Joho the Blogopen data Archives - Joho the Blog

February 26, 2015

[liveblog] Data & Technology in Government

I’m at a discussion at the Harvard Kennedy School listening to an awesome panel of Obama administration technologists. Part of the importance of this is that students at the Kennedy School are agitating for a much strong technology component to their education on the grounds that these days policy makers need to be deeply cognizant of the possibilities technology offers, and of the culture of our new technology development environment. Tomorrow there is an afternoon of discussions sponsored by the student-led Technology for Change group. I believe that tonight’s panel is a coincidence, but it is extraordinarily well-timed.

Here are the participants:

  • Aneesh Chopra, the first federal CTO (and a current Shorenstein Center fellow)

  • Todd Park, White House Technology Advisory

  • DJ Patil, the first US Chief Data Science, five days into his tenure

  • Lynn Overmann – Deputy Chief Data Officer, US Dept. of Commerce

  • Nick Sinai – former US Deputy Chief Technology Officer (and a current Shorenstein Center fellow)

NOTE: Live-blogging. Getting things wrong. Missing points. Omitting key information. Introducing artificial choppiness. Over-emphasizing small matters. Paraphrasing badly. Not running a spellpchecker. Mangling other people’s ideas and words. PARTICULARLY LOOSE PARAPHRASING even when within quotes; these are geeks speaking quickly.
You are warned, people.

Todd Park: I’m now deeply involved in recruiting. The fundamental rule: “If you get the best people, you win.” E.g., the US Digital Service: “A network of elite technology development teams.” They want to address problems like improving veteran’s care, helping immigrants, etc. “If you go to the best talent in the country and ask them to serve, they will,” he says, pointing to DJ and Lynn.

DJ Patil: We’re building on the work of giants. I think of this as “mass times velocity.” The velocity is the support of the President who deeply believes in open data and technology. But we need more mass, more people. “The opportunity to have real world impact is massive.” Only a government could assemble such a talented set of people. And when people already in the govt are given the opportunity to act and grow, you get awesome results. Data scientists are force multipliers.

Lynn Overmann: “I’m a serial public servant.” She was a public defender at first. “There is literally no serious problem you’re concerned about that you can’t tackle from within the federal government.” The Commerce Dept. has huge amounts of data and needs help unlocking it. [In a previous session, Lynn explained that Commerce offers almost no public-facing servies except gathering and releasing data.]

Nick Sinai: Todd, you were the brains behind the Presidential Innovation Fellows Program

Todd: The government is not a lean start up but that approach applied to may problems work much better than if you apply traditional the waterfall approach to computing. Round 1 went well. In Round 2, they brought in about 40 people. There was a subset of the Round 2 who found the program “addictive.” So the Whitehouse used 18F, a digital consulting service provided by the GSA. Demand has now gone off the chart for these new style of consultants. Some of those folks then helped grow the new US Digital Sesrvice. It all started with the Innovation Fellows and grew organically. “The more people we attract people who are amazing into government, the more we energize amazing people already in government, the more air cover we give them” the more awesomeness there will be. Let them create results at 10x what anyone expected. “That methodology is the only replicable, reliable way to change government at scale, at speed, in a way that’s permanent.” “I can’t tell you how much fun this is.”

DJ Patil: My first encounters with the CIOs of existing agencies and departments have been amazing. They’re so open, so eager for disruption.

Aneesh Chopra: The line between public and private sector is becoming very porous. That means that the products of the teams being described are a new form of information that in the hands of entrepreneurs and innovators can be transformative. E.g., Uber wanted their drivers to make better healthcare choices. There’s now a hose of data about the healthcare signups. A startup — Stride Health — took that hose and customized it for Uber drivers; maybe drivers want better back care options. There’s an increasing portfolio of institutions extending these services. A handshake makes more and more of this datea interoperable, and there’s a hand off to entrepreneurs and innovators. “They may not be stamped .gov” but they’ll powered by data from the govt.

Nick: We have an opportunity to do smart wholesaling of data, as well as retailing it: Great services, but also enabling non-governmental groups to build great end-user services.

Lynn: At Commerce we’re trying to do Open Data 2.0. How do we get our data experts out into the world to talk with users ? How do we share data better? How do we create partnerships with the public sector? E.g., Uber shared its data on traffic patterns with the city of Boston.

