Volume 22, Issue 1

Categories intentionally create discontinuities. By breaking the world up into cognizable chunks, they simplify the information environment. But the signals they provide may be inaccurate or scrambled by strategic behavior. This Article considers how law might approach the problem of optimal categorization, given the role of categories in managing and transmitting information. It proceeds from the observation that high categorization costs can be addressed through two opposite strategies—making classifications more fine-grained (splitting), and making classifications more encompassing (lumping). Although continuizing and other forms of splitting offer intuitive answers to inaccurate classification and gaming along category lines, lumping is sometimes a better solution. If category membership carries multiple and offsetting implications, the incentive to manipulate the classification system is dampened. To take a simple example, insurance that covers only one risk is more vulnerable to adverse selection than is an insurance arrangement that covers two inversely correlated risks. Making categories larger, more durable, and more heterogeneous can produce such offsets. These and other forms of bundling can arrest damaging instabilities in categorization .

For every legal concept X, there are clear instances exemplifying an X and clear instances exemplifying a non-X. The cases that come before courts are those that seem to lie in between, being neither clearly an X nor clearly a non-X. It is tempting to think that, being in-between, they should receive an in-between treatment, that is, to the extent that they are an X they should be treated as an X. If they are sixty percent toward being an X, they should get sixty percent of the treatment due an X. But this presupposes that in-between cases can be rank-ordered at least roughly according to the degree of their X-ness. This Article explains why that generally cannot be done and why courts therefore go for an either/or approach: something gets treated either as an X or as a non-X. The explanation is rooted in the kind of phenomena explored in the theory of social choice and multi-criterial decision-making .

According to an influential and intuitively appealing argument (the Continuity Argument), (1) morality is usually continuous, namely, a gradual change in one morally significant factor triggers a gradual change in another; (2) the law should usually track morality; (3) therefore, the law should often be continuous. This argument is illustrated by cases such as the following example: since the moral difference between a defensive action that is reasonable and one that is just short of being reasonable is small, the law should not impose a severe punishment when the action is almost reasonable and no punishment at all when the action is reasonable (as positive law sometimes does). In this Article, I consider two doubts regarding this argument. First, the premise that morality is continuous in such cases is incompatible with the common view that the moral status of actions is not continuous since there is an important difference between actions that are permissible and actions that are wrong— even if this difference is due to a difference that is very small, such as the one between an action whose consequences are the best and an action whose consequences are just slightly less good. This view extends also to the overall moral status of agents given the common assumption that it depends on the moral status of their actions. This is an important challenge that the Continuity Argument should confront. However, I argue that the best account of morality is more scalar than the common view in these respects. Therefore, I conclude that the first premise of the Continuity Argument is correct in this regard, although it is based on a minority view. The second doubt concerns the scope of the second premise: since there are reasons both in favor and against legal continuity, and the applicability and force of these reasons depend not only on various moral propositions but also on contingent non-moral facts, we often lack the evidence to determine the degree to which the law, at a certain place and time, should be continuous, and specifically that it should often be continuous .

Criminal law conceptualizes guilt and the finding of guilt as purely categorical phenomena. At the end of trial, the defendant is pronounced either “guilty” or “not guilty” of the charges made against her, excluding the possibility of judgment of degree. Judges or juries cannot calibrate findings of guilt to various degrees of epistemic certainty by pronouncing the defendant “probably guilty,” “most certainly guilty,” or “guilty by preponderance of the evidence.” Nor can decision makers qualify the verdict to reflect normative or legal ambiguities. Findings of guilt are construed as asserting factual and legal truths. The penal results of conviction assume similar “all or nothing” properties: punishment can be calibrated, but not with the established probability of guilt . The prevailing decision-making model, with its ‘on-off’ formulation of guilt, is so broadly established that it is considered an axiom— but there is nothing natural or pre-political about it, nor about the derivative distribution of punishment. This Article attempts to expose the hidden potential rooted in the construal of criminal verdicts as judgments of degree, by drawing three hypothetical manifestations of a linear conceptualization of conviction and punishment in the criminal trial and plea-bargaining arena. It also offers a normative assessment of converting criminal verdicts from categorical decisions to continuities .

