Privicating privacy: reflections on Henry Greely's commentary.

• Author: Zittrain, Jonathan
• Position: Response to article by Henry Greely in this issue, p. 1595

My article in this issue of the Review(1) makes the claim that an appreciation of the power of "trusted" or "privication" architectures could help break political and conceptual logjams suffered within long-running debates over privacy. Trusted architectures include systems of hardware and software that take note of various entitlements to the data they store--and automatically enforce those entitlements. I use "privication" to describe a structure of self-enforcing entitlements that provides access to data to a large audience as traditional publication does, while narrowing the scope of access in a way that precludes easy wholesale copying and retransmission of the data. The information thus retains private (and privately-controlled) qualities.

Prof. Greely's commentary evinces a healthy skepticism about the applicability of these architectures to the problems of medical data privacy. He makes several points; I reflect on each below.

Greely's first worry is that even a small portion of a record passed by word of mouth can cause harm or embarrassment. This is surely true, and while trusted systems can help by offering audit trails that may clearly define who would have been in a position to breach confidentiality, my concern is less with the occasional damage that gossip can wreak on privacy than with the wholesale abuse made possible by the internetworking of medical data. A comprehensive interoperable medical database will entail a sea change in the ability to search for damaging information on an individual, and to put millions of records to uses never before possible, from telemarketing to employment screening. Trusted architectures can be central components to a firewall that, together with traditional legal protections, can best advance the benefits of the information age while minimizing its intrusions.

Second, Greely finds much of the work on trusted systems to be conjectural. The problem, however, is that the technological forces at work here may have only two phases: too early to tell their impact, and too late to do anything about it. It is technically trivial to integrate a notion of patients' rights into the coming medical database while it is still on the government's drawing board, but it will be quite difficult to change the standards once they are deployed. I do not urge that we forsake a legal framework for a technical one; rather, I want our values to be reflected in and reinforced by each. In an area as sensitive as...