Battling the Enemy Within

A billion-dollar fiasco
is just the tip of the military's
software problems

A major U.S. Army initiative to modernize thousands of aging computer systems has hit the skids, careening far beyond schedule and well over budget. The 10-year project, known as the Sustaining Base Information Services (SBIS) program, is supposed to replace some 3,700 automated applications by the year 2002. The current systems automate virtually every business function--from payroll and personnel management to budgeting and health care--at more than 380 installations worldwide. But after investing almost three years and about $158 million, the army has yet to receive a single replacement system.

The failure is significant not only because it strands the army with outdated software but also because SBIS is just one casualty among many. In January top Pentagon officials reportedly killed the larger Corporate Information Management (CIM) initiative, which for six years had tried to consolidate and modernize thousands of the armed services' old and redundant computer systems.

The Pentagon has not been tracking either costs or savings of CIM. But the Department of Defense projected in 1992 that CIM would help it cut $36 billion by 1997. The General Accounting Office (GAO), in contrast, concluded last July that "Defense continues to spend about $3 billion annually to develop and modernize automated information systems with little demonstrable benefit. Few redundant systems have been eliminated, and significant savings have not yet materialized."

Why is one of the most technologically advanced organizations so consistently humbled in its attempts to master business software? A close look at the troubles of SBIS reveals that inadequate software technology, industry incompetence, a f lawed procurement process and naive expectations all play a role.

The army conceived SBIS in 1992 to solve a long-festering problem: most of the computer systems that the armed services rely on to raise, organize, train, equip, deploy and sustain their forces are growing obsolete. Designed 20 or more years ago to run on equally ancient mainframes, the systems are becoming prohibitively expensive to maintain. The antiquated programs typically cannot share information with one another, and many force the army to work in ways that no longer make sense.

SBIS was to replace 3,700 largely incompatible systems with about 1,500 new applications. The new systems would all run on the same kinds of computers and networks and would store data in compatible ways. By eliminating duplication, shutting down mainframes and allowing information to flow smoothly, billions would be saved. And best of all, the systems would be based on the industry standards and so would be cheap and easy to upgrade.

The army wisely decided to split its ambitious program into phases. The first contract called only for the common infrastructure and 89 applications, which would take three years to develop. In June 1993 a team of companies led by IBM Federal Systems (which was sold to Loral six months later) beat out several competitors for the contract with a bid of $474 million.

IBM's winning proposal included techniques touted in the industry for their ability to make software development faster, less costly and less risky. Automated tools would boost programmer productivity. Designers would enlist users to help craft prototypes of the applications, so as to avoid expensive design changes later. Computer code already written for other systems would be reused.

Parts of the proposal should have raised questions, however. To back up claims that it could reuse more than 70 percent of existing code (about three times the industry average), IBM cited its work for the Federal Aviation Administration and Westpac Bank of Australia. But the FAA was forced to abandon much of IBM's work, at a loss of nearly $1 billion. Westpac was likewise left with little to show for its nearly $150-million investment and dropped IBM, with some critics accusing IBM of promising technology it could not deliver.

IBM and its successor Loral again face that charge, this time made by a former army official. "IBM had a conflict of interest from the beginning" because it has lucrative contracts to keep the old mainframes running, says Russell D. Varnado, who managed information technology acquisition for the Army Material Command until 1992.

Last December Varnado and a small software firm called Pentagen Technologies filed a federal whistle-blower suit against IBM, Loral and the army officials who manage SBIS. The action accuses IBM and Loral of contracting to perform tasks that they knew were beyond their abilities; it also accuses army officials of failing to enforce the contract. IBM and Loral are fighting the suit.

The charges are based in part on a report filed by Charlotte J. Lakey, who managed the SBIS program from its inception until April 1994. The report describes how the project slipped behind schedule from the outset. "[Loral] missed most of their deliverables," Lakey recalled in an interview, including "their system design plan, software development plan, communications plans--basic things like that."

Annoyed by the delays and alarmed when Loral proposed a software price that was "a lot higher" than expected, Lakey decided that the army should threaten to terminate the contract. But her superior overruled her, and several months later Lakey was removed from her post. In her final report, she suggested that "there needs to be a better contract mechanism...than hoping you get an honorable contractor."

Although Colonel Charles Mudd, the current SBIS program manager, says Loral is using the promised state-of-the-art techniques and limiting systems' designs to fit the budget, the estimated expense has skyrocketed. About $114 million of the $165 million set aside for software and services in the contract has already been obligated, even though no systems have been delivered (four are in testing). The latest estimate released by the army puts the life-cycle costs of SBIS's first phase at $1.4 billion.

For its extra billion, the DOD now expects considerably less: the army has cut back the number of applications to be built from 89 to just 19 and the number of installation sites from 128 to 43. So rather than replacing 985 of the army's 3,700 systems, this phase will apparently upgrade only about 180. Mudd attributes the reductions to budget cuts. But according to House Appropriations Committee staff, the SBIS budget increased 56 percent last year, from $62 million to $97 million. Mudd responds that he has been handed a "major budget cut" for next year. Paradoxically, cutting losses now could raise the price for SBIS, by prolonging the time until expensive old systems are replaced.

One lesson the DOD should learn from this experience--as it casts about for a strategy to replace CIM--is the virtue of patience, says Sanford F. Reigle, who has been investigating the initiative for the GAO. "It took them 30 years to get this screwed up," he says. "We got there slowly, and we'll get out of it slowly."

Indeed, in 1993, four days after William Perry, then deputy secretary of defense, ordered CIM to be accelerated so that all systems would be complete in three years, former director of defense information Paul Strassman objected in a memorandum to Perry. The DOD maintains some 11,000 major applications and perhaps 50,000 databases, he wrote: "The CIM goal to reverse engineer this inventory is 20 to 50 times bigger and twice as fast than anything ever attempted in the commercial sector. The DOD record to date in delivering on time even one million lines of code on schedule and on budget shows a 100% failure rate." Strassman's warning might have had more impact had he not resigned eight months earlier.

by W. Wayt Gibbs in San Francisco