Lynn: In the departments Todd has led, he has worked on the gender balance. Women were in the majority by the end of his HHS appointment. [I couldn’t hear all of this.]

Todd: I’ve learned that the more diverse the team is, the better the team is. We made it a real priority for the US Digital Service to have a team that looks like America. It’s also our hope that we’ll be minting people who become superstars in the tech world and will encourage more youths to enter STEM.

Aneesh: There were a few places we thought we could have done better. 1. Rethinking the role and nature of the infrastructure. Human capital is the infrastructure for the digital economy. 2. We make rules of the road — e.g., Net Neutrality today — that give people a more fair shot to compete. There are foundational investments to be made in the infrastructure and creating rules of the road. That’s part of how we affect policy.

Nick: What about the President’s new precision medicine initiative?

Todd: It’s a new way of thinking about how you get medical service. Increasingly Web sites provide tailored experiences. Why not with science? Should your aspirin dose be the same for someone with a different genetics, exposed to different things in your environment, etc.? Where it gets really phenomenal: The cost of genetic sequencing is dropping quickly. And tons of data are coming from sensors (e.g., FitBit). How do you start getting a handle on that to start getting better treatment? Another side of it: Bioinfomatics has been amazing at understanding genes. Combine that with clinical knowledge and we can begin to see that maybe that people who live near docks with diesel fumes have particular symptoms. We’ll be able to provide cohorts for test studies that look like America.

Nick: Aneesh and Todd, you both quote Joy’s Law: Most of the smartest people in the world work for someone else.

Aneesh: In many ways, the lessons learned from the innovation philosophy have had great effect in the public sector. The CEO of P&G said 50% of ideas will come from outside of P&G. This liberated him to find innovations in the military that resulted in $1B in cash flow for P&G. Also, we’ve learned from platform effects and what the team at Facebook has done. Sheryl Sandberg: There are 3,000 developers at FB, but a query at Google found 35,000 people with the title “FB developer,” because other companies were using the FB platform.

Todd: It’s important to remember Joy’s Law, and the more you can get those people in the world to care about what you do, the more successful you’ll be. I was asked what I would do with the vast amount of data that the govt has. My first thought was to build some services. But about 17 seconds later I realized that’s entirely the wrong approach. Rather, open it up in machine readable form. We invited four innovators into a room. At first they were highly skeptical. But then we showed them the data, and they got excited. Ninety days later we had a health care datapalooza, and it caught fire. Data owners were there who thought that opening up their data could only result in terribleness. At the end of the datapalooza they flipped. Within two years, the Health Datapalooza became a 2,000 person event, with thousands of people who couldn’t get tickets. Hundreds of new applications that could help individuals, hospitals, healthcare providers were created. But you have to have the humility to acknowledge that you don’t know the answer. And you have to embrace the principle that the answer is likely to come from someone who aren’t you. That’s the recipe for awesomeness to be released.

Aneesh: When Secty Sibelius saw the very first presentation, her jaw dropped. The question was what are the worst communities in the America for obesity and who can they talk to about improving it. In seven minutes they had an answer. She said that when she was Governor, it would have taken her staff seven months to come up with that answer.

Q: [a self-identified Republican technologist] President Obama got the right team together. What you do is awesome. How can we make sure what you’ve built stays a permanent part of the government?

Aneesh: Eric Cantor was doing much the same in Congress. These ideas of opening up data and engaging entrepreneurs, lean startups, open innovation have been genuinely bipartisan.

Todd: Mike Bracken from the UK Digital service says: The strategy is delivery. What will change govt is a growing set of precedents about how govt really should work. I could write an essay, but it’s more effective if I point to datapalooza and show the apps that were written for free. We have to create more and more examples. These examples are done in partnership with career civil servants who are now empowered to kick butt.

DJ: We can’t meet the demand for data scientists. Every agency needs them. We have to not only train those people up, but also slot them into the whole stack. A large part of our effort will be how to train them, find them great homes at work, and give them ways to progress.

Nick: It’s really hard to roll back transparency. There are constituencies for it, whether it’s accountability orgs, the press, etc.

Lynn: Civil servants are the most mission driven people I’ve met. They won’t stop.

