Faculty of law blogs / UNIVERSITY OF OXFORD

Reasonable You: Justice and Personalization in Tort Law


Time to read

6 Minutes

This post is part of a special series including contributions to the OBLB Annual Conference 2022 on ‘Personalized Law—Law by Algorithm’, held in Oxford on 16 June 2022. This post comes from Sandy Steel, who participated on the panel on ‘Personalized Law’.

At the root of Personalized Law is an irresistible argument. It goes like this:

  1. Law ought to be sensitive to normatively relevant facts
  2. Facts about particular individuals are sometimes normatively relevant facts
  3. Therefore, law ought sometimes to be sensitive to facts about particular individuals.

The argument seems valid, and all of its premises are true. So, hurray for Personalized Law. But, also, boo for Personalized Law: if that were all the book is saying, then it might be accused of stating the obvious—everyone would agree that the law ought to be sensitive to the facts about particular individuals when those facts are normatively relevant.

The claims of Personalized Law, however, are clearly not obvious or trivial: its significant, interesting, claims are, I think, two: (a) that facts about particular individuals are far more often normatively relevant to certain areas of law than is usually thought or reflected in the law itself, (b) that these facts can be taken into account by law without, or with fewer, costs than are usually associated with an increased focus on particular circumstances.

Why should facts about particular individuals more often be taken into account as normatively relevant? The primary reason given by PL is efficacy. More personalisation leads to better achievement of the law’s justified goals. For example, if the goal is harm-prevention, then a legal directive which requires duty-bearers to modify their conduct in response to different features of individuals, when these features bear on effective means of harm-prevention, will better fulfil the goal. So a legal duty that required warnings to consumers to be based on their ability to understand or, more simply, their language, would be more likely to prevent harm, all else equal.

How would personalisation more effectively achieve the goals of tort law?

PL focuses its analysis on the tort of negligence, and in particular what it describes as ‘the duty to take reasonable care’ (ch 4). In this tort, liability hinges upon the failure to take reasonable care, and this is judged by reference to how an ordinary and reasonable person would have behaved in the circumstances. By and large, the ordinary and reasonable person is not invested with the personal characteristics of the duty-bearer.

In the world of personalized negligence law, however (p 62):

‘There is no longer a ‘reasonable person’ whose fictional attributes anchor the standard of care. Instead, there is a ‘reasonable you’—how each person should act given their individual circumstances, skill, and risks. What is reasonable for an individual defendant would be determined, not by reference to the average traits of some reference group, but instead by the skills and riskiness of this defendant. A young person with good eyesight and quick reflexes, who does not take drowsing medications, would be allowed to drive faster and over longer hours in the same road conditions as someone without these traits’.

PL distinguishes two kinds of personalization of the standard of care: (1) risk-based personalization, which focuses on features of individuals which make them riskier in their interactions with others—‘people with poor eyesight, people who are distractible or habitually tired, or those with less control over their actions’ (p 63); and (2) skills-based personalization, which takes into account variances in skill at reducing risks created, for instance, a person’s ‘technological agility’ (p 63).

Consider risk-based personalization. The idea is simple: ‘requiring high-risk people to take more care reduces the costs of accidents’ (p 64). If we understand the breach requirement in negligence as a cost-benefit standard, this readily follows: for the extra-risky individual, there is a greater potential benefit to the precaution, and so a higher level of cost that could be efficiently spent than for the lower-risk individual. To take a simple example: requiring a low-risk driver with 20-20 vision to have spectacles is inefficient while it is not for a higher-risk driver with impaired vision.

Consider now skill-based personalization. Again, the basic idea as to the efficiency of such personalization is intuitive: ‘If some people have better skills to prevent accidents, it is a waste not to harness these skills through elevated standards of care. Drivers that are more effective in using accident prevention measures should be required to use them more, and doctors who can more easily deploy effective medical devices should be required to operate them’ (p 67). In cost-benefit terms, in other words, the cost of precautions for the highly-skilled is often lower than for the less-skilled; hence it will more often be cost-justified for the precaution to be taken by a more highly-skilled individual compared to a lower-skilled one.