The law inevitably draws lines. These lines distinguish, for example, whether certain conduct reflects ordinary recklessness constituting manslaughter or more extreme recklessness constituting murder. There is no way to meaningfully draw such lines, however, absent shared ways of representing amounts of recklessness or at least knowledge of the consequences of drawing lines in particular places. Yet legal actors frequently draw lines in the dark, establishing cutoffs along a spectrum with little or none of the information required to do so in a way that suits the law’s goals. For example, jurors must decide whether some conduct constitutes extreme recklessness without knowing prior precedent nor the sentencing consequences of drawing cutoffs in particular places. Judges and lawyers cite line drawing precedents from other jurisdictions without considering whether the lines drawn in prior cases had the same consequences as those in the case at bar. And scholars argue about how to classify conduct without making clear what consequences they believe ought to attach once the classification is made, leaving it hard to tell when scholars have substantive or simply superficial disagreements. In this Article, I discuss some line drawing problems and briefly suggest ways we can add meaning to cutoffs. More generally, I argue, we can “smooth” certain features of the law to both reduce our vulnerability to line drawing in the dark and improve the fit between the law and what our best theories of law recommend. Even when we cannot easily smooth the law, thinking about the law in a smoother fashion can help reduce the jurisprudential pathologies I describe.

The law values fairness, proportionality, and predictability. Accordingly, in the context of criminal law, punishments should be carefully calibrated to reflect the harm caused by an offense and the culpability of the offender. Yet, while this would suggest the dominance of “smooth” input/output relationships—for example, such that a minuscule increase in culpability would result in a correspondingly small increase in punishment—in fact, the law is laden with “bumpy” input/output relationships. Indeed, a minuscule change in input (be it of harm, culpability, or any number of other measures) may result in a drastic change in output, creating significant discontinuities. Leading scholars have argued that smooth input/output relationships, which feature careful gradation and calibration, better accord with dominant theories of punishment than do bumpy relationships, which lack fine-tuning. Accepting as a starting premise that smooth input/ output relationships are to be preferred in the criminal law, this Article focuses on the significant doctrinal and practical impediments to smoothing out these relationships. This analysis reveals challenges to smoothing out relationships between inputs and outputs, as well as the difficulties associated with addressing discontinuous relationships among inputs and outputs. Specifically, it exposes the law’s classification of inputs and outputs itself as contestable and responsible for a range of hard-to-resolve discontinuities. In doing so, this Article begins the task of laying the groundwork for further analysis and possible reforms.

Judges and juries often make factual decisions even if the facts are disputed and there is no clear-cut evidence available. Despite this common state of uncertainty, verdicts are thought of as having clear winners and losers––either the plaintiff wins and receives a full remedy, or the defendant wins and the plaintiff gets nothing. In private disputes, factfinders base their binary factual determinations on the preponderance of the evidence. There are, however, several doctrines that allow for partial remedy, discounted by the probability that the facts support the plaintiff’s case, given the available evidence (proportional liability). This Article offers a general theory for proportional liability in private law. It identifies three types of factual uncertainty—mutual uncertainty, unilateral uncertainty, and institutional uncertainty—and shows that legal economists should support proportional liability when the state of uncertainty is shared by the parties and the court (mutual uncertainty), and they should adopt an all-or-nothing rule whenever the information is observable but unverifiable (institutional uncertainty). In cases where one party holds private information (unilateral uncertainty), proportional liability is sometimes, but not always, superior to an all-or-nothing rule.

