We have failed Jay Katz. Like the man looking under the lamp-post for his keys-not because that was where he was standing when he dropped them but because the light is better there-we have labored too long in the light and poked too infrequently into the shadows where the often painful truth is to be found. We have treated as exact the imprecise process of balancing research risks and benefits. We have exalted autonomy and made a sacrament of consent forms, even those that run to hellish lengths, littered with jargon - and forgotten the myriad constraints on subjects' choices. We have realized that, however well-intentioned researchers may be, their individual judgment of when and how to conduct research is usually very partial, in both senses of that word. Yet, from that realization we have moved to the contradictory conclusion that by instituting prior review by Institutional Review Boards (IRBs), we have solved the ethical problems involved in deciding when and how to conduct research.

Above all, we have developed elaborate rules and processes to normalize human experimentation, to treat it as an ordinary activity. We have thus avoided looking clearly at the moral dilemma that lies at the heart of every research encounter: "We are asking you to do this not for yourself but for others, even though we know that the role of human subject entails real and sometimes unforeseen risks including death." Such a statement is significant not becauseor not solely because-it clearly describes the potential harm. I agree with Jay that this is not the critical issue, though it is hardly one that we can ignore, in light both of the historical abuses of research subjects and also of what has occurred much more recently at such renowned medical institutions as the University of Pennsylvania and Johns Hopkins. Rather, its central significance lies in its frank description of the aims of research and, hence, of the potential divergence of interest between the prospective subject and the person offering to enroll him or her in the research project.