Not only would these kinds of personalization be cost-efficient, PL claims, but they would also be just. As a matter of interpersonal justice (PL uses the term ‘corrective justice’), risk-based personalisation is appropriate because ‘[p]eople whose conduct causes more risk are more to blame and should take more care toward their potential victims’ (p 69). Skill-based personalization is more controversial, they think, as some prominent tort theorists have contended that considerations of ‘cost’ should be irrelevant to whether a person has breached a duty of care. But PL rejects this view: ‘A corrective justice account of negligence law must define what losses are wrongful and what conduct by the injurer is unreasonable and blameworthy. This cannot be done meaningfully without considering the burden of care’ (p 70). In a nutshell, whether conduct is unreasonable or blameworthy depends in part upon the costs of avoidance.

Here are four comments.

First, it seems to me that the current law of negligence already de facto effectuates risk-personalization more than PL recognizes. If am inherently clumsy and accident-prone, de facto the kinds of precautions I must take in order to behave as an ordinary, reasonable, person are greater than those that must be taken by those who are not inherently clumsy and accident-prone. In other words, in order for such a person to bring their risk-level to the reasonable level, they will need to take significantly greater levels of precaution than the ordinary person.

Second, although PL presents tort law’s impersonal approach as primarily motivated by administrative concerns—particularly, the absence of reliable information about people’s abilities, and the costs of acquiring such information—it seems clear that, rightly or wrongly, the law sometimes adopts an impersonal approach for reasons that are independent of information costs. For example, in the tort of private nuisance, a hypersensitive use of land is not protected to the same extent as an ordinarily sensitive use. If I use my land to grow rare and delicate orchids in greenhouses, and these are withered by smoke from your barbeque, I have no claim—even if you are aware of my hypersensitive use, and so the problem is not merely one of the availability of information.

Third, consider the justice-based arguments for personalization of either kind. A strand of PL’s argument focuses on the role of blameworthiness. The normative role of blame within interpersonal justice is controversial, however. If we are in the business of punishing, then there is a case that the punished person ought to have committed a blameworthy wrong, otherwise the condemnatory feature of punishment is pro tanto unjustified. But if we are concerned with what one person owes another by way of compensation after a harmful transaction, it’s not clear why blame should be necessary. It seems sufficient that one person had a duty not to inflict the harm and failed to comply with that duty. Since not all breaches of duty are morally blameworthy, it follows that there is already a pro tanto case for compensation absent blameworthiness. If the cost of avoiding some harm or risk is only relevant as an excuse—that is, as a defeater for blameworthiness—then it’s not clear why compensatory justice requires its relevance.

A better argument, therefore, drops the focus on blame and insists only that that what we owe, as a matter of right and duty, to each other depends upon cost. This is difficult to resist. Why is it the case that a person is not duty-bound to sacrifice their arm to save another’s two arms? If this is the case, it seems impossible to explain without reference to the cost to the duty-bearer. In short, what we can reasonably demand of others depends, in part, upon the consequences to those others. This is already reflected in views which accept that duties depend upon a balance of interests.

Finally, there seems to be one strand of argument for personalization in tort which deserves further emphasis than PL gives it. We could call it the impersonal argument for personalisation. The point is that the law’s current largely impersonal approach fails, contrary to appearances, to hold people to the same standard. The current law does not look at the cost to the particular individual, but cost to an ordinary and reasonable person. But if personal cost is what matters morally, and there is a threshold of cost that each person can be expected to bear, then some people will be held under the current law to a standard which requires more of them than what morality demands. If it is possible to acquire more knowledge of the personal cost to the particular individual without undue administrative cost, then personalizing the standard would actually ensure that each person bears the same amount of cost to avoid risks to others. One powerful argument for personalisation is rooted in equal treatment.

Sandy Steel is a Professor of Law at the University of Oxford.


With the support of