Thanks for all these helpful comments! Oddly enough, the study Howard suggested with Satan has actually been run already (by experimental philosopher Mark Phelan). Interestingly, even when it comes to Satan, people say that the bad side effects are intentional and the good ones are unintentional. A number of readers suggested that the effect might not be due to people's true moral judgments, and some suggested that we might address this question by looking at whether people from groups that make different moral judgments actually arrive at different intuitions about intentional action. For a study that speaks to these concerns, one might look to the work of Tannenbaum and colleagues. They found a sample that included both conservatives and liberals. Then subjects in this sample were asked to make judgments about either (a) a case in which Americans are trying to bomb Iraqi insurgents but know that they will kill innocent people who happen to be living nearby or (b) a case in which Iraqis are trying to bomb American soldiers but know that they will kill innocent people who happen to be living nearby. Surprisingly enough, the conservatives say that the Americans killed the innocents unintentionally while the Iraqis killed the innocents intentionally... but liberals say that the Americans killed the innocents intentionally and the Iraqis killed them unintentionally! Does that help to show that the effect really does involve people's moral judgments?