Stock Options? Don’t Need ’Em! I’m Coding For Uncle Sam!
The people behind the new government agency that’s recruiting the nation’s best tech talent to reform its hideous computer systems.
--
There was a time when Mikey Dickerson felt that one could live a satisfying life as a Google site reliability engineer. That was before the fall of 2013, when he got a call to help out with the rescue of Healthcare.gov. Dickerson joined a small team embedded in the bureaucratic maw that produced the dysfunctional website, and became the de facto squad leader, rebooting the site by introducing best Silicon Valley practices into the moribund government IT effort to implement the Affordable Care Act. In one now-legendary moment at a daily standup, Dickerson stood down some traditional types who were resisting the adoption of a tool used widely in developing modern websites. “If I hear one more person tell me I can’t use New Relic,” he said. “I’ll punch them in the face.”
After his temporary stint was over, Dickerson, 36, missed the satisfaction of knowing that his efforts could be life-changing for millions of people. So he accepted the job offer to become the administrator of the United States Digital Service. The brand new organization sits in the Office of Management and Budget, reporting to US Chief Information Officer Tony Scott. (Dickerson also holds the title of deputy CIO.) It works closely with the White House Office of Science and Technology Policy, particularly the tech wing led by US CTO Megan Smith and her deputy Ryan Panchadsaram. (For our interview with Smith and her deputy Alex Macgillivray, see this.) The idea, hatched by Smith’s predecessor Todd Park (who still works as advisor to the president from Silicon Valley) is to replicate the success of the Healthcare.gov rescue and bring the best engineering and design talent from Silicon Valley and elsewhere into the government. Not as bureaucratic lifers but volunteers, working for intense bursts. And if these people become hooked on the impact they can make by using their skills to untangle big intractable messes — maybe so much that they decide to work for longer stints — so much the better.
The Healthcare.gov experience shows how great the differences can be. In newly released figures, the government says that constructing the original enrollment system, known as the Federally Facilitated Marketplace operating system, cost $200 million and would have required $70 million a year to maintain. The new version of the site, revamped by USDS engineers from Google, Y Combinator startups and other commercial tech outposts, cost $4 million to produce, with annual maintenance costs also $4 million.
Assisting Dickerson is his deputy administrator and USDS co-founder, Haley Van Dyck, who has been a key player for the administration’s tech strategies since the 2008 transition, working at the White House, the FCC and USAID. She was also responsible for the first time a government policy effort was developed using the collaborative GitHub tool.
Earlier this month, as their tenure approached the milestone of its first year, I spoke to Dickerson and Van Dyck at USDS headquarters. It’s a stone’s throw from the White House, but situated in a structure that’s more of a townhouse; we sat in the living room, adorned in funky furniture and, of course, whiteboards. Dickerson was dressed in Google casual; Van Dyck kicked her shoes off. The interview is edited for brevity and clarity.
[STEVEN LEVY]How is the recruiting going?
[Mikey Dickerson]: I was worried about getting enough people to come here and work, and also getting the agencies to accept outside help, because both of those things you can’t take for granted. We have not struggled with either of those things at all. I thought we’d be lucky to get a few dozen people interested, which would have been okay because I figured we could probably pull that off with a few dozen. But the reception has been a lot more positive than I wanted to assume. After the spending bill passed in December, for the year 2015 it was expected to be 50 or 60 or so and we’re close to that size now. And we have at any given moment a number of applications waiting to be processed in the low thousands.
Why did it exceed your expectations?
MD: The sales pitch worked better than I had expected.
[Haley Van Dyck]: Our sales pitch has literally one component to it — come serve your country and have an impact at scale. We have had so much success in recruiting because people are interested in using their skill sets to actually help others. That’s really all we’ve got going for us at the moment.
MD: To the extent that we have refined our recruiting message, each version has been more aggressive than the last in telling people: Don’t come here if you’re looking for a fun, easy thing to do, if you think you’re going to make a lot of money, if you think it’s going to advance your interest in some particular way.
Only come here if you want the opportunity to work on the most important problems in the country. I will guarantee you if you come here for that, you’re going to get it. If you come here for anything else, I promise you nothing.
We started getting more aggressive about that because we are inundated with applications. Amazingly that didn’t slow down the flow of applications at all. The more I make it sound difficult, the more it appeals to people who want that challenge. Amazingly, we stand in front of recruiting events and tell them this — and we get more resumes than we can handle. When I recruited for Google, we had money, we had free buses, we had free food, we have all this stuff that we don’t have here — and it was way harder to get anybody interested.
