Gov2.0 Summit, Day 2

Posted Thursday, September 10, 2009 at 9:02 a.m. by Chris Amico in News and Roadside Blogging about gov2.0, government, open source and Tim O'Reilly

It's the last day of the Gov2.0 Summit, and I'm back for another stretch of liveblogging. Today, we'll hear from Carl Malamud, Vivek Kundra and Ellen Miller of the Sunlight Foundation early on. I'm expecting another fascinating day of talks.

As before, I'll be liveblogging below and cross-posting to Twitter. Feel free to leave comments any time.

Updates: oldest first | newest first

  • 9:40 a.m.

    Today's first speaker is Carl Malamud of He's one of the people behind the Eight Principles for Open Government, which outlines exactly how public records and data ought to be released. From today's schedule:

    Open government pioneer Carl Malamud draws on his own experience changing the way numerous government agencies--including the SEC, Patent Office, Smithsonian, and our federal courts--publish their information on the Internet. He puts the changes we are witnessing today into perspective, drawing on 200 years of American history and showing why access to the working of government by private citizens is crucial for our economy and for our democracy.

  • 9:46 a.m.

    Carl Malamud is "really the guy who we owe this entire event to," says Tim O'Reilly.

    Malamud was responsible for getting the SEC online in 1993. "He spent a lot of this career being ahead of the curve," and "a geek thinker."

  • 9:53 a.m.

    "When Abraham Lincoln spoke of government of the people, by the people and for the people, he wasn't just consecrating a battlefield. He was speaking of a transformation in government."

    This was a new transformation.

    Washington and Adams were revolutionaries, "but aristocratic revolutionaries."

    When Jefferson took the White House, "he abolished the formal dining table at the White House, and replaced it with a round one, so no one could sit at the head of the table."

  • 9:55 a.m.

    Lincoln's wave was a second transformation. The day he took office, a new government printing office opened, creating what is now the Congressional Record, the official record of government.

    "The Lincoln wave of transformation was one of fully documenting government."

  • 9:57 a.m.

    The next turning point came in 1911, Malamud says, in the Triangle Shirtwaist Factory Fire. That tragedy led to the first fire codes, then later regulations on electricity, plumbing and safety.

  • 9:59 a.m.

    In FDR's time, Justice Brandeis said: "Without systematic publication of the rules, ignorance of the rules would become a defense."

  • 10:01 a.m.

    We are now in the third wave of transformation, with the government. This wave started in unlikely places, Malamud says.

    In the 1990s, the Air Force let the public access GPS. Later, USGS released extensive digital maps.

  • 10:02 a.m.

    Before the SEC database went online, getting documents meant knowing and asking for exactly what you wanted, and paying for it, in person.

    "In this environment, government produced documents to sell. Information was viewed as a source of profit for the government."

  • 10:04 a.m.

    "For too long, access to public information has been a matter of access to inside information and access to money and power."

  • 10:05 a.m.

    Because of PACER's cost, the executive branch has paid $50 million to access records from the judicial branch. Law students, researchers and others limit their use of PACER because of the cost.

    "How can we be a nation of laws not a nation of men" if we can't get to the law itself?

  • 10:08 a.m.

    Opening data is "an opportunity for citizens to make government more efficient."

    Three propositions that should be true in a free society:

    • "If a document is to have the force of law, it should be free for all to read."
    • "If a meeting is part of the law-making process is truly public, that means it must be on the internet." "If you want to attend a hearing today, you best live in the beltway, and you'd better hire someone to guard your place in line." "Shielding hearings from the public eye reduces the credibility of Congress."
    • "The principal that primary legal material should be available to all is a principal that should be led by the executive branch and should apply to all levels."

    We need to open source society's operating system.

  • 10:12 a.m.

    Standing ovation for Carl Malamud as O'Reilly calls him back on stage. "Carl is a man on a mission."

    "The mechanisms of government come out of the problems of our society," O'Reilly adds. "They really came from burning needs."

    "Transparency really leads to better government."

  • 10:13 a.m.

    Getting specific now, Ellen Miller of the Sunlight Foundation will speak with Vivek Kundra:

    Moderated by Ellen Miller, Co-founder and Executive Director of the Sunlight Foundation. What is the state of and other government transparency initiatives? Open data is one of the foundations of the digital commonwealth. It fosters re-use and innovation, and also helps citizens hold government accountable. But it also raises privacy and security issues. What data should be made available? How are innovators making it more useful? What policy issues need to be addressed? What can government learn from Web 2.0 and cutting edge technology companies about how the data-driven organization works, and what it makes possible? Ellen and Vivek will tackle these issues and more in a conversation on the key issues of the Summit.

