Kundra uses stock market model to determine viability of IT projects.
Vivek Kundra, appointed chief technology officer (CTO) by Washington D.C., Mayor Adrian Fenty in 2007, is determined to make the city's IT organization act like a startup instead of a bureaucracy. Through a concept called OCTO (Office of the CTO) Labs, Kundra turned his 700-person department into a test tube for cloud computing, open source applications and other ideas.
GT: You've developed an interesting model for evaluating IT project viability. Describe it.
Kundra: I created a portfolio management model where we treat every project in government as a stock. We evaluate the management team; we evaluate how happy our customers are; we evaluate the performance, whether it's on time, on budget and delivering value. We use that data to decide whether to hold on to the project, invest more or kill it - similar to what portfolios in a private market would do as far as sell, buy or hold. That's brought a lot of scrutiny and moved us aggressively in terms of killing projects that won't deliver.
GT: You use Google for desktop applications and e-mail instead of traditional desktop software. Why?
Kundra: The government ultimately is a consumer of technology, and a lot of times government gets into the business of developing massive applications. When I moved to Washington, I had more computing power on my laptop at the local coffee shop than the average police officer or teacher. We looked at the cloud computing model and the consumer space. Compared with the cost of owning infrastructure, it's far cheaper.
GT: So the future of the computing platform in the district is software as a service and less hardware ownership?
Kundra: Absolutely. You look at our most recent procurement around project management. We decided to go with the cloud model instead of buying a ton of servers that would have taken me six to seven months to procure, configure and deploy. We were able to do that immediately. When we look at that platform in terms of collaboration, everything's going to be in the cloud. As we're looking at the whole data center model, the question really becomes, why do we need a data center?
GT: Government typically is risk averse. How do you change that?
Kundra: I think controlled risk is very healthy. That's why I created OCTO Labs. The idea was that we would throw hundreds of ideas on the wall and even if five of them survive, they will be transformative.