Everyone agrees that the rollout of Healthcare.gov has been something between a fiasco and a disaster. One of the mysteries is how a famously tech-savvy administration, headed by a president whose campaigns broke new ground in using digital technology to accomplish their goals, could have presided over this kind of screw-up. The answer is nearly as complicated as the website itself, but as the administration has said, the problems are not insurmountable and the site will be fixed (hopefully sooner rather than later).
The next important question is what we can learn from this episode. There are vital lessons to be absorbed about how our government functions—not the Obama administration in particular per se. Instead, we got a good peek at what happens when private companies adept at squeezing billions from the taxpayers are hired to build something big.
There's plenty of blame to spread around, from the White House to the Department of Health and Human Services to the contractors who built the site. But the whole mess can be reduced to this: The administration created Healthcare.gov in the same way the government does lots of similarly large projects, from revamping a department's computer system to building a fighter-bomber plane. And that's why it failed.
Few details about this whole mess are more revealing than the fact that while the principal contractor for Healthcare.gov is a Canadian company called CGI, there are no fewer than 55 separate contractors responsible for parts of the project. How could anyone imagine that the work of all those different companies could be seamlessly integrated, particularly when there was a specific (and not so far away) date when the whole thing had to come together? That's where the bombers come in.
In government contracting on large projects, the use of dozens or even hundreds of different contractors for a single project not only isn't unusual, it's standard practice. The wealth gets spread around to lots of companies and lots of congressional districts, ensuring that as many members of Congress as possible have a stake in continued funding. To use a relevant tech-industry expression, the complexity isn't a bug, it's a feature.
In defense contracting, cost overruns and missed deadlines are so common that if a large project was delivered on time and on budget, the officials who work on procurement would probably faint from shock. Deadlines pass, budgets are blown through, and friendly appropriations committee members keep the money flowing. Things get done eventually, but there are few consequences for not bringing the project in on time. When your bomber goes over budget and has glitches, there's an article about it in Jane's Defence Weekly, but it isn't on the front page of every newspaper in the country. And it isn't as though we have a war starting next October 1st, and if that bomber isn't ready we won't be able to fight.
Take the B-2 stealth bomber. When it was built in the 1990s, subcontracts on the project were distributed to a remarkable 46 states and 383 of the 435 congressional districts. The cost of the program went so far beyond the original estimates that the Air Force ended up buying only a fifth of the planes they had originally ordered; each B-2 ended up carrying a $3 billion price tag, with a cost to fly it estimated at an incredible $135,000 per hour. Building the military's next-generation aircraft, the F-35 Joint Strike Fighter, requires 1,300 different suppliers spread across 45 states. It has been bedeviled by more bugs than you can count, which helps account for the fact that the plane won't actually be in full production until 2019, seven years after the original target date. When it began, the F-35 program was supposed to cost $233 billion; the total is now over $400 billion, and by many estimates its eventual cost will top $1 trillion. Oh, and it can't fly at night or in the rain. Seriously.
And it isn't just weapons. Large government IT projects have an astounding record of failure. According to one analysis, in the last decade there have been 3,555 government IT projects that cost over $10 million. Over half had major problems, either going over budget, being behind schedule, or not performing the way the users intended. Another 41 percent were such failures that they were either scrapped entirely or had to start over from scratch. That left a mere 6.4 percent of projects that came in on time and on budget and worked the way they were supposed to.
No IT failure was more notorious than the one that occurred when the FBI decided to revamp the archaic system agents used to access and enter case information (on September 11, agents didn't even have the ability to pull up photos on their computers, if they even had computers). After five years and $170 million dollars, the system, created by the outside IT firm, SAIC, was such a disaster that the FBI scrapped it and started over. Another $451 million and six more years went down the system’s gullet before it was put in place in 2012—but only after they took the project over from the second contractor and brought it in-house. SAIC, by the way, is doing just fine; in 2012 they were the eighth-largest federal contractor, bringing in $7.4 billion in taxpayer money.
Companies like SAIC and CGI don't get billions of dollars in contracts because they employ the most brilliant and creative programmers in the world. They get those contracts because they're experts in working the federal contracting system. In fact, the Healthcare.gov contract was awarded to a CGI subsidiary called CGI Federal, so named because its reason for existing is to win federal contracts. If you're an NPR listener, you've probably heard, "This program is made possible by General Dynamics IT." You may have thought, "A weapons company does IT? That's weird." But if there's anything weapons manufacturers know, it's how to get a hold of taxpayer money, and there's big money there. General Dynamics' IT division took in $10 billion last year, mostly from the government. Lockheed Martin, the largest government contractor ($37 billion in federal contracts in 2012) has an IT unit as well; its 2012 revenue was $8.8 billion. They were the ones who did the second phase of the FBI's tech revamp and had to be kicked off the project (and guess who's the lead contractor on the F-35).
Hindsight is 20/20, of course. But given the way federal contracting works—and the abysmal record of government IT projects—if the Obama administration had the proper foresight, it would have realized that the ordinary procurement process would probably doom this project, at least in its initial stages. So what could they have done differently? The smartest thing might have been to recruit bidders from outside the world of government contracting, with a public push that highlighted the absurdity of the existing contracting system, made clear the technical hurdles Healthcare.gov would have to overcome, and challenged companies to come forward and see if they could achieve what the big government contractors probably couldn't. There are surely hundreds of technology companies that could have completed every aspect of this project—companies with the technical expertise to handle the job, but whose business models aren't predicated on making their projects as complex and costly as possible. There would have been hoops to jump through to make the process compatible with existing procurement rules, but it could have been done.
Before Republicans start feeling too vindicated by the failures of Healthcare.gov, they should remember that they were the biggest advocates of outsourcing government tasks to private contractors like CGI. They told us that private companies would always do things cheaper and more effectively than government could. Government contracting exploded during the Bush years; the total value of federal contracts was just over $200 billion in 2000, but increased to well over $500 billion by the end of Bush's term (it has declined slightly during the Obama administration, but remains over half a trillion dollars a year).
This isn't just a government problem. It isn't just a private-sector problem, either. It's a result of what happens when the two come together and there are ungodly amounts of money to be made. But we should never forget that large systems—and all their pathologies—are the product of individual decisions made by human beings. There's nothing inevitable or eternal about them. They can be changed and improved. It isn't easy, but we just have to try.