  • 10:15 a.m.

    Miller: "Vivek and agree on some fundamental things of what open government means."

    "Public data means online. I mean this is the 21st century."

    One big question: Open data means open in real time, Miller says. Six month old data is not as useful.

  • 10:17 a.m.

    First question: Government as a wholesaler of data?

    Kundra: "It's very important that government democratizes data. It makes data available in a machine-readable format."

    It's not as useful for government to "slice and dice" data and be a "data retailer," Kundra adds.

    "What we found is that people, the American people, are going to be able to do that far better than an individual trying to present on side of that data."

  • 10:18 a.m. was built on top of Sunlight's, though at significantly higher cost. But cost aside, how do we get data that's more timely and more accurate?

    Kundra: "The infrastructure that a lot of these agencies run on ... a lot of those agencies have aging systems. They're built on COBOL infrastructure. They're built on systems that can't handle the demand."

    Second, he says, we need to change the culture of data in DC. People want to go through more layers of trying to clean up information before releasing it.

    Kundra said is faster to release the data quickly and let the public find gaps. "Within 30 days...we noticed the quality of data was going up."

  • 10:21 a.m. is "one of your signature accomplishments, and a great one," Miller tells Kundra.

    But, "you picked the low hanging fruit. You put it online and nobody got hurt and it looks terrific."

    What's the plan to expand it and "get into the meat of data?"

    Kundra: " is the first brick when you think of open data and government." It's the first step to being a platform.

    The government needs to shift from technology and move debates toward public policy. If you bought a camera, you could compare megapixels and cost and quality. But you can't do that with health care, or energy. "How do you compare and effectively make public policy decisions that are data driven?"

    Government is not the only source of good ideas, says Kundra.

    "How do we get people to give us better ideas?"

    "We always knew that the most interesting ideas were at the intersection of multiple data feeds," Kundra says, so it's no surprise, he adds, that DataMasher won Apps for America.

  • 10:25 a.m.

    What's next?

    Figuring out how to "hard wire releasing data as part of the agency culture."

    But not every data feed is created equally. It's about getting useful data, not just a record of press releases.

  • 10:27 a.m.

    Government and OpenID?

    "The US government is still investing in platforms it maybe shouldn't be investing in."

    If you wanted to reserve a campsite at a national park, you'd have to create a unique account. Then create another one for every other agency you use.

    "How do we leverage platforms that already exist?"

    Customers--citizens--already have accounts with Yahoo and Google. Why not use those accounts to authenticate across multiple government services.

    "This will allow us to move from government websites that are just brochureware" to applications that provide interactive services.

  • 10:30 a.m.

    Miller: Why does cost $7 million to rebuild? Why so much?

    Kundra: "This has never been done on this scale." The new will track spending of the entire $785 billion stimulus, down to the county level.

  • 10:31 a.m.

    "We need to find gamechanging strategies to solve these problems." Kundra says the government is looking at cloud computing solutions to save money.

    Here's another example: When TSA wanted to launch a blog, the initial cost was estimated at $6 million. But "wiser heads prevailed," Kundra said, and "we realized we could host it free online." Here's the blog, running on Blogger.

  • 10:34 a.m.

    Speaking of cloud computing in government:

    What might be happening that could clarify things or make complexity knee bending and overwhelming for government agencies? In a discussion we call Clouds Over Washington: The Cloud And How It Can Serve Government, we explore how the move to cloud computing might provide significant benefit to openness, transparency, and easier collaboration-or result in significant problems.

  • 10:35 a.m.

    Casey Coleman, the CIO of the GSA, says this is about fundamentally changing the way government operates. Procurement and security are complex and expensive. "Our goal is to abstract some of those issues away."

  • 10:37 a.m.

    Plenty of government data is "low-risk" and those bits are candidates for commercial cloud hosting.

    Dr. Werner Vogels, Vice President & Chief Technology Officer at "Amazon itself needed to virtualize some of its own services."

    Engineers across the company were solving similar problems, so Amazon consolidated its server infrastructure so its developers wouldn't have to think about it individually.

    "Scalability isn't just about scaling up. It's also about scaling down when we don't need those resources."

  • 10:40 a.m.

    "Folks don't want to be bothered on things that don't differentiate their business."

