—to resolve without

trading off one person’s vital interests for another’s, in the stylized versions

that populate the philosophical literature they can be resolved by reference to

7 A partial list of contemporary moral and legal philosophers who fall in this camp would include

Judith Thomson, Frances Kamm, T.M. Scanlon, Robert Nozick, Richard Epstein, Jules Coleman,

Joel Feinberg, David Owen, John Goldberg, Ben Zipursky, Arthur Ripstein, Ernest Weinrib, Johann

Frick, Sophia Reibetanz Moreau, and Rahul Kumar. I discuss many of their arguments in more detail

in Fried (forthcoming 2020). It should be noted that Kant himself believed that the rightness or

wrongness of an action does not depend upon its consequences. It depends instead on whether the

action complies with the ‘maxim’ guiding it. In this one respect, Kant shares more with ex-

pected-value consequentialists than with contemporary Kantians, although the content of their

respective ‘maxims’ is worlds apart. I thank an anonymous reviewer for pushing me to clarify

this point.


Dow



other nonaggregative criteria (a maximin decision rule, doing versus allowing,

intending versus foreseeing harm, contributory negligence, identifiable versus

nonidentifiable victims, etc.). They can, that is, if we do not inquire beyond the

stipulated boundaries of the problem.8

The second and more important explanation is what is missing from the

nonconsequentialist account, both in its academic version and lay morality: the

problem of decision-making under uncertainty (risk). For a long time, phil-

osophers ignored risk entirely. In recent years, a small group has turned its

attention to the problem, but (in my view) has yet to show that nonaggregative

principles can supply a viable, systemic solution to it.9

2 . T H E P R O B L E M O F R I S K

As Sven Ove Hansson (2007, p. 30) has remarked, “Throughout the history of

moral philosophy, moral theorizing has for the most part referred to a deter-

ministic world in which the morally relevant properties of human actions are

both well-determined and knowable.” Given that the consequences of our ac-

tions are always epistemically uncertain ex ante, nonconsequentialists’ fixation

on choices with specified, determinate outcomes seems to require some

explanation.

In his well-known article “Conspiracy and the Paranoid Style: Causality and

Deceit in the Eighteenth Century,” Gordon Wood (1982, p. 413) supplies an

historical one, which traces back to the seventeenth century. The mechanistic

conception of the physical world produced by the Scientific Revolution, Wood

argues,

became the paradigm in which the enlightened analysis of all behavior

and events now had to take place. Cause was something that produced

an effect; every effect had a cause; the cause and its effect were

integrally related. Such thinking created a new world of laws,

measurements, predictions, and constancies or regularities of behavior

all dependent on the same causes producing the same effects.10

Causal relationships that lacked such regularity and predictability, it was

thought, did not deserve the name of science. In Samuel Stanhope Smith’s

8 For an extensive discussion of the trolley literature, see Fried (2012b).

9 For a discussion of that literature, see Fried (2012a; 2016). For reasons I discuss there, I don’t believe

such a solution is logically possible.

10 I am grateful to Sharif Youssef for bringing to my attention the relevance of Wood’s thesis to mine.


Dow



words, chance was “only a name to cover our ignorance of the cause of any

event.”11 (1812, p. 22).

Applied to the realm of human action, the mechanistic story located the first

mover in human intentions and motivations, from which action and its con-

sequences followed. In Wood’s (1982, pp. 412–413) words,

Although the new science tended to remove man from the center of

the physical universe, at the same time it brought him to the center of

human affairs in ways that even classical and Renaissance thinkers had

scarcely conceived of. It promised him the capacity to predict and

control not only nature but his own society, and it proceeded to make

him directly and consciously responsible for the course of human

events.

God still hovered in the background as the theoretical First Mover. But God

was notably absent from the only causal story most people cared about: the

story of cause and effect in everyday human affairs.

The Enlightenment’s mechanistic account of cause and effect in the physical

universe has given way over the past two centuries to more complex and prob-

abilistic understandings of causation. But in the social world, the mechanistic

account remains very much alive. In modern analytic moral philosophy, it is

lodged in two foundational assumptions, the first explicit, the second not: our

willing an action is the (morally relevant) cause of the action, and the conse-

quences of our actions follow deterministically from the actions themselves.

The first of these assumptions supplies the moral undergirding for the

Kantian notion of a perfected will, and with it the congeries of condemnatory

responses that reign supreme in contemporary American politics and common-

sense morality: blame, retribution, guilt, etc. If we are not the cause of our own

intentions—if they are caused by God, or social forces, or biological ones—then

of what moral interest is the quality of our willing or the ends we seek to bring

about? The Enlightenment view of the individual will as an uncaused cause may

eventually buckle under mounting evidence to the contrary from neuroscience,

genetics, etc., but it has so far managed to ward off similarly powerful evidence

of social determinism.12

My concern here is with the second of these assumptions: that the conse-

quences of our actions follow predictably and deterministically from the actions

11 In Wood’s words, “Just as devout Puritans believed that nothing occurred without God’s provi-

dence, so the liberal-minded believed that nothing occurred without some person willing it. Earlier,

men had sought to decipher the concealed or partially revealed will of God; now they sought to

understand the concealed or partially exposed wills of human beings.” (1982, p. 420).