It’s interesting that people are jumping when the history of government IT has been so dismal.
HVD: If it was just, “come for the mission and run into a bureaucratic wall over and over again,” it wouldn’t work. This is also working because we have a strategy that we’ve developed over years of figuring out what has worked and what hasn’t to make sure that there is an opportunity for impact. The work of the engineers and designers that we’re pulling in from across the country will actually have a chance to influence and change the way the government is operated.
How about your second concern, getting the various bureaucracies here to accept your approach?
MD: That’s not even close to our top handful of things we’re worried about at this point. We had a plan built around the 2016 budget proposal, which was used as an excuse to meet with all of the cabinet agencies. There’s a lot of them — these are big, ceremonial meetings — and it takes a long time to organize them. We walked away from that exercise with, I don’t know, conservatively 60 to 80 projects that we felt were meaty.
For 2015, you were budgeted at what?
HVD: Twenty million.
And you asked for how much in 2016?
HVD: The total request was a hundred and five. Because this is such a large industry we’re trying to disrupt, we need a bunch of pressure points of how to change the system. So that $105 million includes the central funding request for the HQ, the headquarters team, as well as discrete funding in each of the agencies for their own teams as well.
Does that require separate Congressional approval, or is it basically just part of the big budget?
HVD: It depends on each of the committees. It’s very much a shrapnel approach. Some of these line items [won’t] get funded. Some of them will. Either way we believe that lack of money is usually not the problem that we face in federal IT. This was a way to develop a strategy that allowed us to have the necessary mandate for these teams inside the agencies at a very senior leadership level.
Have you managed to get the point across that not spending these modest sums will doom them to bigger expenditures, for stuff that probably won’t work?
MD: We’re for sure trying. That’s basically our pitch for why Congress should fund us — it’s so tiny. We had a formula by which we calculated those requests and it works out to about one fifth of one percent of what the agency would have spent anyway.
HVD: Which is to say we see this as a very prudent insurance policy.
MD: We’re very, very small. For the tiny amounts of money we’re talking about here, if this is sufficient to get even one of the agencies’ smaller IT programs on track, and not running over budget and over schedule and everything, it will pay for itself ten times over.
And we’re not taking on the small things at the agencies — we’re taking on the biggest ones in most cases. Immigration is a great example. We’ve had a handful of people for several months in the US Citizenship Immigration Service, which is in the Department of Homeland Security. They have helped to make what’s called the I-90 process — how you get a replacement green card — into what is now an online electronic process. And now we have fans in Homeland Security. It’s hard for people to understand, how could five people make a difference? We point to [the rescue of] Healthcare.gov and say, “Well that was five people, and we need to replicate that model,” and they sort of buy that. But it helps a lot when somebody who is their peer at another agency says, I wasn’t so sure about this whole crazy scheme either but I decided to give it a shot and six months later they’ve made a tremendous impact.
Do you have a metric that shows the difference between that immigration program before and after your small team revamped it?
HVD: Well, the metric before was that the integration system was entirely paper-based. To actually apply to the system it cost about $400 per application, it took end user fees, it took about six months, and by the end, your paper application had traveled the globe no less than six times. Literally traveled the globe as we mailed the physical papers from processing center to processing center. This transformation process existed well before we showed up [but wasn’t succeeding]. There was actually a billion dollar contract that was out at one point to start this modernization process. At the end of the five-year contract, which included [an additional] two-year requirement-gathering phase, zero code was delivered that worked. They were in the second year of another five-year long contract when we showed up. And I-90 is the first functioning release that has existed on this project in almost seven years. [After the interview, a government spokesperson clarified that the first contract was a seven-year deal for $1.2 billion, and the result was “behind schedule and slower than paper.”]
What did I-90 cost when you folks did it?
HVD: The salaries of five people.
MD: They worked with an existing organization, but what we added to the project was five people. Not even measurable against a $1 billion contract.
Is it fair to say that USDS is the organization that Healthcare.gov founded?
HVD: I think it’s fair to say that we would not be here without Healthcare.gov. That was an incredible wind in the sails. It accomplished three really important things for the entire movement. First and foremost is that it showed us that status quo was no longer an option. There was no choice to do things the same way. Healthcare.gov was an unfortunate crisis I wish never happened, but it wasn’t unique. It was very representative of a much broader problem across government, and we needed to have that deep understanding across all layers of government from senior leadership down to the engineering room floor to really actually get the message to start doing things differently.
