Partial to dark suits and wraparound shades, Depew kept his gray hair closely cropped and a pistol holstered on his belt. He was a G-man's G-man. And he embraced technology with an almost evangelical zeal. When he was working the New Jersey fuel oil case in the early 1990s, Depew not only coded his own case management database using the FoxPro program, but he put it on floppy disks and gave it to any agent who asked for a copy.

Depew joined a team of seven that assessed the Web interface SAIC was designing for the ACS system. When completed, the interface would let agents point and click their way through the tedious process of filling out official forms, but not much else. Recognizing the limitations of the interface and ACS, Chiaradio and Depew met with Dies. They convinced him, and later the director himself, that the bureau needed an entirely new database, graphical user interface, and applications, which would let agents search across various investigations to find relationships to their own cases. The new case management system would host millions of records containing information on everything from witnesses, suspects, and informants to evidence such as documents, photos, and audio recordings. To address concerns being raised by intelligence experts and lawmakers in the wake of 9/11, these records would be accessible to both the FBI's agents and its intelligence analysts. Chiaradio dubbed the new system the Virtual Case File.

Dies wanted to provide agents with this software as fast as possible. In Depew's view that meant "shooting from the hip." This cavalier approach to software development would prove fatal to the VCF. Today, many organizations rely on a blueprint—known in IT parlance as an enterprise architecture—to guide hardware and software investment decisions. This blueprint describes at a high level an organization's mission and operations, how it organizes and uses technology to accomplish its tasks, and how the IT system is structured and designed to achieve those objectives. Besides describing how an organization operates currently, the enterprise architecture also states how it wants to operate in the future, and includes a road map—a transition plan—for getting there.

The problem was, the FBI didn't have such a blueprint, as numerous reports from the Government Accountability Office, the DOJ's inspector general, and the National Research Council subsequently pointed out. Without it, the bureau could not, as a 2004 report from the NRC stated, "make coherent or consistent operational or technical decisions" about linking databases, creating policies and methods for sharing data, and making tradeoffs between information access and security.

With no detailed description of the FBI's processes and IT infrastructure as a guideline, Depew said that his team of agents began "to feel our way in the dark," to characterize investigative processes such as witness interviews and surveillance operations and map them to the FBI's software and databases. Over a six-week period in the fall of 2001, Depew's group defined how agents worked, how they gathered information, and how that information was fed into ACS. Working with engineers from SAIC, they drew up diagrams and flowcharts of how the case management system operated then and how they wanted the new case management system, the VCF, to operate in the future. Mueller himself attended one of these meetings to tell the agents to design a system that would work best for them and not to feel constrained by 50-year-old business rules.

Depew's team also called in people from across the FBI: a dozen in the first few weeks; 40 by the end of November. These "subject matter experts" explained how their divisions or units functioned internally and with the rest of the bureau.

In December 2001, the FBI asked SAIC to stop building a Web front end for the old programs. (Later, FBI computer specialists would create a Web interface as a stopgap, which is still used by agents today, until the VCF was delivered.) Instead, SAIC was asked to devise a new application, database, and graphical user interface to completely replace ACS.

To formally define what users needed the VCF to do for them, SAIC embarked on a series of Joint Application Development (JAD) sessions. In these meetings, Depew's team of agents and experts got together with a group of SAIC engineers to hash out what functions the VCF would perform. Ideas captured in these sessions formed the basis of the requirements document that guided SAIC's application designers and programmers.

In January 2002, the FBI requested an additional $70 million to accelerate Trilogy; Congress went further, approving $78 million. DynCorp committed to delivering its two components by July 2002. SAIC agreed to deliver the initial version of the VCF in December 2003 instead of June 2004.

SAIC and the FBI were now committed to creating an entirely new case management system in 22 months, which would replace ACS in one fell swoop, using a risky maneuver known in the IT business as a flash cutover. Basically, people would log off from ACS on Friday afternoon and log on to the new system on Monday morning. Once the cutover happened, there was no going back, even if it turned out that the VCF didn't work. And there was no plan B.

But while the Trilogy contracts were changed to reflect the aggressive new deadlines, neither the original software contract nor the modified one specified any formal criteria for the FBI to use to accept or reject the finished VCF software, as the Inspector General reported earlier this year. Furthermore, those contracts specified no formal project schedules at all, let alone milestones that SAIC and DynCorp were contractually obligated to meet on the way to final delivery.

In reaction to the new deadline, SAIC broke its VCF development group into eight teams, working in parallel on different functional pieces of the program, in order to finish the job faster. But the eight threads would later prove too difficult for SAIC to combine into a single system. Nevertheless, in an interview at SAIC's McLean, Va., office complex, Rick Reynolds, vice president and operations manager for SAIC, defended the decision to change tactics. "People forget the urgency that we were underand our customer was under. And we were right beside them," he declared. "Wewere in the foxhole together."