The Knowledge Gap

The only network the FBI ignores is its own


Last month, at just about the same instant the FBI was granted yet more power to snoop around on the private lives of citizens, the watchdog arm of Congress reported that the Bureau has yet to patch up its own information infrastructure. The General Accounting Office wound up its latest look at the FBI's much-delayed computer upgrade and found it still a mess.

The cost of this deficiency is difficult, if not impossible, to measure. After all, if you don't know what you are missing, you don't know what you are missing. But it is safe to infer that data remains locked up in walled-off crannies of the Bureau, out of the reach of investigators trying to piece together related info. In effect, Osama bin Laden's last known Tora Bora mailing address could reside on some forgotten disk drive with no way to find it.

This sad state is nothing new. Back in September 2000 the GAO found that "the FBI had over 13,000 desktop computers that were four to eight years old and could not run basic software packages." In December 2001 a Department of Justice review found 234 isolated, "stove-piped" applications on 187 different servers. The GAO observes that each of these servers "had its own unique databases and did not share information with other applications or with other government agencies."

In June 2002 the GAO was back to report that the FBI's information technology remained backward, lacking even a "fully functional E-mail system." This led the GAO to conclude that "these deficiencies served to significantly hamper the FBI's ability to share important and time-sensitive information internally and externally with other intelligence and law enforcement agencies."

And just about a year ago the FBI's computer missteps were again chronicled by the Justice Department's Inspector General. The IG found transition to the Trilogy data-sharing system hamstrung by a lack of Microsoft Word macros. Microsoft Word macros?

The crux of the problem seems to the Bureau's inability or refusal to come up with what is known as an enterprise architecture. Scratch any multinational operation with thousands of employees and you'll hit EA. This roadmap for computer systems lays out just what it is computers try to accomplish in the operation, and supplies the framework for who will do it and how it will be done.

Big enterprises need such a blueprint to keep from duplicating functions, to ensure interoperability, and to make sure that the mission fits the master. You wouldn't want your great widget-demand forecasting software to allow your widget-factory managers access to your corporate payroll, as an absurd example.

Yet despite the obvious need to come up with some sort of plan for a 27,000-strong, 450-office enterprise, the Bureau seems steadfast in its swaggering Alfonso Bedoya stance: "Plan? We don't need no stinking plan!" The GAO notes that the FBI has just now, 32 months into the current "reform," begun to populate the architecting effort with warm bodies.

The GAO calls the foot-dragging an indication of a failure to make EA an "institutional management priority." That's bureaucrat-speak for nobody cares. A sure sign of low regard is that the FBI is still developing its "as-is" architectural descriptions. This is another way of saying that the FBI does not even have a firm grasp of what it spends $800 million a year on now, let alone what it would like to do in the future.

More ominously, the GAO found that what little EA work is being done lacks input from the counterterrorism and counterintelligence fiefdoms within the Bureau. Both of those units no doubt house some of the Bureau's most nifty black boxes, so it might seem that trying to tie them to the rest of the enterprise is like hitching racehorse to an oxcart. But this would be shortsighted in the extreme.

As the flight training school fiasco demonstrated, with the FBI unable to act on tips generated across the country, it is not enough just to have a handful of incident reports sitting in file cabinets while Washington calls the tune. There must be a way for field offices to pull data together and easily share it, a peer-to-peer network for G-men. This is the opposite of what the Bureau's default architecture assumes. A few specialized outfits in D.C. are being tasked with hunting down a list of known bad guys in near real-time.

Such a top-down approach is almost sure to fail while subjecting millions of Americans to needless invasions of their privacy. Better to step back and do the decidedly unglamorous, sure to frustrate, and downright boring job of making all the existing boxes work before employing sexy new subpoena powers to fill up shiny new boxes with gigs of fresh, unread data.