12 For further discussion, see Fried (2013).


Dow



themselves. In contrast to the first assumption, which has been debated end-

lessly in contemporary philosophical arguments over free will versus determin-

ism and compatibilism versus incompatibilism, philosophers have said little

about the second. Indeed, the central role given to actions with (ex ante) well-

determined and knowable consequences is less an argumentative assumption of

moral philosophy than a near-universal and unexamined practice.

Those who have explicitly defended moral philosophers’ singular focus on

conduct stipulated to have (ex ante) determinate consequences have generally

done so on one of two mutually inconsistent grounds. The first is that the right

way to handle risk can be derived from the right way to handle determinate

harm, because the imposition of risk is itself a completed harm or other wrong,

or because the wrongness of risk is parasitic on the wrongness of the harm it

threatens.13 Hence, by resolving the latter we resolve the former as well. The

second is that choices made under conditions of epistemic certainty are a dif-

ferent moral kind from those made under conditions of epistemic uncertainty,

and that only the former are the proper subject of moral theory. The latter are

taken to implicate less lofty considerations.

Both arguments for the proposition that risk is unproblematic, in my view,

get things backwards. The moral principles nonconsequentialists have extracted

from their extended engagement with trolley problems, far from solving the

problem of risk, are viable only as long as they are not extended to garden-

variety risk. And, as I suggested above, if classic trolley problems present a

difficult choice for nonconsequentialists, garden-variety risks present an im-

possible one.

In On What Matters, Derek Parfit (2011, p. 162) offers a rare, explicit defense

of the “different moral kinds” argument:

In trying to answer [what acts are right and what wrong], it is best to

proceed in two stages. We can first ask which acts would be wrong if

we knew all of the morally relevant facts . . .. After answering these

questions, we can turn to questions about what we ought morally to do

when we don’t know all of the relevant facts. These questions are quite

different, since they are about how we ought to respond to risks, and

to uncertainty. As in the case of non-moral decisions, though these

questions have great practical importance, they are less fundamental.

These are not the questions about which different people, and different

moral theories, most deeply disagree. Given the difference between

these two sets of questions, they are best discussed separately. So I shall

often suppose that, in my imagined cases, everyone would know all of

13 For further discussion of the literature, see Fried (2012a).


Dow



the relevant facts. We can then ask what we ought to do in the

simplest, fact-relative sense.

This explanation raises more questions than it answers. Insofar as what is

wrong with (potentially) harmful conduct is the (potential) harm itself, it is not

apparent why “certain” and “uncertain” harms would raise “quite different”

moral problems. All acts involve consequences that are (ex ante) more or less

uncertain. Trolley problems, in which the consequences are stipulated to be

known with certainty ex ante, simply represent the limit case at one extreme.

Given that we are dealing with factually continuous phenomena, why would we

think they raise morally discontinuous problems? Why is the difference between

(say) a 95 percent probability and a 100 percent certainty as to any of the

morally relevant facts (the probability that harm will result, the identity of

the victim(s), etc.) a difference in kind rather than a difference in degree, and

a very slight one at that?

In addition, assuming that the two do present different moral problems, it is

not apparent why we should regard the moral problems raised by uncertain

harm as “less fundamental” than those raised by certain harm. As Parfit ac-

knowledges, it cannot be due to their relative practical importance. While cer-

tainty is king in the hypothetical world of trolley problems, in the real world, the

consequences of our acts are always uncertain ex ante. This is true even of harms

that are intended (in the strong sense of desired or the weak sense of foresee-

able). If I point a loaded gun at your head and pull the trigger, I am over-

whelmingly likely to kill or seriously injure you, but I am not certain to do so.

The gun could misfire; I could have been mistaken about whether it was loaded;

the bullet could be deflected by a metal plate in your skull. So also, if I divert the

trolley, I may believe I will thereby save five from certain death at the cost of one

life, but I can never be certain, ex ante or (in many cases) ex post. Perhaps

diverting the trolley will cause it to tip over before it reaches the one; perhaps if I

had done nothing the five would have seen the trolley in time and moved out of

the way. A fortiori, what is true of knowingly inflicted harms is true of accidental

harms. Thus, from an ex ante perspective, the problem of all harm, accidental or

not, is the problem of risk.14

In what sense then does risk present a “less fundamental” problem? Parfit’s

comment suggests two possible answers. One is that the solution to the problem

of risk cannot be of fundamental moral interest, because, as of the moment we

14 As David Luban (2009; 2014) has noted, this is true as well of the moral dilemmas of war that are

often taken to be the real-life analogues to trolley-type hypotheticals: whether it is permissible to

torture others to obtain information that will (‘certainly’) save innocent lives, or to kill enemy

noncombatants to save the lives of one’s own combatants or citizens.


Dow



must choose how to act, we lack a “morally relevant fact” bearing on the

wrongness of the act: what its actual consequences will be. If this is the claim,

it is problematic on a number of levels, but I focus on just one here. Whether

the actual consequences of an act are in fact morally relevant to its rightness or

wrongness needs to be established. If what we seek is an (ex ante) action-

guiding norm, they decidedly are not. (I return to this issue below.)

