Tackling Tax Technology

The project is large by necessity. After all, the IRS tax system produces more than $2 trillion in gross revenue each year and has the largest active customer base of any organization of comparable size. The system depends on a collection of more than 300 computer systems developed by the IRS over 35 years that store more than 100 terabytes of legacy data. The system comprises a network of 20 mainframes (until recently there were 67), 871 mid-range computers, more than 100,000 desktop PCs, 2,770 vendor-supplied software products and more than 50 million lines of computer code.

This combination of outdated, inflexible and incompatible systems results in inconsistent and unreliable information. It hinders IRS efforts to make changes that new tax laws require. Moreover, the older systems don't work well with the Internet, a necessity in light of IRS plans for improving service and cutting costs.

To solve these problems, the IRS and the PRIME Alliance, a group of contractors headed by El Segundo, Calif.-based Computer Sciences Corp. (CSC), must replace the 1960s systems that maintain taxpayer records, as well as other systems added as the tax code evolved, with Internet-compatible technology. The IRS-contractor team must consolidate hundreds of databases supporting IRS functions and business processes. The relational database system being put in place will, among other things, make online transactions possible. To achieve its goals, the team must develop one of the largest data stores ever created. In addition, the team must come up with a strategy for moving data from the old systems to the new while the IRS is running both systems concurrently. The team also must figure out how to retire aging databases in the proper sequence. While all of that is taking place, Rossotti's executive staff must prepare the organization to understand and support the new systems.

The project, always envisioned as grand in scope, has taken on a life of its own over the past few years. Although IRS officials have declined to set a specific date for getting the majority of tax payments online, Rossotti has said publicly that the program is intended to put IRS business systems on equal footing with the best private and public sector organizations.

Today, Rossotti is confident when he talks about the present and future success of modernization. "Every program goes through down periods and up periods. During the down periods, you have to remember that it's never as bad as it seems, and during the up periods, you have to remember that it's not as good as it seems," he says.

Business systems modernization will succeed, he says, if IRS leaders manage it correctly and acknowledge that the path is fraught with risks and blind alleys. "It's only people who are blindly optimistic who fail, because they don't realize they have to make adjustments along the way."

At this point, leaders inside and outside the IRS are generally positive about the outcome of modernization, which began in June 1999 and will continue for several years. Many goals have been met, and even Congress and the General Accounting Office-two of the IRS' harshest critics-believe Rossotti and his team are on the right track.

The IRS executive staff and the PRIME Alliance, headed by CSC and including IBM, KPMG Peat Marwick, Lucent Technologies, Northrop Grumman, SAIC and Unisys, have made important progress in instituting the controls necessary to effectively manage modernization, says Randolph C. Hite, GAO's director of IT systems issues. Hite also notes that the IRS has moved smartly ahead on many of the system projects involved in modernization.

Work in Progress

Great strides have been made, but they're not enough, according to Larry Levitan, chairman of the IRS Oversight Board, a nine-person independent panel formed in September 2000 to provide an independent strategic view of the agency. The oversight board was created by the 1998 Revenue Restructuring Act, which directed the IRS to revise its mission statement to provide greater emphasis on serving the public.

"Today, [the IRS] is still broken. It provides poor service to the taxpayers who come to [it] for help and does a mediocre job at best of enforcing the tax law," Levitan says. He is quick to point to some of the reasons for the agency's continuing shortcomings, including staff cuts-from 115,000 to 98,000 over the last decade-while the workload increased at least 20 percent. During that time, the IRS failed to put into use new technology, compounding its problems. "If you apply what happened to the IRS to any large commercial enterprise, that business would be out of business," Levitan notes. "But the IRS can't go out of business, because it collects the money that drives the government."

Levitan says the IRS is developing a good strategy and technology migration plan. He points to the strengthened team of contractors and IRS personnel, which has been bolstered by the addition of high-level technical experts and executives-including Rossotti. Levitan notes that Rossotti, a founder of Fairfax, Va., IT consulting firm American Management Systems Inc., is the first IRS commissioner who didn't practice tax law. Instead, he is skilled at running large, complex technology organizations. Levitan also approves of Rossotti's attempt to bring outside expertise into the program to strengthen and broaden its knowledge and experience base.

Recent GAO testimony generally agrees with Levitan's assertions. Published in May, "IRS Modernization: Continued Improvement in Management Capability Needed to Support Long-Term Transformation" (GAO-01-700T), noted that the IRS is making slow but steady progress in overhauling its structure, performance management system, business processes and information technology.

Before moving forward on any other front, the IRS must complete a comprehensive blueprint of its enterprise IT architecture, congressional auditors say. Only such a blueprint-a map of all projects and their progress in meeting the goals of the overall enterprise architecture-will ensure success, GAO has found. "These projects don't exist in a vacuum, but in relation to one another," notes Hite. He says the IRS should have finished the comprehensive blueprint 12 months ago and that although progress has been made, the agency is not where it should be. IRS officials promised the revamped blueprint by June 30.

That blueprint and the commitment it represents-to continually assess what the program needs, what each individual project needs and what the taxpayers need-is key to the overall success of the program, experts believe. "If you don't assess what you need and what the taxpayers need, you'll keep building systems that are either outdated or outmoded, and you won't know why you are doing what you're doing," says a Senate Governmental Affairs Committee staffer who has monitored modernization.

