InformationWeek

January 12, 1998

Crash-Landing Ahead?

Rocked by a series of disastrous mega-projects, federal agencies are working to change how they manage IT. But the FAA, one of the worst offenders, insists on going it alone.

By Edward Cone

asten your seat belts: The Federal Aviation Administration is about to launch another huge computer system. The FAA's $1 billion Stars (Standard Terminal Automation Replacement System) project, scheduled to begin deployment late this year, is intended to update the antiquated, failure-prone computers and radar displays that monitor all airspace within about 50 miles of U.S. airports. As part of a constellation of interrelated systems that could cost another $11 billion through 2003, Stars will exchange flight-plan data with air-traffic automation systems, terminal-control facilities, and surveillance radars across the country. Stars comes on the heels of the FAA's colossal modernization failure, the Advanced Automation System (AAS). That effort, along with several other failed FAA projects, took 16 years of effort and cost taxpayers $23 billion. But it failed to revamp much of the FAA's IT infrastructure, which supports the nation's air-traffic flow. Too bad -- some of these systems still run on vacuum tubes.

'Learned From Failures'
This time, though, the FAA swears it's going to get things right. "We're light-years ahead of where we were," says Steven Zaidman, the FAA's director of system architecture. "We have learned from our past failures."

Beyond Stars' front-line importance to air safety, the system is a crucial test of the FAA's management competence. But given the FAA's history of failure and continuing resistance to managerial change, many close observers find it hard to be confident it's a test the FAA will pass. "FAA management at top levels in Washington is short-changing the public and endangering their lives," said Sen. Alfonse D'Amato (R-N.Y.), speaking at a congressional hearing last summer that included testimony on near misses and dysfunctional systems. "It's a joke, a travesty, and it's not getting better-it's getting worse."

Adds the General Accounting Office, the investigative arm of Congress, in a recent report: "Management structure [at the FAA]...is not effective." The GAO cites the FAA's lack of a CIO as a major problem and adds, "No FAA organizational entity is responsible for developing and maintaining the technical Air Traffic Control architecture." That leaves the agency "at risk of making ill-informed decisions on critical multimillion-dollar, even billion-dollar, air-traffic-control systems," the GAO adds.

Unfortunately, calamitous IT mega-projects at federal agencies ranging from the Internal Revenue Service to the Department of Defense are nothing new. A 1994 report by the Senate Committee on Governmental A ffairs described a grim and widespread situation: "The federal government continues to operate old, obsolete computer systems while it has wasted billions of dollars in failed computer modernization efforts." (But one agency with a reputation for IT success is the Social Security Administration; see related story, " Social Security Gets It Right .")

The Senate report, "Computer Chaos," blamed the failures on "poor management, inadequate planning, and an acquisition process that is too cumbersome." Says David McLure, the GAO's assistant director, "The government's track record on large systems is bleak." In fact, sources close to the federal budget process sayfunding for several major IT programs at government agencies could be cut from the 1999 budget if they don't show signs of progress.

The government's prospects are mixed. On just one issue-year 2000 preparations-the government's Office of Management and Budget (OMB) found in December that fewer than 30% of Washingto n's nearly 8,600 critical systems are year 2000 compliant. While that marked a substantial increase over the previous quarter, the federal government's total estimated cost for year 2000 conversions rose in the same quarter by $100 million, to $3.9 billion.

But even in this dismal context, the people responsible for making your flight land safely stand out for their ineptitude, critics say. "The FAA is mind-boggling," says Don Mullinax, chief investigator for the Senate's Governmental Affairs Committee.

What Went Wrong?
The FAA has little to show for the $7.6 billion spent on AAS, the one-time centerpiece of Air Traffic Control modernization. What went wrong with AAS? Everything, according to the GAO: "FAA did not recognize the technical complexity of the effort, realistically estimate the resources required, adequately oversee its contractors' activities, or effectively control system requirements."

Similarly, the FAA's record on year 2000 conversions is especially bad. While nearly all federal agencies have completed year 2000 assessments and are now renovating code, the FAA has assessed only 38% of its systems-and that does not include an additional 245 critical systems the agency has recently identified, according to the OMB.

Still, thanks to its political muscle, the FAA successfully lobbied for the privilege of conducting its own management and procurement reforms instead of submitting to the congressionally mandated reforms that followed the "Computer Chaos" report (see related story, " Federal CIOs Look Past Failures "). William Cohen, who while a senator was the report's sponsor (he is now the Secretary of Defense), cited the FAA's "consistently poor management" and called the agency "ill-prepared to accept the responsibilities" of fixing its own problems. Nevertheless, the FAA was allowed to go its own way. While other federal agencies are required to have CIOs by the IT Management Reform Act of 1996-also known as the C linger-Cohen Act-the FAA was exempted and still does not have a true CIO.

'Like A Business'
Still, the FAA claims its internal reforms, which include a major overhaul of procurement procedures and some managerial shuffling, are paying off handsomely. Zaidman calls the FAA's acquisition-management reforms, instituted in April 1996, an important success. "We no longer procure systems, we acquire them," he says, eschewing the old governmental purchasing jargon to emphasize the agency's move away from cumbersome bidding rules. "We don't send out specs, we look for functionality. It lets us operate more like a business and lets industry help us and negotiate with us." Zaidman adds that the revised practices have cut in half the time needed by the FAA to award IT contracts.