Alternatively, Parfit can be read to suggest not that the problem of risk is

unimportant, but that the answer is uncontroversial, at least as compared to

trolley problems. (“These are not the questions about which different people,

and different moral theories, most deeply disagree.”) If so, I am uncertain what

he has in mind, but perhaps it is the fact that many nonconsequentialists have

concluded that risk must be ceded to aggregation.

But, those who have given up on risk have not given up on nonconsequentialism,

implicitly assuming that a substantial domain of harm-producing conduct remains

for nonconsequentialist principles to operate on. I turn to that assumption now.

3 . R I S K A N D A G G R E G A T I O N

The problem that risk poses for nonaggregation is easily demonstrated. Suppose

you are planning to drive to the movies. You always drive carefully (holding in

suspension for the moment what exactly we mean by carefully). But you know

that even driving carefully, there is a 1 in 50,000 chance that in the course of the

trip you will seriously injure or kill someone. If you could know with certainty

that, of the tens of thousands of times that you are going to drive over your

lifetime, today is the day when driving will result in serious injury or death to

another, then you could simply stay home today. Doing so entails a personal

sacrifice, to be sure—you’ll have to put off going to the movie. But it is hardly a

tragic one, and it is one that aggregationists and nonaggregationists alike would

make. This is the seductive fantasy world into which trolleyology invites us: a

world in which all tradeoffs can be made by pairwise comparisons of the actual

harms that will result, allowing us to ignore all the other instances of identical,

risky conduct in which, as the odds would have it, no one ends up being

harmed.

But in a world in which you cannot know, ex ante, which (if any) of the tens

of thousands of times that you will drive over your lifetime you will end up

injuring someone, the chance that you take with others’ lives today is the same

as the chance you take every other day that you get in your car and drive

somewhere. As a result, whatever we decide about the permissibility of your

driving today will apply equally to all instances of driving. That reality leaves

nonconsequentialists with two unacceptable alternatives, whether they are


Dow



proceeding under the maximin decision rule common in contemporary non-

consequentialist philosophy (greater harms always trump lesser harms) or lib-

ertarians’ absolute prohibition on boundary crossing. If the mere possibility

that an act could result in death or grave harm to others is sufficient to rule it

out, then no act is permissible (except perhaps when necessary to save others

from harm more severe than the proposed act threatens to impose).15 On the

other hand, if the mere possibility that an act, undertaken with due care, could

result in serious harm to others is not sufficient to rule it out—if harm must be

absolutely certain to eventuate on this particular occasion—then in a world of

indeterminate consequences, all actions are permitted until it is too late to avoid

their consequences.

Writ large, the same two unacceptable choices present themselves when we

make the identical moral calculus over a large number of identical acts—for

example, in choosing rules of general application to govern whether and under

what conditions people may drive. The only difference between the two cases is

the mode of presentation. In large-number cases, the risk is conventionally

stated in frequentist rather than probabilistic terms. Thus, rather than asking

whether a 1 in 20,000 chance that Ms. Jones will injure or kill someone when she

drives X miles today is too great a risk to allow her to take, we ask whether the

statistical certainty that 1 out of every 20,000 car trips of X miles will result in

serious injury or death is too great a harm to allow the general population to

inflict by driving. If we refuse to let the numbers count when the choice is posed

in frequentist terms, we end up with the same two unacceptable alternatives. If

we regard statistical certainty as morally equivalent to absolute certainty (what-

ever we take the latter to mean), then driving is never permissible, because over

the course of a year, however prudently we drive, collectively we are certain to

kill or injure tens of thousands of people. On the other hand, if statistical

certainty isn’t certain enough—if we have to know with absolute certainty

that a particular instance of driving is going to result in grave injury or death

before we can rule it out—then we cannot say anything about the permissibility

of driving at all, or under specified conditions, until the harm has eventuated.

Consider the Ford Pinto case again. As I stated at the outset, the moral many

have drawn from the Pinto case is that it is unethical to allow some people to die

or suffer serious injury because it would cost “too much” to prevent it. But what

exactly does that mean Ford should do? However safe a car design, for some

15 Scanlon (1998, p. 209) acknowledges the absurdity of this result. But he believes it can be avoided

without having to discount harms by their probability of occurring (aggregation), if we allow

probabilities to be factored into the definition of the reasonable precautions that “the agent has

to take to avoid causing harm.” As I discuss elsewhere, this maneuver simply relocates aggregation; it

doesn’t diminish its role (Fried 2012a, 253–257).


Dow



amount of money it could be made safer. And however much money Ford

spends on making it safer, it can never eliminate the possibility that a design

feature will be the but-for cause of someone’s death. Does that mean that Ford

may not manufacture and sell cars? If not, what does it mean? Suppose that by

investing $5 billion in additional safety features, Ford could reduce expected

deaths or serious injuries by 0.01 percent. Should Ford (must Ford) make that

investment? If so, how about $50 billion? $500 billion? In a world of finite

resources, we have to draw the line somewhere. I don’t think any nonconse-

quentialist would disagree with that, and I imagine all would draw the line

considerably short of $500 billion, given that such a cost—which in the end

consumers will bear—would price most consumers out of the car market and

hijack scarce societal resources that could save thousands if not millions of

additional lives if deployed in a more cost-effective manner. If the nonconse-

quentialist response is, “then save the millions of others too,” we will in short

order bankrupt the world.16 But wherever we draw the line, we will knowingly

be choosing to allow Ford to increase the number of preventable deaths

“merely” to save money (to be used for some other, generally offstage, pur-

poses). The rule that determines where to draw the line therefore cannot come

from the paradigmatic nonconsequentialist thought that “human beings have

dignity and not mere price,” and that they have an “unconditional[] . . . worth

that . . . is not subject to trade-offs.” (Hill 1992, pp. 292, 296).

