Charles Worthington
‘That’s What Our Superpower Can and Should Be’: Charles Worthington on the U.S. Digital Service Working Side-by-Side with Federal Agencies
| Charles Worthington joined the U.S. Digital Service in October 2014, serving for almost three years and holding the titles of Digital Services Expert, Director of Product, and Acting Deputy Administrator. Prior, Charles was a Senior Advisor in the White House, a Presidential Innovation Fellow, and a member of the Federal Communications Commission’s National Broadband Plan team. |
Charles Worthington formally joined the U.S. Digital Service two months after it launched, but had a hand in shaping the organization for many months prior — first as a Presidential Innovation Fellow, and then as a Senior Advisor to Todd Park in the White House Office of Science and Technology Policy in the months leading up to the USDS launch.
Charles wore a number of hats at USDS, from engineering and hiring to leadership. He also played a key role in creating the Digital Services Playbook, which served as a blueprint for USDS and several other civic tech organizations.
Below, Charles discusses digital transformation tactics, how to create momentum for change within government.
August 15, 2019
Emily Tavoulareas:
Charles, tell us how you ended up at the U.S. Digital Service.
Charles Worthington:
I was a Presidential Innovation Fellow (PIF) in the second-ever fellowship class. I joined in July 2013 and was assigned to work on open data implementation at the Department of Energy. The fellowship itself wasn’t very impactful: I was buried in a low-level part of the department. But over the course of the fellowship, the PIFs worked with each other. And so I got to work with Mollie Ruskin, Ben Willman, and other folks at the Department of Veterans Affairs (VA). I also got to know Haley Van Dyck as a part of the open data work.
Jen Pahlka started at the White House around the same time, as Deputy Chief Technology Officer (CTO). And December of that year — after the HealthCare.gov crisis — Jen shared her U.S. Digital Service proposal. At the time it was called the “GovX” team, and it was going to be in the General Services Administration (GSA). She and CTO Todd Park were working on a memo, and she invited a few people to brainstorm. That was the moment I first joined the USDS bandwagon. The meeting was a little chaotic. Toward the end, I asked Jen if I could have the laptop and started wordsmithing the memo.
Jen and I continued to talk, and a few months later she said she could use more help in the CTO’s office. So I ended my fellowship a few months early and switched my detail to the White House.
Emily:
You were one of the few people who was present before the concept for USDS had crystalized. What were some of the earlier versions of USDS, and how did they evolve?
Charles:
By the time I was involved, we had coalesced around the idea that this would be a part of the White House. We had the UK’s Government Digital Service (GDS) in our heads as the thing we were trying to emulate. We started experimenting with “discovery sprints,” which are time-boxed efforts to assess and help fix a specific issue an agency was having. So we had a model. There was the HealthCare.gov experience, but also these diagnostic assessments that we did with a few agencies, like the VA and U.S. Citizenship and Immigration Services (USCIS).
Emily:
Do you recall when USDS went from “This is a cool thing we should do” to “Ok, we’re doing this”?
Charles:
It was around winter 2013, after the HealthCare.gov rescue. There was a memo describing the idea that Jen and Todd were developing, and my sense was that approval happened sometime the following spring. By the time I had joined, we were focusing on what this new office would focus on at first. We intended to pick three to five important things that the government was doing which relied on a successful use of technology, and try to increase the odds of their success.
Kathy:
What was the first discovery sprint?
Charles:
It was an assessment that some USDS folks did at the VA on the disability claims backlog, which was a huge priority for the White House and for the VA. As I understand it, that sprint landed really poorly. It was about the Veterans Benefits Management System (VBMS), which is the big claims processing system. VBMS was new at that point and had cost a lot of money, and the report was critical of VBMS. Today, we’re still battling the reputation that report created for USDS. The report was very critical of the project, but there was not an obvious way that it would help improve the situation and the VA resisted it. Then we did two discovery sprints at USCIS.
Emily:
In those early discovery sprints, when you were coming over as a detail from the White House, what was the reception like at the agencies? And did that change over time?
Charles:
I was most involved with the sprints at USCIS, where we were viewed with skepticism. Anytime the White House gets involved, people are wary. I think in many people’s experience, engagements with the White House are rarely helpful to their mission, but they always cause more work. Ultimately, though, I do think agency staff who were advocating for more modern approaches found value in our presence, as a way to get more traction for the things they had been attempting to push from within.
Emily:
One of the existential questions about USDS was where the teams would be and where they got their power and their air cover. What do you think?
Charles:
Well, a huge White House priority at the time of the DHS sprint was immigration reform. The immigration reform bill was making its way through the Senate and there was this big (correct) concern that if the bill passed, the underlying systems at USCIS wouldn’t be able to handle the implications. The fact that it was a White House priority and the Domestic Policy Council was very interested in immigration reform helped us a lot. People took us very seriously.
