Summary

Billions of dollars in current and future Defense Department (DOD) weapons and command, control, communications, and intelligence systems depend on high-performance, correctly functioning, real-time computer systems that hold up under severe stresses. Yet the Pentagon's mission-critical systems continue to be plagued by a host of long-standing software development problems, including cost, schedule, and performance shortfalls. This report provides an overview of earlier GAO work on mission-critical systems. Many studies both by GAO and DOD have pointed out a variety of deficiencies, ranging from a lack of management attention to ill-defined system requirements to inadequate testing. Why has this situation persisted for so long? GAO notes that the understanding of software as a product and of software development as a process is not keeping pace with the growing complexity of existing and emerging mission-critical systems. The Pentagon is trying to overcome this dilemma through two main programs--the software action plan working group and the Corporate Information Management initiative. Whether these efforts will solve the military's formidable software problems is uncertain; there are no easy answers.

GAO found that: (1) management, requirements definition, and testing problems contribute to significant schedule delays, cost increases, and performance shortfalls of DOD mission-critical systems; (2) current system acquisition policies do not allow for more incremental or evolutionary development; (3) it is difficult to accurately measure the progress of developing software products; (4) mission-critical systems development attempts are accompanied by serious limitations; (5) the software action plan working group is addressing management issues without a formal written plan to guide their activities; (6) the DOD technology plan is intended to assess DOD software needs for mission-critical systems and to define technology investments over the next 15 years; and (7) the data administration program is attempting to provide consistent, unambiguous, and easily accessible data and minimize the cost and time required to make systems and data compatible.