For the past several years, the New York University Economics Department has conducted a basic investigation of the economics of information. This work, sponsored by the National Science Foundation, has delineated the difference between the peculiar characteristics of information as an economic commodity and the characteristics of ordinary goods and services and has explained why a private market for information products may not function properly. A special area of study has been the transfer of information, in particular through scientific and technical journals. In this report the authors apply their basic research on the economics of information to the production of computer programs.

Conclusions

1. The discipline of economics offers a basis for making analytical statements regarding the pertinence of intellectual property rights in general, and copyrights in particular, to the production of computer software.

2. As the American economy relies increasingly on information products and electronic data processing, the importance of software will grow. Examples suggest that private production in response to incentives may not entirely meet the nation's needs and that some public subsidy may be justified. A failure to develop an adequate policy towards computer software could conceivably have an inhibiting effect on the overall growth of the economy.

3. With proper specifications, and under certain conditions, copyright can provide an effective incentive for the production of computer software. The authors prefer a system of copyright protection to the currently prevailing reliance on trade secrecy on a variety of counts. Trade secrecy, which works better for intermediate than final products, restricts the range of direct users in a way that copyright would not. Trade secrets necessarily inhibit the flow of information about computer programs, thus making it more likely that separate efforts will result in wasteful duplication, making it more difficult for buyers to search out suitable products, and possibly making it more difficult for new firms to go into the programming business. Trade secrecy may also result in the bundling of programs with other services or products as part of an overall package, to the detriment of customers or consumers. Finally, the need to maintain secrecy leads to building certain undesirable qualities into software, such as obscure codes and unnecessary complexity. Copyright is claimed to have superior characteristics in all of the above interests.

4. In general, the New York University economists support broad specification of property rights through the copyright mechanism, so as to allow the copyright owner to exploit as many markets as possible. In this vein, the practice of charging some customers a higher price than others-which sometimes involves an antitrust violation-may have merit if it permits an otherwise unprofitable enterprise to make money and hence be undertaken. The exemption of certain users of copyrighted works, whether through fair use or library or educational pro-visions, results in an implicit subsidy for those {Page 126} favored users, a subsidy whose burden is felt partly by other users who are fully subject to the provisions of the copyright law. Economists generally prefer open subsidies borne by the general public through taxes as both more efficient and more equitable.

5. This report specifies a model to estimate the best length of copyright protection to provide maximum benefits for the public but taking into account the need to provide adequate incentives to producers. The length of protection should be greater than the time needed for a producer to recover costs and make a profit, but less than a work's useful lifespan, so that some software will be in the public domain while still useful. The variables employed in the calculations included the average useful lifetime of programs, the possibility of economies of scale in production, the responsiveness of demand to changes in price, the rate of decay of what customers will pay for a given program over time, and the social interest rate. Since the values of these variables were not known definitely, the best length of protection was estimated as falling in the range between two and fourteen years.

6. Any legislature has only two basic considerations in designing a copyright law to provide incentives: the breadth or scope of protection, and its length. Increasing either one increases the opportunity for profit but also imposes a greater cost on the public. There exists a tradeoff between these two dimensions: the more there is of one, the less there needs to be of the other. The Copyright Act of 1976 stands at one extreme, with a very long period of protection but filled with multiple exemptions. However, a quite different system might work for computer software: very short but very tight protection.

Next section: Legal Protection of Computer Software, An Industrial Survey