When Peter Levine, the Defense Department’s deputy chief management officer, was challenged at a House subcommittee hearing last week on what he’s doing to improve the department’s less than stellar track record for deploying business IT systems, he began by disputing the premise of the question.
“We don’t have a not-stellar record, we have a horrendous record,” he said. “I think that of all the things the department does badly, that’s one of the things we do the worst.”
Levine, a former longtime Capitol Hill staffer who’s been on the job as DCMO since last May, offered a fairly clear diagnosis that boils down to a too-many-cooks-in-the-kitchen problem: His office is in charge of overseeing and approving spending on business systems, the DoD chief information officer certifies the technical bits, but the Pentagon’s acquisition apparatus, in the end, calls most of the shots that drag out systems’ schedules and make them more expensive than necessary.
“Let’s say we’ve identified a $20 million fix to a problem where all we’re looking to do is tinker with an existing system. We have a process today where we go through an investment review and we identify a solution, but we then have to throw it over the transom to the acquisition community which might spend $20 million just to set up a program office. And they start that process all over again,” he said. “And then they’ll come to us with a solution, which is, ‘Let’s build a whole new system from scratch.’ Well, that’s crazy.”
But Levine said his office, the DoD CIO and the office of the undersecretary of Defense for acquisition, logistics and technology have agreed to “harmonize” their separate processes. He offered few details about the ongoing effort, but said in his written testimony that the end goal is to ensure the folks handling business process approval, technical requirements and acquisition are talking to each other from the very start of a decision to upgrade a business system or buy a new one.
“It’s going to begin in the requirements process, carrying through to the consideration of alternatives and supporting the actual acquisition of a final IT capability,” he said. “Our goal is to better deliver capability, with appropriate oversight, while removing many of the burdens levied on the IT functional and program managers.”
Levine also signaled that DoD is likely to ask Congress to adjust the thresholds that trigger additional scrutiny and review from top acquisition officials when the department is buying large IT systems. Under current law, many major automated IT systems (MAIS) with budgets as low as $40 million are sifted through the same paperwork processes and acquisition rules as large, multibillion dollar hardware.
“What that means is that when we have a business system investment, we trigger a process on the acquisition side which is as big and as clumsy as the process we have on the acquisition side when we’re buying an aircraft carrier or a fighter aircraft or something like that. And if you’re buying a business system, I’m not sure that makes sense.”
Congress apparently thought it made sense when it enacted additional oversight and reporting requirements for major Defense IT systems. Many of the current rules came into play as part of the 2007 Defense authorization bill, about the time the Defense Integrated Military Human Resources System (DIMHRS) was starting to come off the rails at a cost of $1 billion with no usable capabilities, to cite one example (See also: Air Force Expeditionary Combat Support System).
The Government Accountability Office reported last year that many of DoD’s continuing problems with business systems have to do with the Pentagon’s impulse to customize off-the-shelf software in order to suit its longstanding business processes, rather than the other way around.
Levine said he agrees and is addressing the issue by refocusing DCMO’s oversight so that it takes account of broad portfolios of technology and how they affect DoD’s missions, as opposed to one-by-one reviews of individual business IT systems and whether they suit their immediate users.
“One of the things that we need to do is to recognize the business systems themselves are not going to solve our problems,” he said. “If you try to automate an old process without looking at it and figuring out how it works, you’re doomed to failure. We have tried many times to buy an off-the-shelf system and then the users of the system come in and tell us, ‘Well, that’s not exactly the data we want. We want this other data, because that’s what we had before,’ so we start tearing apart the guts of an off-the-shelf system. And before you know it, we’ve spent five times as much to re-engineer the system and to rebuild the system as the cost of the system itself. We have to control our own appetite.”