Digital Media in the Cloud: A Conversation with Vivek Kundra

It's great to be here, thanks for having me. And Vivian, it's great to be here with you. I love the way life works, we surf here, we surf there, and it all folds back in together. It's great to be here with you this morning. I hope your son is sleeping well.

Jambo, habari gani? Jambo, habari gani.

There are no memorials on that side of the world, but I think a big part of what's happening as far as the digital world is concerned, it's an issue that we're going to be discussing as you think about the physical world. What does it really mean in terms of continuity of history when more and more content is actually digital-born? I think that's one of the things that we're grappling with. I remember with the National Archives, one of the big struggles is to figure out how do you actually catalog all that history for centuries to come. Things that can be born digitally can die digitally, and die relatively easily.

Well, I think a lot of it is actually not different in terms of the work that I did when I was at the White House. In a lot of ways, I thought of myself almost as a digital journalist in the sense that I believed, as you look at the foundation of our democracy, a more open, transparent, and participatory government is what we really want to aspire to and build. I spent a lot of time using the very tools that have empowered whether it's businesses to fundamentally drive efficiencies within their own ecosystems, or it's people and the ability to connect. I very much tried to figure out how do we make sure that as we're thinking about government, we recognize two principles. Number one, there's a fundamental change in technology and a big part of it is going to be driven by Darwinian pressures. And number two, we need to recognize that we don't have a monopoly on the best ideas. Based on those principles, a lot of the investments we made were very much about making sure that journalists and the American people actually had the information they needed to tell the story of what was going on with their government.

Absolutely not. Because it can be used in a lot of other ways, which is to actually close and limit access and make sure that a few powerful actors have asymmetrical access to information. Just because it's digital, it doesn't mean that it's inherently open and transparent.

Well, I'll never forget my first day on the job. What happened was I was handed a huge stack of PDF documents, and my team that had been printed out and stapled. And I said, 'Congratulations, you're now responsible for overseeing 80 billion dollars in IT spending, and here's a stack of 27 billion dollars worth of IT projects that are millions of dollars over budget and frankly years behind schedule.' The second thing that happened was I found out that my team was assigned BlackBerries based on the number of years they had served in the government. And if you remember, even the President had to fight tooth and nail to get a BlackBerry. You wrote, 'My neighbor's ten-year-old could look at the latest stats of his favorite baseball player on his phone on the school bus, but I couldn't get an update on how we were spending billions of taxpayer dollars, 80 billion, while at my desk at the White House.'

Well, the first thing I recognized was that no amount of resources within my office were actually going to be able to solve this problem alone. So I looked outwards in terms of other models where you had huge efficiencies and transparency. And the model I looked at was the stock market. What's really interesting about the stock market is that you have transparency in terms of what's happening as far as the price of a stock at each millisecond, each nanosecond. You can buy, sell, or hold, make decisions. And the simple question I asked myself is, well, why can't we apply that to billions of dollars in federal IT spending? I recall testifying in front of Senator Collins, and she asked me the very question you asked me. And I said, 'Well, Senator, what I'm going to do is an IT dashboard, and I'm literally going to put every major IT project online with the picture of every CIO in the United States government that's responsible for the project and how it's performing.' So within a 60-day period, I launched the IT dashboard. We got the President to take a picture of him looking at the IT dashboard, and I blogged about it. Within minutes of doing that, I had so many phone calls from CIOs across the federal government, from the Department of Energy to Health and Human Services, who said, 'Oh my god, for the first time my cabinet secretary called me and asked me why a project is red.' And that was the power of transparency. We're literally shining light on the operations of government. It ended up having a huge impact on how we were spending taxpayer dollars. The Department of Veterans Affairs actually halted 42 IT projects, terminating a number of them, and we were able to save 3 billion dollars as a result of shining light on the operations of federal government.

Well, I think this is where ONA can play a big role. One of the things I recognized after having launched the IT dashboard and the data.gov platform was that journalists weren't really using those resources to tell the stories of what's really going on within the government. I think what can happen in a big way is actually making sure that you're also building an ecosystem of users to be able to take that data. On the dashboard today, you could literally go online and download the raw data of how every major IT project is performing. On data.gov, there are over 400,000 data sets on every aspect of government operations, from health care to education to what's going on with the environment. And what I haven't seen is a huge pickup and a number of stories that are being told in a digital way using that raw data that's been made available publicly.

Absolutely. And I remember also when I was in that DC government, one of the things that we did there was I literally took from an IT perspective the operations of what was the police department, the public school system, and what we were doing in our office, and put all this information out there in terms of homicides, in terms of garbage collection and the time it was taking, in terms of permits that were being issued or school performance, and also contracts that were being awarded within the technology department. So you could literally see the government official who awarded the contract, the company that won it, and how much the government was paying them on an hourly rate. We saw immediately huge disparities in terms of who was getting paid how much on the permitting side, were able to detect all these disparities in terms of preferential treatment and so forth. So it also serves as a mechanism of fighting corruption as people use that data. There are a number of really interesting use cases that came out of that open data movement that was in the District of Columbia.

Well, I think it's available in every aspect of government operations. And the principle that I mentioned, that the government doesn't have a monopoly on the best ideas, through that what I really mean is that I have no idea what value you'll be able to create or the stories that will be able to be told at the intersection of multiple data sets. So it's not like you can take data, let's say, on the environment or in health care, but the real value is going to be to bring those two data sets and recognizing that there's real value that's going to be created and stories that can be told.