Q: Everyone has talked about the need for common approaches. We need identities that are confidential and interoperable. I see lots of activities, but not a plan. You could do a moonshot here in the time you have left. It’d be a key part of the infrastructure.

Aneesh: When the precision medical provision was launched, a critical provision was that they’ll use every regulatory tool they have to connect consumers to their own data. In 2010 there was a report recommending that we move to healthcare APIs. This led to a privately funded initiative called Project Argonaut. Two days ago we held a discussion here at Harvard and got commitments for public-private efforts to create an open source solution in healthcare. Under Nick, the same went on for connecting consumers to their energy info. [I couldn’t capture all this. I’m not sure the above is right. And Aneesh was clear that he was speaking “as an outsider.”]

DJ: If you check the update to the Podesta Big Data report, it outlines the privacy aspects that we’ll be pushing on. Energy is going into these issues. These are thorny problems.

Q: Cybersecurity has become a high profile issue. How is the govt helping the private sector?

Aneesh: Early on the President offered a framework for a private-public partnership for recognigizing digital fingerprints, etc. This was the subject of a bipartisan effort. Healthcare has uniform data-breach standards. (The most common cause of breaches: bad passwords.) We need an act of Congress to [he went too fast … sorry].

Lynn: Cybersecurity requires an international framework for privacy and data security. That’s a major challenge.

Q: You talked about the importance of STEM. Students in astronomy and astrophysicists worry about getting jobs. What can I say to them?

DJ: I was one of those people. I lot of people I went to school with went on to Wall Street. If you look at the programs that train data scientists, the ones who are super successful in it are people who worked with a lot of messy data: astrophysicists, oceanographers, etc. They’re used to the ambiguity that the data starts with. But there’s a difference in the vocabulary so it’s hard for people to hit the ground running. With 4-6 weeks of training, these people crush it. Tell your students that there are great opportunities and they shouldn’t be dissuaded by having to pound the pavement and knock on doors. Tell them that they have the ability to be game changers.

Q: How many of us are from the college? [surprisingly few hands go up] Your msg about joining the govt sounds like it’s tailored for young professional, not for students. The students I know talk about working for Google or FB, but not for the govt.

Todd: You’re right. The US Digital Service people are young professionals who have had some experience. We will get to recruiting in college. We just haven’t gotten there yet.

Lynn: If you’re interested in really hard problems and having a direct impact on people’s lives, govt service is the best thing you can do.

Q: When you hire young tech people, what skills do they typically not have that they need?

Lynn: Problem solving. Understanding the problems and having the tech skills to solve them. Understanding how people are navigating our systems now and asking how we can leverage tech to make that process much much easier.

DJ: In Sillicon Valley, we’re training people via internships, teaching them what they don’t learn from an academic environment. We have to figure out how govt can do this, and how to develop the groups that can move you forward when you don’t know how to do something.

Aneesh: There is a mindset of product development, which is a muscle that we haven’t worked enough in the policy arena. Policy makers too often specify what goal they want and allocate money for it. But they don’t think about the product that would achieve that goal. (Nice shout out to Karim Lakhani. “He’s in the mind set.”)

Q: [leaders of the Kennedy School Tech for Change] Tech for Change has met with administrators, surveyed students, etc. Students care about this. There’s a summit tomorrow. [I’m going!] What are the three most important things a policy school could do to train students for this new ecosystem. How can HKS be the best in this field?

DJ: Arts and humanities, ethics, and humility.

Todd: One expression of humility is to learn the basics of lean startup innovation. These principles apply broadly

DJ: There’s nothing more humbling than putting your first product out there and watching what people say on Twitter.

Lynn: We should be moving to a world in which technology and policy aren’t separate. It’s a problem when the technologists are not at the table. E.g., we need to be able to track the data we need to measure the results of programs. This is not a separate thing. This is a critical thing that everyone in the school should learn about.

Todd: It’s encouraging that the geeks are being invited into the rooms, even into rooms where no one can imagine why tech would be possibly relevant. But that’s a short term hack. The whole idea that policy makers don’t need to know about tech is incredibly dangerous. Just like policy makers need a basic understanding of economics; they don’t have to be economists.If you don’t have that tech knowledge, you don’t graduate. There will be a direct correlation between the geek quotient and the efficiency of policy.