The most telling evidence that nonaggregative principles cannot deal with

garden-variety risk has come from nonconsequentialists themselves. Some have

been explicit about that incapacity. Consider this, from Robert Nozick (1974, p.

75): “It is difficult to imagine a principled way in which the natural-rights

tradition can draw the line to fix which probabilities imposed unacceptably

great risks upon others. This means that it is difficult to see how, in these cases,

the natural-rights tradition draws the boundaries it focuses upon.” Or this,

from John Rawls (1971, pp. 114, 339–340). In discussing how to reconcile

natural duties (which include “the duty not to harm or injure another”)

when they “come into conflict, either with each other or with obligations,”

Rawls reluctantly concedes that with utilitarianism off the table, “I do not

know how this problem is to be settled, or even whether a systematic solution

formulating useful and practicable rules is possible.” Happily, he concludes, he

needn’t answer the question in order to derive the general principles governing

16 For an impassioned example of such a response, see Wood (2011, pp. 79, 80). It is hard to resist the

sentiment that propels it—that all persons have an unconditional worth that is not subject to

tradeoffs—but it is also impossible to deliver on it.


Dow



the basic structure of society.17 Rawls’s dismissal of the utilitarian solution is

brief but telling. He considers and rejects use of the “utilitarian principle to set

things straight,” not because it isn’t a viable solution, but because it would very

quickly lead to utilitarianism full stop: “Requirements for individuals so often

oppose each other that this would come to much the same thing as adopting the

standard of utility for individuals”—a solution that he has already ruled out on

the grounds that it leads to “an incoherent conception of right.” (Rawls 1971, p.

339). Scanlon (2013, p. 510) sounds a similar note in explaining his choice to

adopt an ex post view of the problem of harm to others in What We Owe to Each

Other: “I did that because I was concerned that the introduction of ex ante

benefits and burdens would open the door to aggregation of a kind that I aimed

to avoid.”

Many more have folded implicitly, by qualifying our prima facie uncondi-

tional duties (not to sacrifice the one for the good of the many, not to cross

others’ boundaries) in terms that sound in rights-based (deontological) values

but get cashed out as aggregation. This is surely a case in which covert tools are

not good tools. If concepts like “undue risk” or “reasonable precautions” have

an operational meaning distinct from aggregation, it needs to be put on the

table so it can be assessed side by side with aggregation. If they don’t—if (as I

believe) they boil down aggregation manque—then we can cross that purported

disagreement between nonconsequentialists and consequentialists off the list

and move on.

But as Rawls foresaw with respect to harm more generally, given that virtually

everything we do poses at least some remote risk of seriously compromising the

fundamental interests of others, ceding risk of harm to aggregation comes at a

very high price for the nonaggregationist project. Every time we jog down a city

street, or decide to release a prisoner early, or decide not to breach patient

confidentiality to warn someone else that he or she may be at serious risk,

however prudent our decision, we are taking others’ lives in our hands. As a

result, ceding risk to aggregation saves nonaggregative principles by rendering

them largely irrelevant.

Not surprisingly, then, those nonconsequentialists who have taken risk ser-

iously have sought ways to keep it in the nonaggregationist fold. I have dis-

cussed two at length elsewhere (Fried 2012a; 2012b): (i) smuggle aggregation in

by the back door through qualifications placed on our absolute duty not to

harm others (e.g., except when we act with “due care”); and (ii) choose among

different persons’ fundamental interests based on factors that sound in

17 For other representative statements ceding the problem of risk to aggregative techniques, see

Coleman (2002, p. 210); Feinberg (1987, pp. 190–193).


Dow



deontological values (e.g., statistical versus identified persons, proximity, doing

versus allowing) but not the one that ostensibly underwrites our duty not to

harm: that every person has a worth “that cannot be . . . subject to trade-offs.”18

A third approach has been to seek a meta-principle that provides nonaggre-

gative justifications for embracing aggregation—e.g., the retreat to reasons in

the neo-Kantian literature; reliance on what Nozick (1974, p. 14) describes

disparagingly as “a utilitarianism of rights” before proceeding to deploy it

himself to derive the just minimal state.19

All three of these strategies to rescue risk from aggregation have been dis-

cussed in the literature, some extensively so. For the balance of the discussion, I

want to focus on a fourth that has played a much larger but generally unnoticed

role in the nonaggregationist literature in both law and moral philosophy:

recharacterize ex ante uncertainty as certainty through what amounts to hind-

sight bias.

It would be an exaggeration to say that the case against aggregation is built on

hindsight bias. But it would not be that much of an exaggeration. Strip away

hindsight bias, and the conclusion that nonaggregative principles cannot supply

a viable, systematic alternative to aggregation is hard to avoid.