One FAA program has moved ahead under the new purchasing regulations: Oasis (Operational and Supportability Implementation System), a $120 million project that provides emergency assistance and weather briefings to pilots.

Bill Greenwalt, a Senate staffer who helped write Cohen's "Computer Chaos" report, acknowledges that there has been a "major shift" on buying and contracting at the FAA, but he questions whether this alone can turn things around. "Now we can buy faster and cheaper. Smarter? That's still up in the air," says Greenwalt. "Buying faster can mean paying a premium so something can sit in a warehouse."

Fixing the way the FAA buys computers is only half the battle. Says one former Senate staffer who now works at a major federal vendor, "It's the management, not the rules. Their systems are poor because management isn't doing its job."

Zaidman responds that the FAA has made significant managerial improvements. "We have changed our senior-level management," says the 23-year FAA veteran. "We have restructured ourselves into integrated product teams." Under Zaidman's boss, George Donohue, the FAA has begun to focus on off-the-shelf technology, maintenance costs, and return on investment. "We need modular sy stems and interoperability," says Zaidman. "We have more technical and business skills." He points to the agency's master plan for its "system of systems" as an example of the FAA's progress. The next iteration of the plan, National Airspace System architecture version 3.0, is due early this year.

No Detailed Plan
Still, the FAA's persistent critics demand more. The GAO report praises the FAA for its logical blueprint of the sprawling National Airspace architecture, but points out the lack of a detailed technical plan for fitting it all together as a major ongoing problem. "Without a single, unified technical architecture," says the report, "compatibility and interoperability across and among all systems is highly unlikely."

No big deal, says the FAA. "We don't envision having a complete document of technical specifications that we would publish," says Zaidman. While the GAO recommends that a complete technical architecture be completed before the FAA replaces its 30-year-old Host Computer System-"the heart of our infrastructure," says Zaidman-that system is on schedule for replacement next year. The Host gets data from other systems such as weather computers and air traffic radars and processes it for controllers' use. Its feeder systems use several different communications protocols and data formats, which must be translated for Host by an expensive intermediary system.

Culture Of Denial
That proliferation of protocols and formats is typical. GAO investigators identified the FAA as a software version of the Tower of Babel, with 53 programming languages in use for 54 systems surveyed. The Host computer runs on machine code and the Jovial language (widely used in the military), while other systems use languages including AIX, Unix, and DOS.

A majority of the 10 distinct National Airspace modernization programs also lack technical architectures, and the ones that did have a blueprint had mutually incompatible communications protocols.

One thing that doesn't seem very different at the "new" FAA is the way it deals with criticism. "They go into their shell and wait for people to go away," says the Senate's Mullinax. "It's unbelievable that a government-funded organization has that kind of culture."

Defense Secretary Cohen once described an FAA culture of "denying there is a problem, defending its poor performance, and deflecting criticism away from the agency." Says Zaidman, "I am not familiar with Sen. Cohen's criticisms, so I can't help you there." Yet despite repeated requests for interviews with other managers, Zaidman was the only FAA official InformationWeek was permitted to interview. The FAA said Jane Garvey, a highly regarded administrator who took the helm at the FAA in August, was unavailable for comment.

It's against this backdrop of failure and obfuscation that the Stars program is proceeding. A joint project with the Department of Defense-another agency known for poor systems management-Stars is a successor program to AAS, which Zaidman describes as "one of our biggest failures." The collapse of AAS extended the life of the antiquated ARTS (Automated Radar Terminal Systems) project, built on '70s and '80s technology and increasingly difficult to maintain as old parts become harder to find. Stars includes new processors, software, and keyboards at more than 170 Terminal Radar Approach Control facilities.

Zaidman says Stars shows the FAA's new face. His first example: The FAA awarded a contract to Raytheon Corp. of Lexington, Mass., in just six months rather than the 18 months the agency had expected it would take. "We're using a lot of commercially available hardware and software, including approximately 800,000 lines of computer code," says Zaidman. The first Stars unit is scheduled to be installed in Boston in December 1998, with 172 facilities to get Stars installations through February 2005.

Flying Feud
Compounding the pressure over Stars is the continuing bad blood between FAA management and air-traffic controllers. It's a feud th at predates the disastrous Patco controllers' strike of 1981, which ended with President Reagan's dismissal of more than 11,000 striking air-traffic controllers. The controllers, now represented by the National Air Traffic Controllers Association, are stressed by the huge increase in air traffic since deregulation and distrust a management hierarchy that they say displays no more people skills than it does technological prowess. "Stars isn't going as well as they say, and it's not the answer to our problems anyway," says a labor representative who requested anonymity. In other forums, air-traffic controllers have criticized Stars for cumbersome, complicated controls and unclear terminal displays.

Getting Stars out on time and on budget would be a huge step forward for the FAA, but it would be just the first step of many. The system of systems that make up the National Airspace System architecture has almost a dozen different functional areas, including a digital communications system, an upgraded en-route tracking system, and a navigation and landing system based on a global positioning system.

With Congress taking a much tougher stance on funding major computer systems, the FAA can't afford any more crashes.

Back to News in Review