Think $US500 million is a lot of money? That's how much the FBI may end up spending over the next six years on its new case management system, dubbed Sentinel. Maybe you remember Sentinel's predecessor, Virtual Case File. The VCF project lasted four years and cost a mere $170 million. Of course, VCF was such a mess that when the project was killed one year ago this month, it was completely useless. Four years and $170 million later, the FBI had nothing to show for it. Now that's an expensive project.
Can the FBI do any better today with $500 million? Maybe.
The difference won't be the money. Three times the money, spent the same way, would just produce a hole three times as deep.
The difference will come in how the FBI uses that money. And right now, the bureau is acting as if every cent counts.
For example, instead of building a bright, shiny, all-new vehicle like it did with VCF, the FBI is going the used-car route for Sentinel. It's using the enterprise IT architecture it salvaged from the wreck of VCF. It's borrowing an experienced IT project manager from the CIA and an acquisition contract from the National Institutes of Health. And it has squeezed $97 million out of divisional budgets for the first of the project's four phases.
That's another difference with Sentinel: it has four clearly defined, overlapping phases, each with deliverables. VCF didn't have that. Its developers kept ploughing ahead as technology changed and requirements shifted because of 9/11. After four years, VCF delivered nothing useful.
But at the end of Sentinel's Phase I, 18 months from now, FBI employees will have a Web portal for access to their creaky old Automated Case Support system, plus some improved case- summary and indexing features. Later phases will layer on workflow, indexing and paperless case management. But even if the next three phases never happen, users will have received something for their money.
All this represents a very smart approach, especially because the FBI has two problems to solve. One is delivering Sentinel. The other is restoring its credibility when it comes to big IT projects. With $170 million already down the tubes, the bureau can't afford another botched effort.
Unfortunately, at a casual glance, it looks like Sentinel is already in trouble. The contract was supposed to be awarded last November, then in December. It finally went to Lockheed Martin last week.
That's a nearly six-month schedule slip already. So much for project management cred, right?
Now look closer. It turns out that the original schedule didn't leave enough time between the request for proposals and the bids. Bidders raised technical questions. The bid evaluation team needed more data from bidders. Some elements of the project that were originally planned to be done before bidding ended, including the system security plan, depended on technical details of the bids. The project plan had to be reshuffled.
So, the FBI's management faced a tough decision early on: hit the deadline and risk the project, or miss the deadline and do it right. Instead of cheating on project management to accommodate an aggressive schedule, the FBI sacrificed the Gantt chart to make sure the project got a good start.
Sure, it looks bad when bid-award day arrives months later than it was supposed to. But unresolved problems just get more expensive over time. If slipping the schedule at the start is the price for nailing down problems early, that little bit of embarrassment and delay is a bargain.
It may just require educating some members of Congress about the realities of software development when it's time to ask for the rest of that $500 million.
So, has the FBI got it right? This time, we won't have to wait four years to find out. Whether bureau users have their case management Web portal by the end of 2007 will tell us a lot about how well the FBI can do a big IT project the second time around.