Steve Kelman looks at four fundamental factors that can determine whether a large-scale IT project fails or flourishes.
Ever since the Obama Administration's health care websites crashed last fall, Washington has been trying to figure out what's wrong with the federal government's ability to do large-scale IT projects. Last week's story about the cave in Pennsylvania where federal retiree applications are processed manually, as reported in The Washington Post, only added to the overall impression that when it comes to information technology the federal government is the gang that can't shoot straight.
But if you think about it for a moment the government often achieves results ranging from credible to spectacular through contracting, including for IT. Just to take another story very much in the news — NSA snooping and hacking of electronic communications — the criticism against the agency is hardly that it haplessly lacks the capabilities to do its mission. On the contrary, its technological innovations and successes are the driving force behind the public outcry. Ditto with the drones developed for the Department of Defense. The Defense Advanced Research Projects Agency (DARPA) has, over the years, signed contracts with researchers doing both basic research but also development that have played central, and sometimes the total, roles in developing the Internet, GPS technology, and Siri voice recognition software now used on smartphones. NASA has deployed a succession of amazing technologies over the decades — think of the moon mission and space walks.
There are even some pretty solid achievements among civilian agencies. The VA developed for its hospital system (in-house actually) an IT application that, as a patient is about to receive a drug, scans the patient's wristband and the drug container to make sure the drug is the one actually prescribed by the doctor and to check for interactions with other drugs. The CDC's National Healthcare Safety Network provides real-time data to hospitals, and to policymakers, about trends in hospital-acquired infections, so hospitals can compare their situation with other hospitals as well as with their own history. We may note, furthermore, essentially all other back-office HR systems in the government are automated and function fine — for example, Social Security records and payments, a far bigger system than OPM's retirement system for federal employees.
So this leads us to the following question: can we make any generalizations about government contracting successes versus debacles? A few thoughts:
1. The "coolness" factor
A common factor in many successes is that the agency mission, or the specific program being supported by contracting, is itself attractive and even cool. NASA regularly scores at the top of the "best places to work in the federal government" listing published by the Partnership for Public Service, as do agencies in the intelligence community. VA hospitals and the CDC save lives. This probably means that, on the whole, these agencies attract more talented and committed employees, who do a better job figuring out what the organization needs when it contracts and do a better job managing the contract (and the government's interaction with the contractor). It probably also means that the most talented contractor employees tend to get assigned to projects at these organizations. Contrast this with just how inspiring it is to work on OPM's mandate, the government's incredibly bureaucratic and uncreative personnel system. So probably the simplest answer is that these contracts — and probably government more generally — has an easier time succeeding the more attractive the mission and, hence, the better the employees.
2. Multiple agencies
One common feature of the two most-recent high-profile failures Healthcare.gov and the OPM retirement processing system is that both require coordination across many agencies, to deal with input into a central system that exists in non-standardized databases and uses different technologies. Healthcare.gov, despite the popular perception it was a mere website, actually had to bring together huge databases from the IRS, 50 states, and other organizations. OPM has to deal with multiple formats of data about employees who are retiring that different agencies compile. It is difficult for one agency, particularly a relatively weak one such as OPM or the Center for Medicare and Medicaid Services, to get all these other agencies to collaborate. (Two giants, DoD and the Department of Veterans Affairs, have feuded for decades about whose system should form the basis for a unified health record that would transition a soldier from active duty to veteran status.) Success stories are pretty much always done in just one agency.
3. Dysfunctional procurement practices
The whole government — the successes as well as the failures — also lumbers with a ball and chain around its neck in the form of some dysfunctional procurement practices, which makes success more difficult for everyone. (Note that the intelligence community and DARPA are not subject to these rules.) The most strategically damaging of these is the overly bureaucratic way the government evaluates vendor past performance in making contract awards. In the private sector, probably the key incentive for vendors to serve their customers is a brutal system of past performance — if the customer is dissatisfied, the vendor is history. Prior to the Clinton administration, a misguided fear of giving government officials an ability to exercise discretion prevented any use of past performance in making new contract awards at all. Happily, this is no longer the case. But the system is far more bureaucratic than the brutal past performance system that is applied in the commercial world. An exaggerated notion of due process protections for contractors means that the regulations allow them to contest unfavorable ratings to the superior of the rater; this means that for a rating official, giving a contractor a bad report card is an invitation to spend countless hours defending one's judgment. The system has also suffered from a lack of senior management attention, despite its centrality as a lever dramatically to improve contractor performance, allowing excessive caution by most contracting officials to rule. The result has been a situation where most past performance ratings are plain vanilla, making it hard to use them as a differentiator in awarding contracts.
4. Big bang procurements
Many of the government's IT projects — including OPM's benefits automation — fail because the government believes it needs to specify all the features of the new system at the beginning of the procurement, rather than developing projects with incremental functionality that improves over time. The arcane and multiple rules (for different employees) federal pension system is hard to modernize in one fell swoop, but that's what OPM tried. This occurs for two reasons. One relates to the absence of a robust past performance system: deprived of a powerful past performance incentive, the government feels, in order to avoid being ripped off, it must overly specify every feature of an IT project A second problem is that the government has trouble complying with statutory requirements for competition among suppliers that would normally exist between increments; the more that is specified upfront at the time of competition for the original contract, the less the need to re-compete each stage of the process. There are ways to deal with this second problem, but they require creativity that the government has often been unwilling to apply.
Reforming the contracting system
What remedies does this analysis suggest for improving the value the government gets from contracting? The first is a management challenge for the agencies with less-glamorous missions, to nonetheless enthuse employees about their missions the way, say, Lay's potato chips manages to do for its (at least) equally mundane product. Unfortunately, inspiring management is not necessarily something for which the government is famous.
Can we do anything to give more powerful incentives to contractors to perform well? Transforming the government's paper tiger past performance system into a real tiger should be a priority for anyone who cares about the performance of the contracting system. A simple regulatory change (or legislation if the regulators won't act) could eliminate the excessive privilege awarded contractors to contest past performance ratings, and replace it with simply the ability of the contractor to enter their own version of events into the contract file.
Also, there are various experiments going on in government for "pay for success" or "share in savings" contracting, which should be extended. The growth of this kind of contracting has been stymied by some arcane budgeting rules, which should be waved by statute for these kinds of contracts on an experimental basis. There has also been objection to potential excess contractor profits achieved by the generation of very large savings, from those who would apparently rather see a contract fail than have the contractor make too much money. These contracts are also complicated to negotiate, and the government needs special expertise (including perhaps outside experts) to help.
In the most-positive news in the procurement system in recent years, the government has begun to use procurement contests, where the agency says what it wants and gives a pre-determined monetary prize to whomever comes up with the first, or best, solution. If no solution is developed, the government pays nothing.Not surprisingly, DARPA pioneered this technique with a contest a number of years ago to develop an all-terrestrial vehicle that could successfully navigate a desert obstacle course on Mars. Contests mean the government pays only for success. Contests also avoid many of the arcane features of the conventional procurement process. And they encourage the entry of non-traditional, garage-type players into the system, rather than just the usual suspects: lots of amateur teams entered the DARPA contest, and a Federal Trade Commission contest to develop an application to screen robocalls was won by a high school student. GSA's contest website, which makes it easier for agencies to organize procurement contests, won the Ford Foundation/Kennedy School Innovations in American Government award this year.
Contracting success is not impossible. Even in this era of highly publicized failures, success exists. But to make it more likely, we need to pay more attention to management and to remove some of the millstones around the government's neck.
Note: This piece also appeared on the Brooking Institution's FixGov blog.
NEXT STORY: 18F unveils first pilot app