4 . R I S K A S S E E N T H R O U G H H I N D S I G H T B I A S

The term ‘hindsight bias’ refers to our psychological propensity to believe,

without adequate evidentiary support, that the actual (ex post) consequences

of our choices could have been predicted ex ante.20 Monday morning quarter-

backing is one colloquial description of the phenomenon.

The standard tragic choices problems in the nonconsequentialist literature—

trolley problems and their many cognates—obviate the need for explicit hind-

sight bias by building perfect 20/20 hindsight into the definition of the problem.

Because we are endowed ex ante with perfect knowledge of the ex post conse-

quences of our actions, ex ante and ex post epistemic perspectives converge. Ex

ante, we know that act X will produce Y consequences. Ex post, when it in fact

produces Y consequences, we have learned nothing new. As suggested above,

that convergence allows nonconsequentialists to indulge the illusion that the

costs (for example) of strictly applying a rule that greater harms always trump

18 For a trenchant critique of Scanlon’s attempt to distinguish between certain and uncertain harms on

this basis, see Norcross (2002).

19 For an excellent summary of many of the arguments, see Alexander and Moore (2007).

20 More precisely, it refers to the propensity to conclude, ex post, that actual consequences had a

greater probability of resulting than the evidence that was available ex ante would have supported.


Dow



lesser harms irrespective of the numbers involved, or of treating rights as

trumps over aggregate wellbeing, will be limited to those few occasions in

which serious harm will result.

But in a world of epistemic uncertainty, the two perspectives come apart by

definition. Yes, the ex post consequences of our choices can yield important

information that allows us to update the prudence of similar choices going

forward. In some number of cases, the consequences can shed light on the ex

ante prudence of the particular decision that produced them. Our responses to

bad outcomes are, alas, not that discriminating. They generally take the much

cruder and logically indefensible form of concluding that if something bad has

happened and some human agent is a but-for cause of it, that agent should have

acted differently and hence is to blame (in the moral, not causal, sense). That is

to say, we should have acted so as to avoid harm. Our impulse to reevaluate

conduct in light of its consequences is hardly surprising. We live in the world

that is, not in alternative worlds that might have been much more likely ex ante

to come to pass. When one is a but-for cause of some tragic consequence, it is

almost impossible not to run the film backwards and think, if only I had gone to

the store an hour later, none of this would have happened. That retrospective

recharacterization of the tragic choice we actually faced—how to trade off the

convenience of driving against the remote possibility that on any occasion we

drive, including this one, we will harm uncountable and unidentifiable others—

offers the illusion that the choice needn’t have been tragic at all. After all, would

it really have been a great inconvenience to go to the store an hour later? That is

to say, it invites us to reconstruct every choice as a trolley problem, in which we

can choose with perfect knowledge of the consequences of our available choices.

That illusion is based on simple hindsight bias. As suggested above, faced

with the choice we actually have to make on any given day—whether to drive at

all, and if so, with what level of precaution—all we know is that there is some

possibility that this will be the day when, driving prudently, we will nonetheless

kill someone. Of course that remote possibility gets factored into our rational ex

ante choice. But it is only one factor, and in weighing it against others, we

rightly discount it by its probability of occurrence—that is to say, we let the

numbers count. It manifestly does not prove dispositive for the tens of millions

of Americans who drive on any given day, or for the governmental unit charged

with deciding under what conditions they are permitted to do so.

I turn now to a closer examination of hindsight bias at work in the non-

consequentialist literature on harm to others, embedded in the moral impera-

tive, First, do no harm.

One can find numerous statements in the nonconsequentialist and conse-

quentialist literatures to the effect that the consequences of our actions deter-

mine whether we acted wrongly. Many of those statements appear to be


Dow



throwaway comments or misstatements, or are using right and wrong in an

unconventional sense, and I am not concerned with them here.21 I am con-

cerned only with cases in which hindsight bias actually drives the author’s

determination whether an act was, as of the moment of choice, permissible

or not.

The problem of retrospective reevaluation of an action in light of its conse-

quences has received significant attention from philosophers in one context in

the nonconsequentialist literature: the so-called paradox of moral luck. In the

classic moral luck case posed by Bernard Williams (1981, pp. 20–39), a bus

driver sets out one day, following his accustomed route and driving with due

care (meaning, with the level of care society expects of him). A young child darts

out in front of the bus. The driver, who does not see her and could not have seen

her until it was too late to stop, hits and kills the child.22 If asked ex ante (that is,

before the child was hit) whether the bus driver was driving appropriately, the

average person would presumably say yes, because he was driving as we would

have him drive. But if asked the same question ex post, after things turn out

badly, the average person (or so it is assumed) would conclude that the driver

had acted wrongly after all.

