Nissenbaum’s thesis is that computerization is undermining accountability. Accountability provides value to society by motivating those most likely and able to prevent risk and harm to society to do so. Accountability, of course, implies potential for liability to punishments, such as monetary damages.
The paper discusses four reasons why computer systems are diminishing accountability:
“Many hands”--responsibility is diluted because many people in many organizations work to produce the final system. Collective responsibility issues of this kind are neither unique to computer systems nor easily solved.
Bugs--because software errors are viewed as inevitable, the attitude has developed that it is unreasonable to hold developers accountable for any imperfections.
The computer as scapegoat--the tendency to blame the computer rather than a human.
Ownership without liability--many software producers want ownership rights (such as copyright and patents) to software, but no responsibility or liability for it. Others want software to be free, or free of property rights. Few people in these discussions talk about responsibilities of ownership.
Nissenbaum concludes with three recommendations for promoting accountability. First, separate accountability from liability. This distinction would help because it is the fear of monetary damages that makes avoiding accountability so desirable. Nissenbaum argues that for now, accountability is best addressed on a case-by-case basis. Second, clarify and promote a standard of care. Developers and organizations not following such standard practices would be negligent and held liable for certain kinds of errors. Third, impose strict liability on some software. Under strict liability, you are liable if your product harms me, even if you followed all reasonable standards of care. Other products that are consumer-oriented or that have a large impact on society or individuals are held to strict liability. Why shouldn’t software be held to the same standard?
These recommendations would be difficult and controversial to implement--who defines the standards of care, certifies that they were followed, and so on? Moreover, the recommendations only partially address the problem. Strict liability partly addresses the problem of ownership without liability, but perhaps by going too far in the opposite direction, and seems in conflict with using standards of care to address accountability for errors. But none of the recommendations do much to resolve the issues of collective responsibility or the computer as scapegoat.
Nonetheless, the discussion is thoughtful and thought-provoking. The proposals should serve as a useful basis for further discussion and debate. Nissenbaum is to be commended for addressing a timely issue. This paper is worth reading by all.
