Last October, most of America was surprised by the glitches, delays and breakdowns of HealthCare.gov, the federal government's website for health insurance enrollment. But for those who follow how government acquires, develops and deploys computer-based services, the results weren't so much surprising as disappointing.
Why? Because it's happened often, despite a strong understanding of how to avoid such calamities (more on that below). While the federal government has its fair share of infamous IT failures (the IRS's tax modernization program and the FAA's NextGen air transportation initiative, for example), state and local governments have their own, rather long list of dubious achievements that keeps growing.
New York City watched the costs of its payroll modernization project grow from $63 million to $700 million before pulling the plug. The state of Texas had its seven-year, $863 million outsourcing deal with IBM that was plagued by problems. Recently, the Commonwealth of Massachusetts has tried to gain control of its new, $46 million unemployment claim system that has suffered from, you got it, glitches, delays and breakdowns.
Besides their huge costs, what troubled IT systems have in common is complexity. They involve many stakeholders, often from many kinds of organizations including governments, all kinds of data from many disparate sources, and a bewildering array of technologies, software applications, business rules and organizational cultures and contexts. Solving today's complex public problems rests on our ability to develop successful complex systems. Unfortunately complex systems suffer from a high rate of failure. The success rate for all IT projects is only 39 percent, according to the Standish Group, and just 10 percent for projects with budgets over $10 million.
Still, as the Boston Globe
pointed out in its investigation of the numerous state IT projects that have crashed and burned, government has a bias towards building big when it comes to IT. Part of it has to do with funding. A reliance on capital budgets to fund IT projects results in big, one-time systems that can take years to complete. But government IT projects have also grown big because of the complexity of delivering more holistic solutions to constituents. Just look at the online health exchanges, built by both the feds and the states, which pull data from numerous sources to verify the various levels of eligibility for enrollees. Similarly, the jobless system launched in Massachusetts last year is a complex piece of technology that replaces a system that was three decades old, and was designed to be deployed in phases.
Like this story? If so, subscribe to Government Technology's daily newsletter.
To make matters worse, large projects not only fail more often, they deliver less, according to McKinsey, the consulting firm. Specifically, half of IT projects with budgets of over $15 million run 45 percent over budget, are 7 percent behind schedule and deliver 56 percent less functionality than predicted, according to a report released in 2012. And it's not just government that suffer this fate, but private sector IT projects as well.
Greater complexity leads to greater risk, says Theresa Pardo, director of the Center for Technology in Government at the University of Albany, SUNY, which, for the past 20 years, has been helping governments around the world navigate public sector information technology. Pardo, who has a Ph.D. in information science, says complex public sector IT innovations projects require a very high level of coordination across various levels of governments, agencies, units, non-governmental partners and contractors. That makes them unique. "Most traditional system development efforts do not account for the level of coordination necessary in the development of such complex systems," she says. "Our work shows, the more boundaries crossed, the more critical coordination becomes."
When you add high public expectations and political pressure to act quickly, then things can get even more risky. "Failure occurs when the pressure to act exceeds the opportunity to understand," says Pardo. From her perspective, in such complex systems efforts, full appreciation for the consequences of each action is difficult to achieve. To do so requires finely-tuned and expert coordination. This all takes time, time that doesn't seem to have been available to those developing HealthCare.gov.
Another problem that bedevils complex system development in government is ambiguity when it comes to responsibilities. "When many actors are involved, investing time to clarify roles and responsibilities and to revisit those roles over time is absolutely critical. From what I've read about HealthCare.gov, creating sharp clarity about relative roles and responsibilities and coordination mechanisms to manage and revisit those roles as conditions changed was a challenge." says Pardo.
So, what can government do to get IT projects out of the newspaper headlines and into operational mode? It starts with what Pardo calls a "before the beginning" analysis that tests assumptions and, most importantly, establishes who the true stakeholders are in the system, not just the main groups involved, but the ultimate user. "It sounds obvious, but we have found that the answer to the question -- who are the stakeholders and who are the users -- can vary according to who you talk to," says Pardo. It's the same with assumptions. When assumptions are made, they must be tested before they fuel design decisions. If they go untested and are built into the system, the consequences can be unexpected and significant.
The HealthCare.gov project team assumed that on the first day the exchange went live, the number of people using the system would be low. And while that may have been correct, in terms of those who planned to actually enroll, the team clearly did not factor in all those millions of people who were curious and wanted to see what it looked like. The initial assumption became a business rule that affected HealthCare.gov's load factor. The result was a huge digital traffic jam on the first day that overwhelmed the system.
Pardo says the success of complex government IT systems depends on several other factors, such as understanding the context within which a system is being developed; and having people on staff who have the capability to manage work across department (and governmental and sectoral and professional, to name a few) boundaries. In other words, it requires top-notch management and leadership skills. "Attention to the time and skills required for expert collaboration and coordination is often overlooked," she says. "You need 'super' project managers, who have the skill-sets to ensure all actions are coordinated across multiple boundaries and are sensitive to shifting realities."