Rossotti agrees that a blueprint is essential, but says GAO's expectation that such a document can be compiled quickly is unrealistic. "The reality is that you have a hard time just doing architecture without actually having the experience of working on real applications," he says. Over the next few months, the tax modernization team will be filling in more detail on various projects, leading to an increasingly detailed blueprint that will be continually revised as progress is made, Rossotti says.

Another hot button for GAO was the lack of effective configuration management controls, which help ensure that a project as complex and large as this one runs smoothly. Simply put, configuration management ensures progress on each piece of the project is kept current with progress on all the other pieces. "They were relying on people talking to each other to share the current configuration of their respective systems. You just can't do that with an engineering project," Hite says. "You have to put in place the configuration management process of controls and databases to do this in a more formal way, so everyone will know one configuration item was changed on Project A and that it will affect these three other projects."

Jim Kennedy, program executive of the PRIME Alliance, says the contractor-IRS team already has made significant progress toward addressing the configuration management challenge and expects problems associated with it to be resolved shortly.

Although an increased focus on configuration management should go a long way toward improving the way modernization is run, more work is needed to implement formal program management, especially as critical projects reach delivery dates during the next year or two. Thus, the contractor-IRS team has set up an integrated program management office and has instituted other checks and balances.

The modernization team now holds monthly program management reviews. The reviews ensure that each new version of a system progresses smoothly from development to the operational or production stage, and provide status checks for each project.

About 150 people typically attend these open reviews, which are becoming more necessary as the complexity of the integrated projects grows, Kennedy says. But developing stronger program management and focus will take time. "Program management isn't something you achieve by reading a textbook," Rossotti says. "You have to play the sport to get competitive."

GAO also found the modernization program needs a performance management system, including rewards and incentives for IRS staff to provide good service and comply with system requirements. To help assure such compliance, the team is using a method called Enterprise Life Cycle (ELC) that the IRS and the PRIME Alliance developed jointly. The ELC functions as a how-to guide for program and project managers, defining the paths they can take, the steps they must complete and the items they must address to deliver a project successfully. It is based on CSC's Catalyst methodology.

Modernization program managers are so emphatic that every tenet of ELC be followed that projects come to a screeching halt if the methodology falls off track. "Failure to adhere to ELC guidelines will cause us not to allow a project to proceed through a major milestone transition until we have satisfied ourselves on matters of quality and business case support," explains John Reece, IRS deputy commissioner for modernization and chief information officer. Reece says that although the tenets of the ELC are stringent, they are worth following. "It's comprehensive and can sometimes be burdensome, but we try to manage our way through it," he says. The impending departure of Bert Concklin, who manages the Business Systems Modernization Office, isn't making program management any easier. Concklin, former chief of the Professional Services Council, an association of federal services contractors, was appointed associate commissioner for business systems modernization in June 2000. He has headed the program management office since then and is in charge of implementing IT solutions to modernize and improve IRS business systems.

Concklin announced in May that he would be leaving the IRS in early September to pursue other opportunities. Between the time Concklin leaves and a replacement is found-and that could take six to 12 months-Reece will don Concklin's hat as well as serving as CIO and modernization czar. Reece downplays the suggestion that he's spreading himself too thin, saying he'll contribute to many phases of modernization for as long as necessary.

Learning Curve

Virtually everybody connected to the undertaking-contractors, IRS officials and oversight leaders-realizes that a project as large and complex as tax systems modernization is bound to encounter rough spots. But with constant vigilance and a willingness to learn from experience, most believe the IRS can accomplish what once was considered impossible.

Even Levitan, one of the IRS' most outspoken critics, believes the agency will succeed. "I'm optimistic, but realistic. It's going to take a long time, it will be very expensive, and it will have significant risk."

Learning from experience is the key to success, PRIME program executive Kennedy believes. "There has been maturity on all sides. We've gone through the growing pains of starting up a program and understanding the roles and responsibilities of both parties. We've learned how to create a partnership within the PRIME Alliance itself, with outside contractors, and with IRS," he says.

GAO's Hite points to the recent completion of the design for the corporate account data engine (CADE), a repository for enterprise data, as a sign of progress. The data store, as CADE is often described, will provide a basis for creating and managing the information underlying all taxpayer-related software applications. It will displace the 30-year-old tape-based system of master files and related components containing 170 million taxpayer accounts.

CADE will replace the master files with a single database, the tax account data store. The data engine also will accept, validate and store tax return and account data, generate refund notices, calculate taxes, detect potential fraud, provide universal access to authoritative data with no limitations due to geographic boundaries, and allow processing of electronic refunds within 48 hours of receiving tax return information.

Completion of the CADE design is a big achievement, Hite says. But he notes that other projects have not achieved similar success. For example, the customer communications telephone routing system was not finished in time for the 2001 tax-filing season. The routing system will be as much as six months late, Hite says, but its tardiness isn't due to technical problems. The modernization team simply failed to set a realistic completion date.

Modernization problems will continue to emerge and Rossotti is determined to meet them head on. Current hurdles include building management capacity and succeeding with the first round of projects, but the program's challenges will be quite different a year from now, he says.