Nick: Panel, whats your quick actionable request of the Harvard JFK community?

Lynn: We need to make our laws easier to understand.

Todd: If you are an incredibly gifted, patriortic, high EQ designer, dev, devops, data scientists, or you know someone who is, go to where you can learn about the Digital Service and apply to join this amazing band.

DJ: Step up by stepping in. And that doesn’t have to be at the federal level. Share ideas. Contribute. Help rally people to the cause.

1 Comment »

June 29, 2014

[aif] Government as platform

I’m at a Government as Platform session at Aspen Ideas Festival. Tim O’Reilly is moderating it with Jen Pahlka (Code for America and US Deputy Chief Technology Officer ) and Mike Bracken who heads the UK Government Digital Service.

Mike Backen begins with a short presentation. The Digital Service he heads sits at the center of govt. In 2011, they consolidated govt web sites that presented inconsistent policy explanations. The DS provides a central place that gives canonical answers. He says:

  • “Our strategy is delivery.” They created a platform for govt services: By having a unified platform, users know that they’re dealing with the govt. They won the Design of the Year award in 2013.

  • The DS also gives govt workers tools they can use.

  • They put measurements and analytics at the heart of what they do.

  • They are working on transforming the top 25 govt services.

They’re part of a group that saved 14.3B pounds last year.

Their vision goes back to James Brindley, who created a system of canals that transformed the economy. [Mike refers to “small pieces loosely joined.”] Also Joseph Bazalgette created the London sewers and made them beautiful.

(cc) James Pegrum

Here are five lessons that could be transferred to govt, he says:

1. Forget about the old structures. “Policy-led hierarchies make delivery impossible.” The future of govt will emerge from the places govt exists, i.e., where it is used. The drip drip drip of inadequate services undermine democracy more than does the failure of ideas.

2. Forget the old binaries. It’s not about public or private. It’s about focusing on your users.

3. No more Big IT. It’s no longer true that a big problems requires big system solutions.

4. This is a global idea. Sharing makes it stronger. New Zealand used’s code, and can then take advantage of their improvements.

5. It should always have a local flavour. They have the GovStack: hw, sw, apps. Anyone can use it, adapt it to their own situation, etc.

A provocation: “Govt as platform” is a fantastic idea, but when applied to govt without a public service ethos it becomes a mere buzzword. Public servants don’t “pivot.”

Jen Pahlka makes some remarks. “We need to realize that if we can’t implement our policies, we can’t govern.” She was running Code for America. She and the federal CTO, Todd Park, were visiting Mike in the UK “which was like Disneyland for a govt tech geek like me.” Todd asked her to help with the Presidential Innovation Fellows, but she replied that she really wanted to work on the sort of issues that Mike had been addressing. Fix publishing. Fix transactions. Go wholesale.

“We have 30-40,000 federal web sites,” she says. Tim adds, “Some of them have zero users.”

Todd wanted to make the data available so people could build services, but the iPhone ships with apps already in place. A platform without services is unlikely to take off. “We think $172B is being spent on govt IT in this country, including all levels.” Yet people aren’t feeling like they’re getting the services they need.

E.g., if we get immigration reform, there are lots of systems that would have to scale.

Tim: Mike, you have top-level support. You report directly to a cabinet member. You also have a native delivery system — you can shut down failed services, which is much harder in the US.

Mike: I asked for very little money — 50M pounds — a building, and the ability to hire who we want. People want to work on stuff that matters with stellar people. We tried to figure out what are the most important services. We asked people in a structured way which was more important, a drivers license or fishing license? Drivers license or passport? This gave us important data. And ?e retired about 40% of govt content. There was content that no one ever read. There’s never any feedback.

Tim: You have to be actually measuring things.

Jen: There are lots of boxes you have to check, but none of them are “Is it up? Do people like it?”

Mike: Govts think of themselves as big. But digital govt isn’t that big. Twelve people could make a good health care service. Govt needs to get over itself. Most of what govt does digitally is about the size of the average dating site. The site doesn’t have to get big for the usage of it to scale.