    "Everybody has to build infrastructure. It doesn't add value." Hence, Amazon takes care of that.

  • 10:41 a.m.

    The key to the cloud is "only paying for the resources you really use."

    So, what kinds of data can the government put in the cloud? And more importantly, O'Reilly asks, will agencies share resources?

    Coleman: We really have many small customers in the government, not one large one. "One of the advantages of cloud computing is that you can start small."

    O'Reilly: Can the government do rapid development? "How much room is there for that kind of experimentation?"

    Coleman: "Right now, there is not a culture in rapid development, in rapid release cycles." The friction comes from "the time to complete the procurement, the time to complete the security and requirements processes."

  • 10:46 a.m.

    Vogels: We're really in Day One of cloud computing. This is new, even to Amazon.

    Amazon just launched a "virtual private cloud" that's completely firewalled. That might be useful for more secure government processes.

  • 10:49 a.m.

    Portability and interoperability are a huge concern, Coleman says. The government doesn't want to get locked into any particular technology.

    O'Reilly: "How do you think about interoperability when it's not just infrastructure that you're abstracting?"

    Coleman: "We would welcome services that gave us more of a middle layer."

  • 10:52 a.m.

    Question to Vogel: Why did Amazon start at such a basic level?

    Vogel: It's something we had to develop for ourselves. Part of the reason for keeping services so low-level was keeping costs down, which Amazon counts as one of its core competencies.

    More, there's no requirement to use more than you need, Vogel says. If you just want storage, there's S3, and no need to get EC2 to manage it.

  • 10:54 a.m.

    Audience question: Heard that Interior is starting its own cloud computing services. True? Thoughts?

    Coleman: That's probably Interior's business operations center, which is fee-for-service.

    Part of the thing about cloud computing is that the lines between systems get blurred. It's important to think about needs, and maybe Interior can provide for some needs.

    Vogel: It's possible the cloud can provide more security than you can provide yourself. "Keeping your data secure is a very specialized task." Think about HIPAA. The overall process needs to be secure, not just the infrastructure.

  • 10:58 a.m.

    Eric Ries is up now, talking about doing more with less:

    Startups aren’t just for the garage anymore. A new kind of entrepreneur is emerging from startup hubs worldwide. They are mastering not just new technologies but the discipline to create value for real people. The lean startup is a look at this new trend and ways it can be applied by organizations large and small.

  • 11:01 a.m.

    First thing: Most startups fail.

    "What bothers me isn't that startups fail, after all this is a risky venture, but what bothers me is that most fail for a really bad reason."

  • 11:02 a.m.

    "A startup is an experiment, an inquiry into how the world might look under the vision of its founders."

    Most startups fail not because the technology doesn't work, but because nobody wants the technology when it's ready.

    It turns out management is a key discipline.

    "Most startup founders have horrendously bad ideas when they start out." The key is the pivot. "Within every bad idea is a kernel of truth," you just have to chip away the bad ideas around it.

  • 11:05 a.m.

    "If we can do those pivots faster, then we can complete our process before we run out of money."

    It's not about technology. This is really about management of people who build technology.

  • 11:07 a.m.

    Neither waterfall nor agile really work for lean startups. Both require an understood problem. Waterfall needs a known solution.

    The lean startup loop looks more like the OODA loop, which was pioneered by fighter pilots to minimize time to feedback.

  • 11:09 a.m.

    Question to discuss over the break: What if you could do all the experiments you wanted? What if you could do dozens?

    Talk amongst yourselves.

  • 11:10 a.m.

    We're getting into practical issues now with Elliott Fisher (The Dartmouth Institute for Health Policy and Clinical Practice):

    In this session we examine some of the possibilities that access to data might have for Health2.0. How might we use data as a platform to improve health care and reduce costs? Dr. Elliott Fisher is Principal Investigator of the Dartmouth Atlas of Health Care. He is also Professor of Medicine and Community and Family Medicine at Dartmouth Medical School and Director of the Center for Health Policy Research in the newly established Dartmouth Institute for Health Care Policy and Clinical Practice. His work with the Dartmouth Atlas was cited in a New Yorker magazine article and used by President Obama as a reference in his health care plan.

  • 11:36 a.m.

    "What does higher spending buy?"

    A number of studies showed "quite convincingly that higher cost does not equal higher quality care."

    "Lower spending is compatible with better outcomes," Fisher says.

  • 11:37 a.m.

    So, in high cost areas, where does that money go?