MD: It put a lot of issues into focus. All you have to do is go read the comments on any news article from October 2013 and you will find hundreds of feds and contractors say it’s been this way for decades, all these projects are a giant failure, blah, blah, blah. This emperor has kind of had no clothes for a long time. [But] on the scale of things competing for the President’s attention, better delivery of information technology and agency programs wasn’t in the top handful — until there was no choice but for it to be in the top handful. That’s a painful way to learn. but it beats just dragging the lesson out across another ten or 15 years, which maybe would have happened otherwise.
You are starting this United States Digital Service from scratch. I’m getting a very informal vibe here. Are you consciously trying to set a different culture, more like a tech one, than you see in other agencies?
HVD: We’re very proud of the culture that we’ve developed here.
How would you describe it?
MD: There’s a lot of things that are superficially different. So yes we have white boards, we have DJ equipment kicking around, we have a lot that really isn’t material to how people do their jobs. It’s just signaling value, so that when we bring people in from the outside who have hung around at Google and Facebook and Microsoft offices, they kind of feel, “Oh these people obviously have some of the same DNA because their office is laid out the same.” That’s for appearances’ sake. But the active ingredient I think is we are relentless about trying to hang onto the ruthless mission focus here. We are built for short term appointments. We’re not building a permanent piece of bureaucracy.
If the US Digital Service still exists in 50 years from now hopefully it’ll be working on very different things in a very different way in whatever way is relevant 50 years from now. And if we can’t transform and adapt as the needs change, I would rather we just not exist than become a permanent fixture that exists only to perpetuate itself, which I feel like has happened with huge amounts of well-intentioned efforts that have come before us.
So we don’t make career hires. We’re not building a career organization. A key part of that sales pitch that I told you is super aggressive about trying to turn people away, a key part of that sales pitch, it wouldn’t work if it weren’t for us saying we’re only asking for you for a year or two years maybe, we’re not asking you to make a permanent career change because the kind of intensity of the projects we’re gonna give you to work on and how much we’re gonna ask from you really isn’t sustainable for that kind of period. And I don’t have the incentives to make it a rewarding career for that kind of period.
You are coming here because you want to do service and we will absolutely give you the opportunity to do that service. And it actually helps us maintain that culture here and the group here that every single person we hire we have looked them in the eye and said you understand you’re coming here because you want to serve veterans or serve immigrants or serve students who have loans or whatever your particular constituency turns out to be, that’s what you have to get up in the morning and think about.
And the fact that we’re built on this tour of duty, Peace Corps-type model instead of a permanent piece of bureaucracy I think makes it a lot easier to hold onto that focus.
Do you feel that there is a clock ticking for you to get under the skin of the government and weave your way in there before this particular administration reaches its end?
HVD: I think we definitely feel the sense of urgency to deliver results in the next 18 months.
MD: Eighty weeks. What are we, like 560 days at this point?
HVD: Yep.
Who’s counting?
MD: Oh, we are keeping track.
HVD: We definitely feel a sense of urgency. Our institutional innovation strategy is, if we can prove our value over the next 18 months, we believe it will be asinine for the next administration to not continue to invest in this resource.
MD: I would like us to not need policies or executive orders or things carved into stone tablets in order to force us to continue existing in the next administration. I would rather the next administration see the value for themselves and choose of their own volition to continue doing more or less what we’re doing right now. If we fail to demonstrate that value to them then as far as I’m concerned maybe we shouldn’t exist.
Mikey, I know that after you worked so hard on Heathcare.gov you thought you were done with government service — but when you got back to Google you found the work empty, and that was in part why you returned.
MD: Yup. I went back to my old job and was talking to engineers about the types of things engineers talk about and it was exciting to talk about machine learning again, but at the end of the lunch I was like wait a minute, that doesn’t actually matter. Like, whether we solve that problem or not, nobody cares. Like that will never affect anybody in any way that makes any difference.
Do you feel in general that the people that come here have that same feeling that there is this emptiness in Silicon Valley compared to the stuff you’re doing?
MD: We had an engineer named Will Chan working on immigration for a few months and returned [to Silicon Valley]. He actually ended up not going back to Google, which was what prompted him to write the blog post. It was a very, very similar message — this is Will’s words — that after knowing that the skills that I have can be transformative for some people that really need it, I now can’t live with myself letting it go to waste, working on something that doesn’t matter. We have had several people that experienced this, and we’ve had a surprisingly high rate of people that come here for a short term that decide to extend that term.
