do correctly. Eventually people got the technology right and stopped predicting that semiconductors would displace disk drives. We now know that the progress of disk drives has been at least as rapid as that of semiconductors and in fact they deserve to be given far more credit for their contribution to the present data storage revolution.

The Greater Difficulty of Predicting Software Innovations

Even more difficult to predict, Dr. Gomory suggested, is the progress of software. He recalled a recent conversation about how the inability to produce software might be a stumbling block to progress. He had heard those sentiments as long ago as the 1960s, when IBM launched the little remembered Future Systems project, or FS. IBM labored for three years because it had come to the same conclusion: that hardware costs would drop and the only route to continued profits would be to generate software more cheaply. The result was FS, a system aimed at making programming easier and more profitable. Future Systems never saw the light of day and cost IBM much of its leadership in technology during the three years it struggled to bring it out.

Discontinuous Change

What happened to software productivity was that companies discovered “huge levers of productivity in switching from custom to packaged software.” The lesson, he said, is that there are ways around many seeming roadblocks in technology, but they are hard to predict. While arguing that “the record of prediction is awful on the things that really matter,” there are nonetheless a few notable successes. In this context it is especially important to be modest in predicting the social consequences of technological transformations. He doubted that one could predict when the cumulative effect of technological change would suddenly manifest itself. He used the analogy of a bathtub that fills gradually until the accumulated water produces a point of overflow. This point represents a discontinuous change that does not depend on opening the faucet wider. He compared certain elements of technology that can be predicted, such as the software productivity of a programmer, and certain elements of technology that are harder to predict, such as the effects of new standards. He cited the Internet as the most stunning example of this effect, where agreement on a common protocol enabled the system as a whole to function effectively and to have an enormous impact on society. Such a confluence is almost impossible to predict. Even when we think we understand a phenomenon we might not be able to apply our understanding to the next phenomenon.

Dr. Gomory closed by reminding his audience of the complexity of the task they were attempting. He urged caution and suggested an effort to segregate what



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement