Work in Progress

Feel free to send me a note if you're interested in any of these works in progress.


Against Risk-Aversion

To a first approximation, risk aversion can be understood as a tendency to favor guarantees over gambles, even when the gambles would maximize expected utility. Although orthodox decision theory forbids this sort of risk aversion, it is widely thought that some degree of risk aversion is rationally permissible. To respect this thought, some have proposed alternative decision theories, which do permit risk-averse behavior. This paper explores a difficulty for such alternatives that begins from a consequence of John Harsanyi’s (1955) aggregation theorem: Risk-aversion-friendly views permit risk-averse altruists to violate ex ante pareto, thereby choosing options that are strictly worse in expectation for everyone involved. By itself, this may not seem so bad. But variations on a simple example illustrating Harsanyi’s result lead to more trouble. In particular, these views will permit risk-averse altruists to give someone a ⅙ chance of survival, when that same someone could be given a ⅓ chance instead—without affecting anyone else. This doesn’t look like the rational behavior of an altruistic agent.

Diffuse Harm and Fortuna's Wheel

A diffuse harm hurts many people a little; a concentrated harm hurts one person a lot. Other things equal, a diffuse harm seems less bad than its concentrated counterpart. But this widely endorsed thought is surprisingly susceptible to criticism. Some forms of it can be shown to be in tension with attractive principles, such as anonymity and the transitivity of 'better than'. These demonstrations appeal to cases that involve a sequence of social positions, each very similar to the last, such as the Fortuna’s Wheel scenario. In response, some may wonder whether the parity of diffuse and concentrated harms holds only in these sequential cases. But it can be argued that the parity of diffuse and concentrated harms extends beyond such cases. Specifically, it can argued that in many realistic cases, the expected damage of a diffuse harm is approximately equal to the expected damage of a relevantly similar concentrated harm. Diffuse harm is easily underestimated.

There Is No Inefficacy Problem

The inefficacy problem is said to arise in cases where (i) a morally significant outcome depends, causally, on the behavior of a group and (ii) the outcome does not depend, causally, on the behavior of any of its members. Although some have expressed skepticism about the possibility of such cases, the widely accepted view within moral philosophy seems to be that such cases are indeed possible and that they give rise to vexing philosophical questions—many explored in this volume. This chapter argues against this prevailing view. First, it is argued that the so-called inefficacy problem does not arise in a fully precise world. It is then argued that the inefficacy problem cannot arise entirely out of vagueness.

Statistical Evidence, Accuracy, and Justified Conviction

It seems wrong to convict someone on the basis of purely statistical evidence. But how does this widely held judgment fit with existing accounts of criminal punishment? Retributive and utilitarian accounts, which are both popular, are unable to explain and vindicate our unwillingness to convict on purely statistical grounds. Only the proceduralist approach has the tools to do this. Thus we either should adopt a proceduralist account of criminal punishment or revise our views of merely statistical evidence.

Voter Ignorance, Voter Altruism, and the Prospect of a Decisive Vote

It has recently been argued that voting to change the outcome can often be rational. The argument, though, assumes that the would-be voter (a) knows who the better candidate is and (b) is a perfect altruist. Both assumptions seem false in most cases. How can it be legitimate to make false assumptions? This is a fair criticism. But I'll argue that there is a role to play for both of these assumptions in assessing the rationality of a given type of behavior--whether they are true or not. Ultimately, we will see that there remains a deep and important sense in which the act of voting is rational. At the same time, pessimists may be right that the vast majority of actual voters fall short of meeting this rational standard.

Unorthodox Treatments for Baldness

Bald? Philosophy can help. First, you must grant that removing a single strand of hair from a person's head cannot make them bald. If you don't grant this, the treatment won't work. Now fetch a large number of people. Remove a single hair from each, and transfuse those follicles onto your head. You're now cured—and, by our earlier assumption, we didn't make anyone bald in the process. So what? Interestingly, if everything is configured just right beforehand, then the proposed treatment turns out to commit us to something surprising: baldness is not an intrinsic property—whether one is bald depends on more than the number/arrangement of hairs on one's head. Does this seem implausible? If so, then the observation makes trouble for recent attempts to devise frameworks friendly to the assumption made at the outset.

A Free Lunch for God? Vagueness, Gratuitous Evil, and the Divine/Mortal Asymmetry (with Philip Swenson)

According to most theodicies, the existence of suffering generates certain goods, which can explain why God would allow the suffering to take place. Many of these views face a forceful question: If these hypothesized goods justify God in allowing suffering... what's to say that ordinary mortals like us can't help ourselves to the same justification? Perhaps we are also permitted to allow suffering we're in a position to prevent. One response to this challenge appeals to vagueness: we can be certain that, no matter how we decide, we will not jeopardize the hypothesized goods, if we choose to reduce suffering. But the same principle doesn't hold for God (since God is in a position to prevent far more suffering than we are). This response, however, incurs serious costs: an example involving a series of different worlds, with increasing quantities of evil, shows that the proponent of the foregoing response must maintain that, necessarily, merely by relabeling, God could improve at least one of these worlds without harming any of them.