TECH TRENDS—Web 2.0

Kundra courts the risk of innovation

Call for cloud computing and open data will require shifts in agency IT operations

Cool as a cucumber in the still chilly spring ground: That was our impression of Vivek Kundra the morning we met him backstage after his speech at the FOSE government information technology trade show.

We had set up an impromptu 10-minute interview with the new federal chief information officer for that morning to learn more about his plans for federal agencies. Expanding on his FOSE keynote speech, he talked about how agencies could better use cloud computing and how they should open data to the public.

And he was about to hit the first political bump in the road of his tenure. Nearly the same time we were chatting government 2.0 in the bowels of the Walter E. Washington Convention Center, the FBI was raiding his former office, D.C.'s Office of the Chief Technology Officer, as part of a bribery investigation. Although Kundra was not a target of the investigation, the White House shortly thereafter put him on a temporary leave of absence until it was confirmed that he was clear of the wrongdoing. He returned to the job five days later.

If Kundra did know at the time that the raid was taking place, it did not distract him from getting across his vision for government, which he spoke about with earnest enthusiasm and measured thoughtfulness.

He'll need such grace under fire. The approach he’s calling for will require considerable effort on the part of agency IT shops and will likely meet with at least some resistance. After listening to him speak, two themes keep coming up: cloud computing and open government data. Both could help agencies do their jobs better, though both will require some upfront work.

Cloud computing

First on Kundra's list: take advantage of some of the computing services that are available on the Internet. Kundra has noticed that people are enjoying free Web-based services, such as the Flickr photo-sharing service or Google Apps. "What makes the government so special that it can't embrace some of these consumer technologies?" he asked at FOSE.

Looking back at Kundra's time as CTO of Washington, D.C., can perhaps shed some light on Kundra's position on cloud computing. During his tenure, the city bought Google Apps licenses for 38,000 users in a contract worth about $500,000 a year. Kundra said going with Google Apps could cut the cost of procuring enterprise software while making it easier for a district employee to interact with co-workers. YouTube is another service the district used during his tenure. On the "dcgovernment channel," the city has posted videos of pre-solicitation meetings with interested vendors, for instance.

Although Google Apps and YouTube are obvious examples of cloud computing, Kundra said government could make use of a range of cloud-computing offerings. In an interview with GCN, he distinguished among different types of services. For example, public services, such as Flickr and YouTube, can be used to provide open government data in a cost-effective way. "Government shouldn't be spending a ton of money setting up stuff," when less-expensive alternatives exist, he said.

For more sensitive information, the government could use private clouds that a government agency runs, such as the Defense Information Systems Agency does for the Defense Department. Or agencies could use cloud-computing offerings tailored specifically by contractors for the government market. He also mentioned that agencies could work together to procure common cloud services, which could minimize duplicative services and further reduce costs.

Kundra said he recognized that the cloud-computing approach has some challenges. For instance, how could agencies use cloud-computing services and stay compliant with the Federal Information Security Management Act of 2002? Even before Kundra officially got the job of CIO, the CIO Council started a working group on the issue.

Another challenge for agencies would be adapting to cloud architecture. Traditional in-house systems rely on relational databases to hold data; cloud computing requires a different model for storing and interacting with the data. Much of the logic kept in the database would need to migrate to the application layer. In a recent seminar about Google App Engine, a service from Google that allows customers to run their own applications on the company's services, Wayne Beekman, co-founder of consulting company Information Concepts, addressed the problem.

"The app engine is not a relational database," Beekman said. "You have to think differently."

"Of course, running a legacy application on the cloud would not be possible, as it wouldn't be possible to run it on an internal, modern infrastructure, unless by making significant adaptations," said Andrea DiMaio, a Gartner distinguished analyst.

However, DiMaio also noted that the bigger problem would be changing agency employees’ mind-set about IT. "The problem that we see is mostly the fear of loss of control: location of data, reliability, availability, e-discovery," he said via e-mail. "These are all issues that need to be faced before accepting that mission-critical processes rely on somebody else's virtualized infrastructure."

Open government data

Although Kundra has only been on board for about a month, a buzz phrase already has appeared that reflects another of his priorities: open government data.

The phrase follows from President Barack Obama's campaign promise of transparency. "One of the first things he did after he was sworn into office was sign a memorandum about transparency and open government," Kundra said during his FOSE keynote address. "And that is going to be one of the key agenda items that we're going to be pushing in this administration."

As an example, he pointed to the Recovery.gov Web site, which endeavors to show people how government is spending the stimulus money. "We're going to be publishing government data and [operating on] a default assumption that information should be [available] to the people, not with the default assumption that information should not be in the public domain," he said.

