Software Technology In order to be studied and communicated, algorithms must be expressed in some language (although not necessarily a computer programming language). Το be useful, that language must serve one or more of three goals: 1) It needs to be precise, in that the algorithm is presented unambiguously and the reader clearly understands which details have been eliminated. 2) It may need to be in a form in which the algorithm can be analyzed 3) It may need to be readable and comprehensible for communication to a student or another expert. If precise enough, this language for expressing algorithms can resemble a higher level programming language, for computers are very good at filling in details. OFFICE OF TECHNOLOGY ASSESSMENT Staff Paper on Intellectual-Property Protection for Computer Software to the Committee on the Judiciary Subcommittee on Courts, Intellectual Property and the Administration of Justice U. S. House of Representatives November 2, 1989 This paper draws upon prior and ongoing OTA work, including two assessments requested by the Committee on the Judiciary and its Subcommittee on Courts, Intellectual Property and the Administration of Justice: Intellectual Property Rights in an Age of Electronics and Information (April 1986), Copyright and Home Copying: Technology Challenges the Law (October 1989), Information Technology R&D: Critical Trends and Issues (February 1985), and Information Technology and Research (ongoing). In this paper, we note current issues regarding legal protection for computer software and identify questions that Congress might wish to consider in its oversight of computers, software, and intellectual property. We do not reach policy conclusions. Overview Without The health and vitality of the software industry are crucial to the computer industry, to Government, and to the economy as a whole. software, computers would be unusable.1 Software is vital to defense and civilian agency operations, and to industrial sectors as diverse as telecommunications, electronics, transportation, manufacturing, and finance. In software, the United States still has a distinct advantage in world competition, but this may be in jeopardy in the future. Legal protections for computer software can affect the pace of technological advance in software (and the extent to which these advances are disseminated and used throughout the economy), as well as developments in the i computer-hardware industry. For example, software protections affect the "openness of standards and interfaces. These are important components of firms' competitive strategies in both the software and hardware industries. Thus, the economic implications of under-protecting or over-protecting software extend far beyond the software industry alone. Unfortunately, basic questions about the detailed implementation of intellectual-property protection for software--what to protect? how much? for how long? against what? from whom?--are difficult to answer. Software does not fit comfortably into the traditional intellectual-property frameworks of copyright (which protects writings) or patent (which protects processes and 1 This paper uses the term computer software (or just "software") to refer to sets of instructions--computer programs--for digital computers. Although some analysts use the term "software" to include electronic databases, this paper maintains a distinction between software and databases. In some ways software and databases are merging, however; in the future it may be hard to distinguish between a program and its data. machines). This problem is being manifested in the current round of "look and feel" copyright suits and in the controversy over patent protection for algorithms and computer programs. Another problem in determining where software fits in the intellectualproperty system is that computer software and hardware technologies are changing rapidly, both qualitatively and quantitatively. This makes the crafting of software protections akin to aiming at target that isn't there yet. For example, future advances in computers and computation, especially in artificial intelligence, will force the development of new definitions of "software" and "data". A new type of computer, called a "neural net, " is not programmed as are conventional computers; instead, it is "trained." Moreover, in the future, it may be hard to distinguish between a program and the data on which it operates. Compounding the problem of rapid technological change is a third problem: The legal and technical communities do not have consistent definitions for important terms like "algorithm" or "interface. Without agreement on a common language, discussing protection issues is extremely difficult.2 пре 2 One common definition of the term algorithm is: "a set of rules which |