Then the E-Verify sprint took place in April 2014. E-Verify is the system that employers use to check someone’s immigration status, and the immigration bill was going to make the use of E-Verify mandatory. The question was: Could E-Verify scale up to be used by every employer in the country?
Kathy:
What happened?
Charles:
The immigration bill passed the Senate, but the House never took it up. And so E-Verify never became mandatory.
Kathy:
How did all these events lead to USDS’s launch in August 2014?
Charles:
Around that time, Todd was working to recruit Mikey Dickerson as USDS Administrator. Leading up to that, we wanted to hire between 10 and 20 people, so we planned a recruiting event in Washington DC to meet possible candidates for USDS, culminating in a roundtable event where POTUS made the pitch for public service personally, in a roundtable in the Roosevelt Room of the White House.
That was a huge event to plan. Vivian, Erie, Haley, Jen, and I came up with a list of possible people and called them to talk about USDS. We were leaning heavily into the HealthCare.gov story. Some of the people we invited had already been read into the idea; they had come as intermittent consultants on earlier discovery sprints. But some of the people we cold called, like Dana Chisnell.
One of the big pre-launch activities was closing Mikey and figuring out his start date. A second was the big wave of hires. And a third thing was the Digital Service Playbook, because I was worried about announcing USDS and making a big press splash without us having done anything yet (even though we had a few discovery sprints under our belt by that time).
I worked on the playbook a lot. The consulting company Thoughtbot has something called the “Thoughtbot Playbook,” which is a guide and principles they follow when they’re building products. I had that in my head. The GDS also had a playbook. So we had this idea of, “When we launch USDS, let’s also launch it with this manifesto on how to do digital services better.” The content was based on what the team saw at Healthcare.gov and other agency discovery sprints, and also based on their professional experiences building high quality digital services outside of government.
Ryan Panchadsaram and I were the primary authors. Mollie added a bunch of great thinking about user experience and research. And then the whole cast of people already there added things: Erie, Haley, and Nick. Jen was the one that got it over the finish line. We were working on the playbook over the whole summer, and we had to get it through White House clearance, so every agency had a chance to comment on it.
Kathy:
Oh, wow. What was that like?
Charles:
Honestly, I don’t think many of the right people looked at it or paid attention. For example, I don’t recall the OMB technology policy folks having much input into it. So it wasn’t that bad or good. We also got clear direction from Todd Park that we didn’t have to accept all edits. That was one of our first lessons: If you write the policy, then you get to decide whose edits to accept and whose to reject. Without the top-level air cover from Todd and the President, we never would’ve been able to write and publish the playbook the way we wanted to.
Kathy:
When you look into things, you realize that sometimes these really big decisions are just a meeting with a few people.
Charles:
That’s definitely what it was like at the VA. People would spend months on briefings, making them perfect. But at the end of the day, it’s just the deputy secretary turning to the undersecretary and asking, “What do you think?”
However, the President was really good about reading all the paper that got written for him. If you were writing a presidential memo, he was going to read it and write notes in the margin. That was pretty cool.
Kathy:
The playbook played such a pivotal role in centering our work. Can you talk more about pulling that together, its evolution through the years, and how people used it?
Charles:
It started as a Word doc that we passed around a lot. In the first version, Mollie identified that there was no concept of the user, so she is responsible for those first two or three plays like, “Understand what the problem is. Do lots of research, make sure that it’s working for people.” The first version was much more about the technical stuff, like “How do you do deployments? Where do you host it? What technology stack should you use?” It would have been a huge mistake to publish it without the user experience and user research content, so Mollie deserves a bunch of credit for identifying that omission and fixing it.
We also spent a lot of time figuring out how to present it online. Mollie did the design and Ryan helped a lot. And then I executed it in CSS and JavaScript, with a simple Jekyll build pipeline generating the content from Markdown files. Even to this day, I think it’s hosted by GitHub pages.
Kathy:
Once it was published, how did it influence the foundation of USDS?
Charles:
Well, people in government are constantly citing these arbitrary policies as the reasons why you have to do certain things. Our idea was, “Why don’t we put out a good policy, and then we can cite that?” For example, if some agency is opposed to using the cloud, we could say, “The White House Playbook says that we should default to using the cloud.” Our hope was that the playbook would be helpful for us, but also that other people could use it to advocate for the right thing to do in their corner of the government. And we did see that. One time, the SBA self-assessed one of their systems on each of the plays. And the City of Austin forked it; a number of people have forked it. But the underlying code is the same Jekyll markdown we wrote.
When we first launched USDS, most of the commentary on Twitter was about the playbook. A lot of people were saying, “Oh, shit. This is actually good. Wow, the government actually gets this.” That was really validating, because so much of Twitter is about dunking on the government.