This Article explores settlement incentives under three different burden of proof rules. The conventional burden of proof is a discontinuous step-function, jumping from no damages to full damages at the 0.5 jury confidence level. Continuous burdens of proof, by contrast, would permit sanctions to steadily increase as juror confidence rises from 0 to 1, with no discontinuity. Linear burdens, which have received extensive attention in prior literature, escalate sanctions steadily across the whole range of confidence levels, while the logistic burden takes a nonlinear form. Using a data simulation approach guided by the empirical realities of American civil litigation, I consider the incentives that each of these rules creates for parties contemplating settlement, using a model in which parties make divergent forecasts of their expected outcomes at trial due to optimism bias. Based on this analysis, I conclude that a linear burden would likely raise our settlement rate by a modest amount, except in very large cases and in “easy” cases, in which an unbiased person would predict that a trial factfinder would have a level of confidence in liability quite close to either zero or one. I also compare the expected error rate of the settlements that each rule produces, and find that the linear rule modestly lowers the expected error rate of settlement overall, although this benefit does not hold for easy cases or those with very high damages. Lastly, I conduct a similar analysis for the logistic burden, finding that it induces a similar quality and quantity of settlements as we currently achieve using conventional burdens.

This Article explores the costs and benefits of one subset of continuous and discontinuous rules. These expressions are shown to be distinct from the familiar dichotomy expressed as standards versus rules, but they share the difficulty of dividing the world of law in two. Still, regulatory approaches that focus on discontinuities can often be made more continuous, and vice versa. A speed limit is discontinuous in the sense that one drives above or below (or within) the announced limit. But it is often made more continuous—even with more discontinuities— as when the stated limit is different for various kinds of vehicles. This Article works around these definitional problems to show that law often discourages useful disclosures by encouraging parties with information to offer continuous information in order to avoid after-the-fact lawsuits when specific disclosures prove to have inaccuracies. For example, it is common to hear or be warned that a medical procedure poses the risk of death, when a better-informed doctor or hospital could have given the precise percentages attached to various outcomes. Similarly, a corporation is on safe ground when it follows “generally accepted accounting principles,” when investors would have learned more from information about good and bad outcomes put in probabilistic terms. The Article works toward the suggestion that law might create a safe harbor in which probabilistic disclosures are protected when they are, or are certified to be, more useful than the ready alternative of fairly general disclosures.

There are powerful fairness and efficiency arguments for making charitable donations to soup kitchens 100% deductible. These arguments have no purchase for donations to fund opulent church organs, yet these too are 100% deductible under the current tax code. This stark dichotomy is only the tip of the iceberg. Looking at a wider sampling of charitable gifts reveals a charitable continuum. Based on sliding scales for efficiency, multiple theories of fairness, pluralism, institutional competence and social welfare dictate that charitable deductions should in most cases be fractions between zero and one. Moreover, the Central Limit Theorem strongly suggests that combining this welter of largely independent criteria with the wide variety of charitable gifts results in a classic bell-shaped normal curve of optimal deductions, with a peak at some central value and quickly decaying to zero at the extremes of 0% and 100%. Given that those are the only two options under the current tax code, the current charitable deduction regime inevitably makes large errors in most cases. Actually calculating a precise optimal percentage for each type of charitable donation is of course impractical. This Article suggests, however, that we can do much better than the systematically erroneous current charitable deduction. Granting a 100% deduction only for donations to the desperately poor, along with 50%, 25%, and 0% for gifts yielding progressively fewer efficiency, fairness, pluralism, and institutional competence benefits, promises to deliver a socially more desirable charitable deduction.

President Trump’s decision to move his official state of residence from high-tax New York to no (income)-tax Florida has brought public attention to an issue that has long troubled scholars, designers and administrators of income tax systems: how the interaction of tax rules deferring the taxation of income and tax rules based on residency allows taxpayers to reduce and even avoid taxation of their deferred income. These discontinuities in tax treatment may lead to excessive migration, as well as reductions in state income tax revenues and distortions in the design of state taxing mechanisms. This Article explains what states would have to do to eliminate these avoidance opportunities. However, it also points out that many of these policy changes would create other tax discontinuities. Ultimately, it leaves open the question whether making any of these changes would lead to fewer financial and behavioral distortions.