What do we make of these apparently contradictory evaluations of the iden-

tical act? In his response to Williams, Thomas Nagel (1979) argues that they do

not create a logical paradox, because they are responding to different evaluative

criteria. The ex ante judgment reflects a rational calculus of how we would have

people act, based on the probabilistic information available to them at the time

they act. It is, in short, an action-guiding norm. The ex post “judgment”

expresses the sorts of negative emotional reactions (regret, guilt, remorse,

anger, blame, etc.) we all have when things turn out badly, through no one’s

fault. Our natural propensity to feel guilt or anger in response to the bad

21 A number of eminent consequentialists/utilitarians, including G.E. Moore, J.C.C. Smart, and Peter

Railton, at times appear to commit themselves to the identical form of hindsight bias, equating

“right action” with that “action which does in fact produce the best results.” It is inconceivable to me

that they really mean what they seem to say. Utilitarianism is a practical philosophy that cares only

about optimizing consequences. As Smart (1973, p. 46) states, “the utilitarian criterion . . . is de-

signed to help a person, who could do various things if he chose to do them, to decide which of these

things he should do” in pursuit of that end. That is to say, it is a guide to action. The best

rule-of-thumb for optimizing actual consequences is to optimize the expected consequences of

one’s action, given available information about the probability and (dis)utility of possible outcomes

at the moment one must choose how to act. In contrast, the injunction to choose those actions that

will produce the best consequences is, in a world of epistemic uncertainty, no decision rule at all.

22 In some versions, the hypothetical is altered to make the driver a little bit careless (e.g., he takes his

eyes off the road for a split second) but with bad consequences that are massively out of proportion

with that moment of carelessness. That change does not remove the central problem of outcome

luck; it simply changes the degree of luck involved. For an astute discussion of these issues, see

Jeremy Waldron (1995, pp. 387–408).


Dow



consequences of (all things considered) prudent choices may be unfortunate, in

the sense that it guarantees much unhappiness in life to no good purpose. But it

does not pose a logical paradox. It simply reflects the fact that our rational and

emotional lives operate on different planes and respond to different stimuli.

Similarly, in the nonconsequentialist literature under discussion, the appar-

ently time-inconsistent answers to the identical question—did the actor act

wrongly?—are often answering different questions: did person A act appropri-

ately, and does person A have an obligation to compensate anyone injured by

the act, whether or not he acted appropriately (that is, should we impose strict

liability for the harm). There are a number of efficiency-based justifications for

a strict liability regime, but it is not easy to explain why, if someone acted

exactly as we would have had him act, corrective justice requires that he com-

pensate the victim for the resulting harm (Fried 2012a). Why A and not some

equally innocent bystander, or society at large? However difficult to defend, the

moral requirement that A do so does not create a moral paradox. It simply

adopts different criteria for evaluating conduct and assigning the obligation to

compensate those injured as a result of it.

But clearly many nonconsequentialists mean what they say when they say

that the consequences of our actions determine whether we should have acted

as we did. To quote Richard Epstein (1973, p. 204), “[T]he first task of the law

of torts is to define the boundaries of individual liberty . . . . [T]he liberty of one

person ends when he causes harm to another. [emphasis added]. Until that point

he is free to act as he chooses, and need not take into account the welfare of

others.” Taken literally, the moral imperative “do not harm others” economically

marries the ex ante perspective (do not act in a manner that may possibly/is

likely to/is certain to result in harm to others or violate their rights) and the ex

post perspective (if what you did actually caused harm/violated someone’s

rights, you were wrong to have done it). Given that time moves in only one

direction, the resulting logical paradox leaves us to act from the ex ante per-

spective, and live with whatever opprobrium or legal sanctions come our way ex

post, via hindsight bias, for having done exactly what we were told it was

permissible to do.

Among non-libertarians, the imperative to “act so as not to harm others” has

led to some very odd conclusions, including the second of the two unacceptable

alternatives above: we cannot say anything about the permissibility of the act

until it actually results in harm to others. In John Goldberg and Benjamin

Zipursky’s formulation, “The wronging [meaning harming] of one person by

another is the very essence of . . . [acting wrongfully], and until such an event

happens, there is no occasion to inquire whether an actor can or should be held

to have acted wrongfully by violating a moral or legal obligation of conduct.”

(2007, pp. 1123, 1138).


Dow



Among libertarians, the hindsight bias built into the duty not to harm others

operates differently, because libertarians generally don’t engage the permissi-

bility of conduct from an ex ante epistemic perspective at all. Typically, they

enter the (hypothetical) scene ex post—that is, only after they conclude, on the

basis of the fact that someone’s ‘boundary’ was crossed, that a rights violation

has occurred. Most never address the question of how an agent, operating in a

world in which every action carries some risk of harm to others, should go

about deciding ex ante how to act. As noted above, Nozick’s response in a

moment of candor was that deontological principles could not answer that

question. He is right. But if so, how does it come to be that rights theorists

can offer a confident judgment on the same question ex post?—i.e., what you

did was wrong after all, on deontological grounds, because in the event it ended

up violating another’s absolute right to their person or property. The explan-

ation, once again, is simple hindsight bias.

5 . W H Y S H O U L D W E C A R E ?