Across the tech industry, there are these documents that people revere. Things like the “twelve factor app” or the “4 DORA metrics.” They’re not policy, but they have a similar impact. They’re principles people believe should be followed without good reason not to.
The U.S. Web Design Standards are another little-heralded thing that have quietly had an enormous impact. If a new website gets built in government now, it generally follows that standard. At a minimum, that means that the site has good typography and is relatively easy to understand and navigate, which is totally different than it used to be. Mollie deserves a lot of credit for that.
Kathy:
There’s a theme here, this idea of “the strategy is delivery.” By giving people something to look at, we’re also empowering them to build products and driving culture change.
Charles:
Whether discovery sprints go well or poorly is linked to that lesson. If a discovery sprint is technically accurate and correct, but doesn’t create a clear path forward to actually making a change to the system, then that’s a failure. That first discovery sprint we did at the VA had that problem. It was a bunch of smart people identifying clear problems with a system, writing them down, but USDS was just an idea at that point, and the sprint was staffed by temporary consultants. So there wasn’t a clear theory for how we could actually help VA implement the changes the report identified. Without a clear plan to help with implementation, a discovery sprint can just feel like another form of oversight like what might happen from OIG or GAO.
I think we can all agree that the solution to government tech problems is not more or better-written oversight reports.
Kathy:
Do you think there’s a way to write a very honest report while also creating momentum for change?
Charles:
I do. But I think it is less about what’s written on the page and more about how you conduct yourself in the engagement, the relationships you build as you write the report, and how you make people feel like they are a part of the findings. And what is the plan for the next step?
If the report feels like an external team of smart people auditing an agency, then you’re just like every other oversight group. The way to do it is to find the people in the agency who also want the thing to improve, and make them be the hero of the report. Let them deliver the report, even. That’s one tactic that worked well at USCIS.
Honestly, I don’t know that having a technically true report is actually that valuable. Our fundamental focus is improving services and making people’s lives better. A report is useless unless it makes progress toward that goal. Any technical expert can easily identify the ways in which the government is bad at technology. What’s hard is fixing those problems. And that’s what our superpower can be and should be.
You can apply that thinking to everything. It’s easy to say, “The hiring process is so slow, and the wrong people get hired.” What’s much harder is to create a way to hire in government that does work. USDS has been effective because we learned to roll up our sleeves and fix problems.
Kathy:
Then there are other folks who say, “You need someone to just go and blow things up.” And I actually think there’s a time for that, but in moderation.
Charles:
You need both types of people on the team.
Kathy:
Charles, can you tell us about the early USDS roadshow visiting each of the federal agencies?
Charles:
After we had that first concept of three to five teams and 20 people, there was a debate about what we should do next. We launched the hiring website around that time, and we got a huge influx of applications. People’s imagination of what USDS could be was growing. So we wanted to talk to all the agencies and identify the most important potential projects to work on. We also had this concept of eventually having USDS teams at each agency.
Kathy:
Who was involved in those agency discussions, and how exactly did the idea come about?
Charles:
Mikey and Haley were invited to talk about USDS at the President’s Management Committee meeting with all of the deputy secretaries of the agencies. In that meeting, all the agencies were told to meet with USDS about their problems. I had this idea that we could tier them, and then have a team working on each of the top problems. We would basically use those tiers to figure out where to put resources as we staffed up.
At one point, I was pushing this idea of “five by five” — we should have five people at five agencies, and that’s how we would deploy our first 25 people. We didn’t actually execute that; I think it makes more sense to go deeper in fewer places. Also, dividing by agency doesn’t wind up working. It’s not that useful to have five people at the Department of Justice (DOJ). Instead, you want five people at the Bureau of Prisons, five people at the FBI, and so on. [These were entities within the DOJ.] Same with Health and Human Services: You want those people at the Centers for Medicare & Medicaid Services, not at HHS central.
Kathy:
What was your reception when you talked to all the agencies?
Charles:
It varied quite a bit. Some agencies were super-excited. The Small Business Administration is a great example. The reason we have a team at SBA is because the equivalent of their deputy secretary at the time said, “We need this immediately. Please come now.”
No one was really hostile. Because at that point, we were just asking them about their top projects and explaining the USDS approach. We went through the playbook and we talked a lot about the HealthCare.gov experience. Some were very open and talked a lot about their problems. Others were much more guarded. It was clear who was not interested, which was helpful to us: We got a sense of where we should or should not put people.
At some point in 2014, I shifted to helping build the Talent Team, because as soon as our application page went live we had 1000+ applications and no clear process for sorting through them and making our next 200 hires. Our various theories of how USDS would make an impact wouldn’t amount to much if we didn’t have people to do the work! I wound up working to scale up our talent operations for almost a year before switching back to more delivery-focused roles.
Kathy:
You played so many different roles at USDS.
Charles:
Everyone did everything in the early days.
Kathy:
Thank you for your time, Charles!