Tim O’Reilly on Government 2.0

Pioneering technology publisher Tim O’Reilly tells us that “government as a platform” is the definition of government 2.0. To explain to a non-technical audience what this means, he explains that his company specializes in finding innovations at the edge and amplifying them, through events, publishing and market research. This involves watching alpha geeks like Rob Flickenger. Tim says he knew Wifi was important when he saw Flickenger on the roof of the O’Reilly building using a cantenna to bring Wifi to his favorite coffee shop. Similarly, they were able to anticipate web services by watching developers build screenscrapers, using other websites as data sources.

Tim helped coin the term “web 2.0” and offers a definition of the term. “Top internet sites are built on huge databases which get better the more people participate,” This is a new paradigm – “data, not some sort of hardware, is the ‘intel inside’, the source of lock-in” to appealing platforms.

As an example of how this works, Tim points to Google Voice Search. It gets better each time we use it, learning from user input. And it coordinates three databases – speech recognition, a search database and a location database linked by the Internet into a common platform.

Innovators have begun bringing government into this new paradigm. Carl Malamud helped put the SEC online, using a small NSF grant, data from the SEC and a lot of persistence. Fifteen years later this has helped turn into a vast movement for government transparency. In the UK, Tom Steinberg founded MySociety, and introduced tools like They Work for You, which increases parliamentary transparency, and Fix My Street, which allows individuals to report potholes and ask the government to fix them. This has now been picked up by 311 services throughout the US.

Our new president appears to understand this in a deep and fundamental way. His campaign platform was a self-service organizing platform much as Craigslist is a self-service advertising platform. The question is whether we’ll actually see this in governance. Tim reminds us that “government has always been a platform for collective action,” reminding us of Ben Franklin’s quote, “We must all hang together or we will assuredly all hang seperately.” Franklin’s version of government invited lots of citizen participation, including ideas like a government matching grant – citizens could raise a certain amount of money, and government would match the funds raised.

Somehow, Tim says, we got lost and turned to “vending machine government”, a model where we put in taxes and take out services. Can we undo this, and build government that enables four types of interaction:

– Government to citizen – providing services and information to citizens
– Citizen to government – citizens report on probelms that need government assistance
– Citizen to citizen – not every problem needs to be solved by government
– Government to government – we need better cooperation within government agencies

Tim suggests that there are some lessons from the technology space that could be useful in building Government 2.0

Build open, expandable systems
The rise of the IBM PC platform had to do with the fact that anyone could build compatible hardware, or that Michael Dell could built his own low-cost machines. The web succeeded because Tim Berners-Lee allowed anyone to use his code and build their own website. This is an example of what my colleage Jonathan Zittrain calls “generativity” – the “capacity to produce unanticipated change through unfiltered contributions…”

In open government this might mean open, portable health records, or open data that allows competition by third parties on government contracts.

Build simple systems and let them evolve
The original sketch of Twitter, Tim shows us, was half a sheet of legal paper. The system’s incredibly simple, but there are now 11,000 applications running on top of it, written by third parties. Simple systems like the Internet Protocol can act like hourglass models – they run on a diversity of systems, and support a diversity of applications around a simple protocol.

“Complex systems built from scratch never work. You need to build a simple system and let it grow… Complex problems paradoxically require simple answers.”

Design for cooperation
The Unix operating system was built around the idea that we could join together independent programs with no more than a protocol that allows these programs to work together. This allows for a very different school of software development than in Windows, where 90,000 developers need to figure out how to work together. In Linux, thousands of loosely coordinated little groups build the system together.

The notion of governance via loosely coordinated groups is a Jeffersonian one. And a system like the Internet domain name system looks decidedly Jeffersonian (as David Post points out in his new book.) We can build complex systems, like DHS Virtual Alabama, by encouraging people with lots of data to cooperate and share and build complex maps that allow for recovery from natural disasters.

Learn from your users
Google was late to the game in mapping. But Google is used by 45% of all mashups online. That’s because when innovators started building mashups of Craigslist and Google Maps data, Google didn’t shut the door, but hired the first guy to build a mashup, and then released an API to make the task easier.

Fedspending.org was a site built by OMBWatch, an NGO funded by the Sunlight Foundation. Their tool was so good, it ended up obviating a system the government was building for much more money – the government ended up throwing out their system and using theirs instead.

Lower the barriers to experimentation
The government tends to treat projects like the Apollo 11 rocket launch: “Failure is not an option.” It should be. We fail all the time, and we need to learn from it. He quotes Edison: “I didn’t fail ten thousand times. I successfully eliminated, ten thousand times, materials and combinations that did not work.”

Much innovation comes from a single engineer within an entity like the New York Times, putting archives up on an inexpensive, rented server from Amazon. The low cost of failure made it easier to experiment.

Build a culture of measurement
“If it works, do more, if it doesn’t, stop doing it.” We need to watch how our systems succeed and fail, and build systems that respond to user stimuli. And we need good metrics which we can watch carefully. As Atul Gawande demonstrated with his recent, brilliant article on healthcare, we need to ask quesitons like “How do we measure the success of healthcare?”

Google runs auctions almost continually for it ads, taking advantage of “realtime economics”. Walmart runs a system that connects a consumer purchase to an order from a factory within 14 seconds. Realtime data is the backbome of these “living organisms, responding in realtime to stimuli.”

Throw open the doors to partners
Tim celebrates the iPhone ap store, suggesting that it worked vastly better than more controlled models for aplication development on the Blackberry or Nokia phones. Governments need to stop using tools like earmarks, sole source licensing, and no-bid contracts, which lead to a less open ecosystem.

We also need to make sure eople understand what data comes from the government. He quotes an unnamed congresscritter who asked him, “Why do we need NOAA when we’ve got weather.com?” We need to show what the government can provide and what people can build on top of it. The government launched satellites, and many companies built great GPS tools on top of it.

Tim closes with the idea that government needs to be a vehicle for collective action,
a convener first, and a problem-solver second. He references an effort in Kauai, Hawaii where local businesses faced the closure of a state park due to a washed out road. “They could protest – shaking the vending machine – but instead, they coordinated.” They brought in materials and workers and fixed the road within three days.

Fixing complex problems requires figuring out what government needs to do, what private entites can do and what coordinated citizens can do. If we build systems that allow all these behaviors, we’ll see a great deal of positive change through Government 2.0

Please see John Palfrey’s notes as well for another perspective.