The discussion up until now has focused on the logical problems facing non-

consequentialist principles in a world of epistemic uncertainty. But nonconse-

quentialists’ refusal to face up to risk and the tradeoffs it inevitably entails

carries real-world costs that ought to worry all of us. In our private lives,

most low-probability risks we run will never ripen into harm, because of the

one-off nature of our decisions. Not so in the public policy realm, where the

rules we choose often govern millions or hundreds of millions of events over the

long term. As a result, even very low-probability harms are overwhelmingly

likely to come to pass at some point. If we parole enough prisoners, one of them

will turn out to be Willie Horton. If we develop a vaccine for AIDS that has a

one in a million chance of triggering a fatal reaction and we administer it to

enough people, someone is going to die from it. We can often reduce those risks

by greater precautions. But at a certain point such precautions become pro-

hibitively costly—either in dollars spent relative to the reduction in risk

achieved or in new risks the precautions themselves create.23 And generally

speaking, no amount of precaution will eliminate all risk. If the world’s most

23 One tragic example arose in the wake of the Haitian earthquake in 2010. Responding to one sketchy

relief effort by a group of American Baptists, the authorities halted all evacuations of sick and injured

children as relief workers scrambled to obtain documentation that would prove they were not taking

the children out of the country illegally. The New York Times reported that in the first week alone,

an estimated 10 Haitian children died or became seriously ill as a result of not being able to be

airlifted out of Haiti. (Urbina, 2010). For another widely discussed example of the seeming per-

versity of many risk reduction efforts, see Wolff (2006).


Dow



conscientious child welfare agency has to supervise thousands of children from

dysfunctional families, over the long run some child under its watch is going to

die from abuse or neglect.

The question is, how should we understand and respond to those bad out-

comes when they inevitably come to pass? We know how the average citizen

responds: You (government officials, Ford Motor Company) had a duty to keep

us safe, and you failed to fulfil that duty. (As one commentator dryly remarked,

“Seldom do we hear a company that was responsible for a deadly accident

justify the loss of lives by saying that it was the result of a decision which, in

terms of its effects, produced far more good than harm.” (Hansson 2007, p. 32))

And politicians and policymakers know how the average citizen will respond.

The result is to drive politicians and government employees to channel enor-

mous resources into preventing the high-visibility bad outcomes for which they

know they will be held responsible.

Several years ago, negotiations to ease California’s budget crisis and prison

overcrowding by early release of nonviolent offenders fell apart over this prob-

lem. As one Republican legislator acknowledged with refreshing candor: “If we

let someone out early, and that man commits a crime, the Assembly members

are worried that that will come back to haunt them like the old famous Willie

Horton ads.” (Pogash & Moore 2009; see also Pfaff 2017). Similar concerns

explain why child welfare agencies responded to the rash of high visibility

deaths of children in foster care in the 1990s by redirecting almost all of their

resources to children they thought might be at some risk of death, ignoring the

thousands of others that were facing serious but (they believed) nonlethal

threats of abuse and neglect.

The same desire to avoid ex post blame in the public and private sectors

drives a substantial portion of our health care expenditures, investments in

homeland security, and countless other major public policy decisions. Our

propensity to think that if our choices have bad consequences, the choices

themselves must have been bad, may be immune to reason. As a result, the

only way policymakers may be able to cabin that thought is by manipulating the

psychological salience of bad outcomes or obscuring causal responsibility for

them. But surely, before we reach that question, we should answer the question

whether it ought to be cabined—whether, that is, outcome-determinative non-

consequentialists are wrong to think that if only we act prudently enough, we

could somehow avoid inflicting seriously bad outcomes on some individuals as

the price of realizing lesser benefits for the many.

The full-blown consequentialist will say, of course they’re wrong. It is tragic

when reasonable actions have bad consequences, but it is nobody’s fault. In a

socially interdependent world, whatever we do has potential costs to someone.

The best we can do is to act in a way that we expect will minimize aggregate


Dow



costs relative to aggregate benefits, however we calculate or weight them, and, if

the costs to individual victims are serious enough, remediate them ex post on

welfarist grounds.

If nonconsequentialists believe there is a better alternative, it needs to be put

on the table so that it can become part of the public debate over risk regulation,

and its merits assessed relative to cost/benefit analysis and other versions of

aggregation. If they believe that no amount of money is too much to require car

companies to spend to lower the risk of fatal accidents, or that it would be

immoral to permit driving for the mere convenience of others if some number

of innocent pedestrians will die as a result, then they should say so, and be

prepared to live with the economic consequences of those positions. If, on the

other hand, they understand that in a world of uncertain consequences, any rule

of conduct we adopt imposes tradeoffs among competing and often fundamen-

tal interests, then they have a moral obligation to acknowledge that fact, explain

how they would make those tradeoffs, and how, if at all, their solutions differ

from conventional aggregation. We may not be able to change the widespread

intuition that bad consequences imply bad conduct; it may be an irreducible

part of what it means to be human, and may need to be accommodated in some

fashion in public policy simply in virtue of that fact. But if the intuition is

wrong, philosophers, whether focusing on law or morality, ought to be the ones

saying so most clearly, and doing what they can to counteract the pernicious

public-policy consequences of all of us at least half-believing otherwise.

R E F E R E N C E S

Alexander, Larry & Michael Moore. 2007. “Deontological Ethics,” Stanford

Encyclopedia of Philosophy (November 21, 2007), http://plato.stanford.

edu/entries/ethics-deontological.

Coleman, Jules. 2002. Risks and Wrongs. Oxford: Oxford University Press.

Epstein, Richard. 1973. A Theory of Strict Liability. 2 J. Legal Stud. 151–204.

Feinberg, Joel. 1987. Harm to Others. Oxford: Oxford University Press.

Fried, Barbara H. 2012a. The Limits of a Nonconsequentialist Approach to

Torts. 18 Legal Theory 231–262.

———. 2012b. What Does Matter? The Case for Killing the Trolley Problem

(or Letting It Die). 64 Philos. Quar. 505–529.