Jen: Steven Levy wrote recently about how the Health Care site got built. [Great article -dw] It was a small team. Also, at Code for America, we’ve seen that the experience middle class people had with is what poor people experience every day. [my emphasis – such an important point!]

Tim: Tell us about Code for America’s work in SF on food stamps.

Jen: We get folks from the tech world to work on civic projects. Last year they worked on the California food stamps program. One of our fellows enrolled in the program. Two months later, he got dropped off the roles. This happens frequently. Then you have to re-enroll, which is expensive. People get dropped because they get letters from the program that are incomprehensible. Our fellows couldn’t understand the language. And the Fellows weren’t allowed to change the language in the letter. So now people get text messages if there’s a problem with their account, expressed in simple clear language.


Q: You’ve talked about services, but not about opening up data. Are UK policies changing about open data?

Mike: We’ve opened up a lot of data, but that’s just the first step. You don’t just open it up and expect great things to open. A couple of problems: We don’t have a good grip on our data. It’s not consistent, it lives in macros and spreadsheets, and contractually it’s often in the hands of the people giving the service. Recently we wanted to added an organ donation checkbox and six words on the drivers license online page. We were told it would cost $50K and take 100 days. It took us about 15 mins. But the data itself isn’t the stimulus for new services.

Q: How can we avoid this in the future?

Mike: One thing: Require the govt ministers to use the services.

Jen: People were watching but were asking the wrong questions. And the environment is very hierarchical. We have to change the conversation from tellling people what to do, to “Here’s what we think is going to work, can you try it?” We have to put policy people and geeks in conversation so they can say, no that isn’t going to work.

Q: The social security site worked well, except when I tried to change my address. It should be as easy as Yahoo. Is there any plan for post offices or voting?

Mike: In the UK, the post offices were spun out. And we just created a register-to-vote service. It took 20 people.

Q: Can you talk about the online to offline impact on public service, and measuring performance, and how this will affect govt? Where does the transformation start?

Jen: It starts with delivery. You deliver services and you’re a long way there. That’s what Code for America has done: show up and build something. In terms of the power dynamics, that’s hard to change. CGI [the contractor that “did”] called Mike’s Govt Digital Service “an impediment to innovation,” which I found laughable.

Tim: You make small advances, and get your foot in the door and it starts to spread.

Mike: I have a massive poster in my office: “Show the thing.” If you can’t create version of what you want to build, even just html pages, then your project shouldn’t go forward.

Comments Off on [aif] Government as platform

March 19, 2011

[2b2k] Melting points: a model for open data?

Jean-Claude Bradley at Useful Chemistry has announced (a few weeks ago) that the international chemical company Alfa Aesar has agreed to open source its melting point data. This is important not just because Alfa Aesar is one of the most important sources of that information. It also provides a model that could work outside of chemistry and science.

The data will be useful to the Open Notebook Science solubility project, and because Alfa has agreed to Open Data access, it can be useful far beyond that. In return, the Open Notebook folks cleaned up Alfa’s data, putting it into a clean database format, providing unique IDs (ChemSpiderIDs), and linking back to the Alfa Aesar catalog page.

Open Notebook then merged the cleaned-up data set with several others. The result was a set of 13,436 Open Data melting point values.

They then created a Web tool for exploring the merged dataset.

Why stop with melting points? Why stop with chemistry? Open data for, say, books could lead readers to libraries, publishers, bookstores, courses, other readers…


November 6, 2010

[2b2k] World Bank’s open data … now in contest form!

The World Bank has done an admirable job of opening its data for public access. The World Bank has lots of data, much of it at the national level, and throwing it into the public arena — which it did in April — was a gutsy and right move.

They now have a contest, with $45K in prizes, to encourage the development of apps that make use of that data via its APIs. Here’s more about the data:

The World Bank Indicators API lets you programmatically access more than 3,000 indicators and query the data in several ways, using parameters to specify your request. Many data series date back 50 years, and can be used to create interesting applications. You can read more about the data itself in the API Sources section. The projects API provides access to all World Bank projects, including closed projects, active projects, and those in the pipeline. The dataset includes pilot geocode data on project locations; note that these data are collected through a desk study of existing project documents and are being released as a test database — further work is required for data validation and quality enhancements…

Releasing all this data must have required a lot of cultural transformation work. Wow.

1 Comment »