    In Miami, for example, people see physicians almost three times as often as those in Salem, OR, which has a far lower cost of care.

    Part of the underlying problem, Fisher says, is a failure to recognize the role of local health care systems as drivers of cost. Lack of support for improvement, care management and coordination. Also, the public equates more with better. Finally, the present payment system rewards more care, not better care.

  • 11:40 a.m.

    So, how do we use data to fix all this?

    Here's a map of local health care spending.

    Dartmouth asked: What if all local systems operated like those in the lowest-cost areas?

  • 11:42 a.m.

    Common to all high-performance, low-cost areas: shared aims across the system, use of data feedback to change practices.

    For example, how often do doctors order EKGs? Turns out some order too often, others not enough. By sharing data, doctors talked about when it was really needed, and when they could skip it.

  • 11:46 a.m.

    Much of this thinking fed into Atul Gawande's excellent story in the New Yorker, which dealt with McAllen, TX.

    If you haven't read it, do.

  • 11:48 a.m.

    Q: When will we see data that helps us understand how agriculture policies affect health care policy?

    A: "I think that conversation is coming." But we don't have access to that data yet.

    Q: What's the correlation between urbanization, income and cost of medicine?

    A: "Most of the research we've done has accounted for those differences." There are metro areas with two-fold cost differences, and you can see the same patterns across rural areas.

    "Go back to your communities and use data to transform the system."

  • 11:52 a.m.

    Staying on health care, next is Carleen Hawn of HealthSpottr, highlighting things that work in the current system.

    First example: a physician prescribes antibiotics for a throat infection. Each day the patient gets an email asking how they feel, with links to better, same, worse or resolved. The doctor sees a matrix of patients and can quickly tell what's working for whom.

  • 11:56 a.m.

    Healthspottr is now offering fellowships of $100,000 to $250,000 for health innovation. "We really think the government should be offering contests like this."

  • 11:59 a.m.

    Next, James Heywood of PatientsLikeMe.

    The current doctor-patient interaction "doesn't really work." Instead, "what if we formed a new social contract." Patients share their medical records in a "targeted, meaningful" way.

    His brother has ALS. "I can find the four patients in the world most similar to him." He can ask, "is this normal?" Other patients have been there and can help.

  • 12:01 p.m.

    We're rifling through visualized data here. "Look at this journey across drugs," Heywood says. "Is this real world drug safety?"

    These are real questions: "Can I change my outcome?"

    "What's the best outcome I can expect to achieve? And how do I get there?"

    "Can we use this know if treatments work in the real world?"

  • 12:05 p.m.

    "The real transformation to me I realized that all the barriers we had built...prevents us from using all these innovations in technology." It means we don't have a Moore's Law of medicine.

    "There are so many barriers to doing this well."

    "The transformation that makes this possible is when we as individuals decide to openly share data in a secure community context...It's when we decide to build this well."

  • 12:07 p.m.

    Shifting slightly, H1N1 is back in the news. How do we use data to deal with it? How can citizens help?

    In this session we’ll examine how modern networks permit us to harness information flow to provide early warning and vastly improve our response to system-stressing events such as the H1N1 pandemic. Technologies available now offer governments and engaged citizens much better pandemic surveillance, situational awareness, and decision-making across all levels of response and can provide for much greater resilience. Jim Stogdill of Accenture serves as our moderator; he will be joined by two world class leaders in medicine, collaboration, resilience, and information technology: Michael McDonald, MD; President, Global Health Initiatives, Inc., and Eric Rasmussen, MD; CEO, INSTEDD, a .org begun with initial funding from Google to look for early signs of disease detection through innovative approaches to technology and collaboration.

  • 12:08 p.m.

    Main question: "Can the technologies emerging today help us make a real difference in dealing with something like H1N1?"

    The big difference won't be the vaccine, says Michael McDonald. It'll be in communication and management of the disease.

  • 12:15 p.m.

    But do we have the capacity to act on the information we now have access to? Eric Rasmussen is less optimistic. We need to find a way to put medical capacity where its needed.

  • 12:17 p.m.

    Resilience in the health system is built on horizontal communication, meaning people share information with each other, not just with a central hub. There isn't much sharing vertically within health systems.

  • 12:20 p.m.

    Key take away on H1N1: Behavior and information sharing (especially as enabled by open data) is going to be more important in fighting swine flu than vaccines and other medicine.

    Be sure to watch this session when the video is available.


Comments are closed for this post. If you still have something to say, please email me.