Fried, Barbara. 2013. Beyond Blame. Boston Review.

Fried Barbara H. 2016. Can Contractualism Be Saved? Stanford Public Working

Paper No. 2845451.

Fried, Barbara H. Forthcoming 2020. Facing up to Scarcity: The Logic and Limits

of Nonconsequentialism. Oxford University Press.


Dow



http://plato.stanford.edu/entries/ethics-deontological

http://plato.stanford.edu/entries/ethics-deontological

Frick, Johann. 2015. Contractualism and Social Risk. 43 Philos. and Pub. Affairs

175–223.

Goldberg, John C. P. & Benjamin C. Zipursky. 2007. Tort Law and Moral Luck.

92 Cornell L. Rev. 1123–1176.

Hansson, Sven Ove. 2007. Risk and Ethics. In Tim Lewens, ed., Risk:

Philosophical Perspectives, 21–35. London; New York: Routledge.

Hill, Thomas E. Jr. 1992. A Kantian Perspective on Moral Rules. 6 Philos. Pers. 6

Ethics 285–304.

Kelman, Mark. 2005. Hedonic Psychology and the Ambiguities of ‘Welfare’. 33

Philos. Pub. Affairs 391–412.

Luban, David. 2009. Unthinking the Ticking Bomb. In Charles Beitz & Robert

Goodin, eds., Global Basic Rights, 181–206. Oxford: Oxford University

Press.

———. 2014. Risk Taking and Force Protection. In Yitzhak Benbaji & Naomi

Sussman, eds., Reading Walzer, 277–301. New York: Routledge.

Nagel, Thomas. 1979. Moral Luck. In Daniel Statman, ed., Moral Luck. Albany,

NY: SUNY Press.

Norcross, Alastair. 2002. Contractualism and Aggregation. 28 Social Theory and

Practice 303–314.

Nozick, Robert. 1974. Anarchy, State, and Utopia. New York, NY: Basic Books.

Parfit, Derek. 2011. On What Matters, vol. 1. Oxford: Oxford University Press.

Pfaff, John. 2017. The Never-ending ‘Willie Horton effect’ is Keeping Prisons

Too Full for America’s Good. LA Times (May 14, 2017) available at https://

www.latimes.com/opinion/op-ed/la-oe-pfaff-why-prison-reform-isnt-working-

20170514-story.html.

Pogash, Carol & Solomon Moore. 2009. California Officials Fear Abduction

Case May Hurt Efforts on Parole. New York Times (August 30, 2009) avail-

able at http://www.nytimes.com/2009/08/31/us/31abduct.html.

Rawls, John. 1971. A Theory of Justice. Cambridge, MA: Belknap Press.

Scanlon, T.M. 1998. What We Owe to Each Other. Cambridge, MA: Harvard

University Press.

———. 2013. Reply to Zofia Stemplowska. 10 Journal of Moral Philosophy 508–

514.

Schwartz, Gary T. 1991. The Myth of the Ford Pinto Case. 43 Rutgers Law

Review 1013–1068.

Smart, J. J. C. 1973. An Outline of a System of Utilitarian Ethics. In J.J.C Smart

and Bernard Williams, eds., Utilitarianism, for and Against. Cambridge, UK:

Cambridge University Press.

Smith, Samuel Stanhope. 1812. Princeton Lectures on Moral and Political

Philosophy, vol. II. Trenton, NJ.


Dow



https://www.latimes.com/opinion/op-ed/la-oe-pfaff-why-prison-reform-isnt-working-20170514-story.html



http://www.nytimes.com/2009/08/31/us/31abduct.html

Tetlock, Philip E., Orie V. Kristel, S. Beth Elson, Melania C. Green, & Jennifer S.

Lerner. 2000. The Psychology of the Unthinkable: Taboo Trade-Offs,

Forbidden Base Rates, and Heretical Counterfactuals. 78 J. Per and Social

Psychol. 853–870.

Urbina, Ian. 2010. Paperwork Hinders Airlifts of Ill Haitian Children. New York

Times (February 8, 2010) available at https://www.nytimes.com/2010/02/09/

world/americas/09airlift.html?scp¼1&sq¼haiti%20children%20airlift.

Waldron, Jeremy. 1995. Moments of Carelessness and Massive Loss. In David

G. Owen, ed., Philosophical Foundations of Tort Law, 387–408. Oxford:

Clarendon Press.

Williams, Bernard. 1981. Moral Luck. Cambridge, UK: Cambridge University

Press.

Wolff, Jonathan. 2006. Risk, Fear, Blame, Shame and the Regulation of Public

Safety. 22 Eco. Philos. 409–427.

Wood, Allen. 2011. Humanity as an End in Itself. In Derek Parfit, ed., On What

Matters, vol. 2, 58–83. Oxford, UK: Oxford University Press.

Wood, Gordon S. 1982. Conspiracy and the Paranoid Style: Causality and

Deceit in the Eighteenth Century. 39 William and Mary Quar. 401–441.


Dow



https://www.nytimes.com/2010/02/09/world/americas/09airlift.html?scp=1&sq=haiti%20children%20airlift




FACING UP TO RISK - Oxford Academic

Documents

Transcript of FACING UP TO RISK - Oxford Academic