HVD: And that’s quite honestly one of the bait and switches that we have. People come here for short-term tours of duty, some just get a three-month sabbatical. And we saw about 66 percent of the people that came out for three months ended up going home, quitting their job and coming back full time. And it’s on the rise. I think it’s over 80 percent now.
Some people have said they are concerned that Google is over-represented in the government digital push. Is that right? Does that even matter?
MD: Google is a 60,000-some person company. And a decent number of the people that we’ve hired directly from other companies, like Facebook, left Google [to go to those places]. So Google is just on the resume of a large number of the best people right now. I’m not sure it’s necessarily even disproportionate. We haven’t studied this in any great detail.
HVD: And we have a huge amount of representation from other incredible tech companies, whether it’s Twitter, Facebook, Amazon. We’re hitting a lot of the big Silicon Valley giants.
I understand one of your biggest initiatives is embedded in the Veterans Administration.
HVD: The VA is one of the first agencies where we put boots on the ground, and we have been building out the team there, which today is about 11 people. The VA is interesting because it very much epitomizes a lot of the challenges of working for the federal government. It’s a very big bureaucratic organization so it’s been a challenging one to get a foothold in. The largest project that we’re working on there is focusing on the operability of health records for service members as they transfer from the DOD to the VA.
I understand that someone’s Department of Defense medical file is in a totally different format than the one used by the VA.
HVD: Exactly. And we’re finding that the United States Digital Service model is most effective in those services that transition between agencies because that’s where the coordination gets so challenging and so difficult. We are applying the most amount of pressure right now evaluating the hand off of service records from DOD to VA, which again used to be an entirely paper based process until a couple of years ago. It’s incredible — the actual files for service members are usually over a foot and a half long and would get driven from the DOD over to the VA headquarters building. And there was so much paper that the VA actually had to reinforce the floors of the building to deal with all the paper weight. So that process became digital in some capacity a couple of years ago —
MD: It got digitized as far as taking pictures of the paper in PDF form and sending the PDFs over to the VA. So that’s progress but it’s not really where we meant for that to go.
HVD: Our team is focusing a lot of their energy right now into figuring out how to make that [better].
Besides the technical challenges, what are the biggest obstacles? Do you encounter foes dedicated to the status quo?
HVD: The biggest foe is generally risk aversion. People in government are trained to not do things differently because there’s often really bad consequences when you try something differently and it fails. We run up against this all the time.
MD: I wish there were bad guys with top hats and handlebar mustaches because if there was some super villain behind a humongously dysfunctional project, all we would have to do is identify that person and take them out and everything would get better. That’s not the problem. The problem is just all of the things that inevitably happen when you try to coordinate 60,000 people in the VA to do the same thing at the same time. Even when somebody looks like they’re being a big pain, it’s just a function of their position in the bureaucracy and their role. Their interest is almost always wanting the same thing that we want, which is that they want the veterans to get a better experience, they want the disability claims to be adjudicated faster, but to them that doesn’t mean the same thing necessarily that it means to the person next to them.
What feedback have you been getting from the President?
MD: The number one thing is there’s a lot more that we need you to do and can you find more people and can you take on more projects.
HVD: He’s basically our business development team — he’s going around telling everyone that they need to be working with us, which is incredibly helpful.
So you’re both committed until the end of the administration in January 2017. What will US Digital Services have done by then?
MD: Off the top of my head, Healthcare.gov will be able to stand on its own without the training wheels, without us helping get through the enrollment period or whatever. But by 2016, we should have momentum in solving actual real practical problems between the DOD and the VA, in a way that is also extensible to solve the more general problem of health data inoperability. That would be a big success. And we have a bunch of other irons in the fire.
HVD: Yeah. By the end of this administration, we [would be] excited to see tangible differences for our key users who are veterans, students, people applying for social security, immigrants. So see tangible improvements for those user groups. But more importantly that we will prove that this model works, that it will create a tradition of public service in the technology industry, which didn’t have one before. So that’s our secondary goal.
MD: The super stretch goal would be if it becomes a point of pride [for a techie to work here] in the same way it is on a lawyer’s resume to have been a clerk at the Supreme Court or a doctor’s resume to have done Doctors Without Borders. If you’re really the best of the best, then at some point you will be asked to come work for the US Digital Service for a couple of years.
Photographs by Stephen Voss for Backchannel, unless otherwise credited.
