Go Back.
One of the many paradoxes of the Information Age is that software -- the very essence of the automated systems that move and process information, manage factories, and maintain complex accounting and record-keeping systems -- is produced much as 19th-century handcrafted products were. Shrink-wrapped commodity software for personal computers accounts for about 15 percent of the market; the balance is made up of large, custom systems for such things as financial services, manufacturing, or chemical processing. These systems are custom-built monoliths, produced one at a time at costs of up to tens of millions of dollars. The failure rate for developing these systems -- projects that are started but never become fully operational -- is reported to be about 70 percent. And because each software application is crafted from scratch, software productivity has not improved in decades.
The ATP Component-Based Software focused program is a five-year cooperative effort with U.S. software companies to establish the technology foundation for a fundamental change in software production. The key idea is to create techniques for automated software development that are sophisticated enough to compose software based on the intent of the developer, rather than following a huge number of error-prone instructions. Making this concept of semantically based software automation into a practical reality is an extremely risky undertaking. But if it can be done, the implications for the information revolution are enormous. As the concept of interchangeable parts fueled the industrial revolution a century ago, creating economies of scale and dramatic growth in productivity, this new capability could put software production on track to achieve comparable gains in productivity and reliability.
The ATP focused program in this area is the result of significant input from the information technology industry. There is no shortage of innovative ideas to make software generation dramatically simpler, more automated, more reliable. But the realities of the marketplace are such that the typical software company is struggling to finish its next job on time and on budget and cannot afford to engage in expensive, long-term, high-risk research to change the way software is developed. The widespread frustration within the industry was reflected in the dozens of software-related white papers the ATP received and at workshops sponsored by the ATP.
The Component-Based Software focused program was launched in April 1994, with awards from the first competition announced in October. A variety of companies are now enthusiastically tackling various aspects of the vision outlined above. Many are small, innovative start-up companies. Examples include Aesthetic Solutions, Applied Parallel Technologies, Continuum Systems, Cubicon, SciComp, and Reasoning Systems. These are not household words today, but if the ATP program is successful, one or more of them may well be, five to 10 years from now. Several larger computer companies have already come forward with matching funds for smaller companies who will be teaming with them. These firms will become partners in commercialization if the R&D cost shared by the companies and ATP is successful. These strategic alliances would not have occurred without the ATP.