When the government’s health insurance shopping site Healthcare.gov first launched on October 1st, it was obvious that something was very wrong. The site rejected valid passwords, served up blank drop-down menus, and crashed repeatedly. After millions of visits on the first day, only six people got all the way through.
Even more remarkable than the scale of the catastrophe was the speed with which the site was fixed. An Oceans 11-style recruitment effort pulled together a team of about a dozen technologists and managers who packed their bags, flew to Washington, DC, and worked for two months straight, including on Thanksgiving. By early December, Consumer Reports had reversed its opinion of Healthcare.gov from "stay away" to "it’s terrific." Monday, the official deadline to sign up for health insurance, was one of Healthcare.gov’s biggest traffic days. It stumbled, but quickly recovered and signed up 217,176 people in 24 hours.
The fact that Healthcare.gov now works shows that government can do tech right — it just doesn’t, most of the time. Healthcare.gov is the latest in a long tradition of botched government tech projects: just look at the Air Force’s $1 billion Expeditionary Combat Support System that never materialized, the FBI’s $170 million Virtual Case File system that had to be scrapped and restarted, or the $6 million reboot of USAjobs.gov that cost an additional $1 million to fix.
The fact that Healthcare.gov now works shows that government can do tech right — it just doesn’t, most of the time
Now, six months after Healthcare.gov’s debut gave us a terrifying glimpse inside the government tech sausage factory, there’s some evidence that the government is trying to head off future IT disasters by calling in the crack team first.
After the Healthcare.gov crisis, many people said the government should act more like a startup: keep the team lean, iterate quickly, and test things along the way. By contrast, most large government tech projects start with a bureaucratic contractor selection process called "procurement" that funnels jobs to staid Beltway insiders and locks out small, innovative companies. In addition, the technical requirements for the project are subject to the vagaries of politics, and responsibility is so fragmented between agencies that there’s bound to be reciprocal finger-pointing as soon as anything goes wrong. If something does go wrong, the government shovels money at the same people who screwed up until they fix it or it gets abandoned. Facebook’s internal slogan is "move fast and break things." The way the government builds technology is pretty much the opposite, except for the "break things" part.
There was the sense that government needed to upgrade to more modern software development methods long before the Healthcare.gov saga. One proposal was to experiment with a digital-services agency within the government that could operate with the speed and agility of a startup. The White House has actually had something like this since 2012. The Presidential Innovation Fellows (PIF) program draws talented coders and entrepreneurs from the private sector to hack on federal projects for six months to a year, building applications that could save the government millions of dollars. Unfortunately, the office is isolated and there is limited infrastructure to support the projects after their creators graduate.
Move slow and break things
In October, General Services Administration (GSA) administrator Dan M. Tangherlini recruited 11 PIF graduates to start 18F, a 15-person digital services agency within the GSA that just launched last month. The rhetoric is lofty: the agency plans to prove that rapid prototyping, live testing, and other elements of the speedy "lean startup" methodology will work for the government, delivering better products to taxpayers and saving money.
"This is about building real IT services," says Greg Godbout, one of 18F’s cofounders. "It wouldn't be fair of me to say that we would build Healthcare.gov in the future, but it might be done differently."
18F is still oblique about exactly what it will be doing, but Godbout says it is already booked to build projects for four government agencies that will be announced this summer. Eventually, 18F also wants to encourage agencies to use open-source software and work with contractors outside the Beltway. (It was originally called GovX, but that name turned out to be trademarked by a discount shopping site; it was changed to USDS, which turned out to be trademarked by the US Postal Service. The current name comes from the GSA’s address.)
18F is just one of many innovation groups that exist within the government, one Presidential Innovation Fellow tells The Verge. "Every agency, the GSA most specifically, has a desire to own their own program, establish control, and dictate the terms of ‘innovation’ within government," he says.
A surplus of innovation groups may not be a bad problem to have, he says, but it’s unclear whether it will meaningfully address the reasons Healthcare.gov came out broken. "The underlying issues within the federal government are far from fixed," he says.
Sources with knowledge of the administration’s tech reform effort say that 18F may soon come under the umbrella of a high-level office within the White House, possibly headed by Jen Pahlka, the deputy CTO of the Office of Science and Technology Policy and founder of the civic programming effort Code for America. The White House declined to comment.
"The underlying issues within the federal government are far from fixed."
18F reflects a desire within the White House to make government technology "suck less," to use the parlance of startups. That desire exists in Congress too, where the House just passed the Federal Information Technology Acquisition Reform Act (FITARA), which would address overspending on large projects, as well as reduce the number of CIOs and give each more authority.
FITARA gained momentum after the Healthcare.gov debacle, says Dave Powner, director of IT issues at the Government Accountability Office, as did a number of smaller similar bills. "I had a testimony last July where we highlighted about 13 failed projects," he recalls. "There were lots of similarities [to Healthcare.gov]. But this thing was so visible, so many people just looking at it, that I think that helped get a lot of attention and focus on what went wrong."
Healthcare.gov and the associated media storm raised awareness of the government’s tech problem which led to some action within the government. Those efforts seem to be fragmented, however, and their potential impact is hard to judge. We don’t seem any closer to preventing large-scale tech failures than we were when Healthcare.gov was created, and it’s questionable whether the interest in federal IT reform will last long enough to prevent the next train wreck. "I feel like the lessons of Healthcare.gov were learned," says Greg Gershman, a former PIF fellow who also worked on the Healthcare.gov tech surge. "Everyone gets the fact and understands that this was not the right way to approach the problem. [But] it’s subject to the normal churn of DC, in that people come in, and there’s something that they want to accomplish, and the question is, are they going to be around long enough to actually make the change."