In public appearances, Kundra has pointed out that a new industry of portable location-tracking devices was created when the DOD opened its use of Global Positioning System coordinates. Other industries could flourish in a similar manner as the government opens useful data now locked in databases.

Again, looking at Kundra's work in Washington, D.C., can prove helpful. Last year, the district published on the Web more than 240 data feeds from internal systems, ranging from Metro bus and train timetables to crime reports. The idea was that third parties, such as businesses, nonprofits or individuals, could use the feeds as part of an application, much in the same way location-aware devices use GPS.

To spur applications that use this data, Kundra then kicked off a contest, called Apps For Democracy. Instead of building applications for the public, the district hoped to motivate volunteers to build apps, for the Web, mobile phones or another platform. The effort led to more than 47 new applications, at a fraction of the cost of building them in-house.

On the federal level, Kundra has already proposed a new site, Data.gov, that could be a clearinghouse for federal government data feeds. This site could be a repository of data and directory to government data sources elsewhere on the Web.

Opening data for third-party use is a nimble way of using that data, because people can shape it for particular audiences, said Kevin Novak, a co-chairman of the eGov committee of the World Wide Web Consortium (W3C). He cited how the privately run Sunlight Foundation paired campaign-finance records and congressional voting records.

"It's not a government mission to do that, but there is definitely an interest from the public's perspective for that information," Novak said. "And there are organizations that can get the people together to do those sorts of things."

To help prepare for making data open, Kundra wants agencies to increase server power so that data can be reliably and speedily delivered to third parties. He also wants data accessible in easy-to-use, nonproprietary formats, such as Extensible Markup Language and the Extensible Business Reporting Language (XBRT), or available via Really Simple Syndication feeds.

A third challenge is ensuring the right data gets out. "We've been talking to the CIOs about [starting to] ensure that they don't release any data that is private or confidential,” Kundra said. “You have to have the right security protocols and only make information public that is public."

Questions also are brewing around how and what information should be made public. Certainly, completed records should be made public. But how will agencies ensure that the authoritative version of a document, rather than a superseded copy, is the one widely available? Novak pointed out that the public tends to use the first few returns on an Internet search service as the authoritative version. To tackle this problem, some of the talk among government attendees at a W3C e-government workshop, held last month in Washington, related to the use of Uniform Resource Identifiers, or permanent Web addresses, for documents.

"In general, agencies are pretty far" from the goal of opening their data, said W. David Stephenson, an e-government strategist and co-authored with Kundra a forthcoming book on e-government, “Democratizing Data: Transforming Government, Healthcare, Enterprises, the Environment and Everyday Life.” The good news is that opening data “is relatively straightforward," thanks to a wide range of available standards, he said. "The specific tools will change over time, but [by using] a standard, where you tag data once, and it remains independent, then that data can be drawn by any application."

Charging into the future

Goals such as opening government data and using more cloud-computing services might be ambitious, but the Obama administration seems to have picked someone suited for the charge. After a few short weeks as CIO, federal agency IT thought leaders are buzzing about how they can carry out these ideas. And these are the people Kundra will rely on.

"I know it's not going to be easy, and it's not going to happen overnight, especially if you look at the fact that there are over 4 million federal employees and there are over 10,000 IT systems in the government. A lot of hard work and a lot of good work has happened, and what I intend to do is…leverage the federal employees who have dedicated their life's work to moving forward, to really challenging the status quo," Kundra said at FOSE. "I can tell you, having spent some time within the federal agencies, I've been amazed that some of the smartest people I've ever met in my life are federal government employees."

"Vivek's main strength, I believe, is that he's not afraid of taking risks," DiMaio said. "His attitude can be summarized in one sentence he said he uses when people challenge his innovative ideas: 'Prove me wrong.' This is actually reversing the burden of proof. As opposed to any innovation being risky until proven otherwise — which is why governments have so much trouble with innovation — he takes a novel idea and challenges his opponents to prove it too risky or impossible to implement."

Reader Comments

Tue, Apr 21, 2009

Well said. Prove me wrong is perhaps not the best choice of words. Safety first. Open records are already available under FOIA. Google is still in its infancy, as most newby technologies, and no matter how great, I don't believe we should entrust our National Security and IT systems developed over generations and generations to be exploited via a company that will probably be bought by a "newly" formed group of foreign private equity investors. I'm just saying...

Sat, Apr 4, 2009

While innovation is good, 'Prove me wrong' is actually not an option for those entrusted with protecting the nation's information systems. One must proactively assess the risks and decide what's reasonable in advance of putting systems into production, not simply forge ahead and discover the laws broken after the fact. -T

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above