law. Nevertheless, it seems clear that the European process has moved more effectively than the U.S. process to define the critical issues and subject them to debate. The key is probably the WIPO/UPOV pattern of repeated expert meetings producing substantive proposals that are subjected to external criticism.
Computer programs (software) pose an extremely different set of issues. In the first instance, they appear to be (and are) text, and one is concerned about direct copy of this text. Yet, as Randall Davis of the MIT Artificial Intelligence Laboratory stated, "Programs are not only text; . . . they also behave."19 The type of protection to be provided that behavior is strongly debated, and given the extent to which computer and communications software (broadly conceived) is growing in market size and economic value compared to the corresponding hardware, the character of protection to be provided is extremely important economically.
The first special issue posed by software is its easy reproducibility. Discs can be copied cheaply and converted relatively easily from one computer language to another. This, of course, lends appeal to the copyright approach to software protection—and few would deny that copying or direct translation should in general be prohibited, because it fundamentally affects the incentives needed to develop software in the first place.
However, after this point, the rights that should be given to a software producer become unclear. Should a user be entitled to decompile the program as part of a reverse engineering process? Should software be given protection against other software that uses the same code (in places) or program outline, when one considers that the outline and parts of the code may be defined by the problem and that independent "clean-room" development may lead to the same outline and in some cases the same code? Should there be protection for the program's appearance to the user (the "look and feel")? What about protection for novel algorithms? Also, should standards and interfaces associated with programs be protectable?
These issues have arisen in an industrial structure that is, in general, much closer to that of engineering than to that of literature. Programs are