Well, so there are plenty of tools there to actually take that data and use it and slice and dice and cube it. But there's a decision that I had made when I initially launched it, which was a question of how did we really want to launch the platform. We could either have built really nice-looking applications and said this is what the government believes the truth is, or we could have said we'll make all this information available in its raw format and the American people can decide how that information can be used. I went with the latter approach. As a result of that, what we've seen is not just journalists or, you know, you saw a handful of people using that platform, but also entrepreneurs who are creating applications that they're selling. An example is somebody took a data set from the Consumer Product Safety Commission, which shows you every product that's been recalled, it has real-time results of injuries at hospitals, and they built an iPhone app that allows you to literally go in there. So if you're an expectant mother, you can go in before you buy a crib and see whether it's been recalled or not by scanning the barcode, and you'll get the federal government data on what are the products with that crib. That way, on your way to the surgery room, you can see this hospital's record on the surgery you're about to have. You can see the performance of that hospital in terms of patient ratings, you can see how the surgeries are rated in terms of outcomes and so forth. So that's the amount of the data that's out there, and I believe it's an unexplored field in terms of journalism. And one of my biggest issues when I look at sort of stories today is that why do they have to be a point in time only? There are a number of stories that I believe are evergreen, especially if they're digital, so the numbers can just keep up with the reality.

Absolutely. And then bring in the narrative. So there's always two sides of the story, right? There's qualitative and then there's quantitative. So you could refresh the narrative every so often, but from a quantitative, if that data is already there and governments are making it available, why not start building stories that live forever?

Well, I think there is sort of a battle for the soul of the future of journalism right now in terms of what these technologies that are sort of game-changers and the Darwinian pressure that they're bringing to the market. So you actually see some organizations that are doing an amazing job in terms of innovating and building a capacity, where others ignore this at their peril in terms of a fad that's going to disappear. And the question really is, do you want to be Amazon.com or do you want to be Barnes & Noble?

Absolutely.

Well, I think it's, I can't place a percentage on it, but what I can tell you is that with 400,000 data sets that are there, it's just the beginning. Right? Because there should be millions and millions of data sets. If you look at what's happening throughout our world today, every aspect of whether it's our infrastructure or even our lives to some extent is being instrumented. So think about the transportation grid, there's sensors everywhere that are telling you how trains are running, buses are running, what time they're coming. If you look at our health care system, to drive efficiency, more and more sensors are being deployed. And what this means is that big data becomes a huge problem as far as not just the storage and the ability to slice and dice and cube it, but also to separate noise from signal. And I think in terms of journalism, if you're really interested to see how stories are told by slicing and dicing that big data, taking data, turning it into information, information into insight, and that insight into stories that can be persistent.

Well, I think they should push back. And here's why. I think it's absolutely wrong if people don't know what's happening with the information that they sign up with an agreement and it changes within minutes of signing up for that. I'm a big believer that we need to shift power to individuals, because what you're seeing is also a difference in terms of how generations perceive what is acceptable versus what is not acceptable. So if you talk to people who are born digital, so to speak, they're very comfortable with sharing information that you may not be on Facebook or on other platforms. But at the same time, you know, what we need to do is we need to make sure that we're protecting individuals and their ability to make choices around how their information is actually being used at a personal level.

Well, I think, look, the data is definitely the issue of our time right now in terms of privacy and protections. But at the same time, if you look at the government, one of the reasons, by the way, it's so inefficient in terms of information and service, a lot of people ask, well, why can't the United States government be much more like an Amazon in terms of, you know, which books I like, which services I need? One of the reasons is because there are statutes in place today that prevent the sharing of the information we're talking about. So there are a number of laws already on the books, whether it's with health care data or with tax data. So for example, the IRS can't really share data with other agencies, and they're being effectively translated to the digital realm. Well, it's because it can't share with other agencies, what ends up happening is when you deal with your government online, it's not as good an experience as we should have.

Exactly. But there's nothing wrong with that, right? I think from the very beginning, from the founding of this nation, we've always had a healthy mistrust of a strong central government. Anything in the digital world, that becomes even more true.

So let me start with giving you a data point on the state of government when we first came in. Over the course of the last decade, the United States government went from 432 data centers to over 2,000 data centers in about 10 years. The average utilization of servers in terms of CPU cycles was under 27%. The average utilization of storage was under 40%. And each agency had one of their own, right? But they weren't really using it. Between agencies, bureaus would have their own. And then what ended up happening is you basically look at this and you ask yourself, well, how much are we really spending on infrastructure? Turns out we spent 24 billion dollars every year on IT infrastructure. And I looked at what was happening with the most sort of leading-edge companies and startups around the country, and all of the startups are actually moving towards cloud. So if you went into your board of directors or a venture capitalist and said, hey, I need, you know, five million dollars and six months to hire systems administrators to run my email, to build my website, I need to install my own ERP system, and I need to make sure that I've got an army of people who are integrating all the systems, you'd get laughed out of the room, right? Because the technology has made my life right. And what happened is, historically, if you think about it, is that in the 1960s, the greatest innovation in technology were happening in government. In the 1980s, that innovation moved to the enterprise. And in 2005, something big happened, all the innovations shifted to the consumer side. And that is one of the reasons I instituted a cloud-first policy, because the US government could save billions of dollars by moving away from building these redundant, duplicative systems and this infrastructure that was costing taxpayers billions. Just to give you a data point on some of the results, in something as simple as email, USDA and GSA were able to save 40 million dollars by moving to the cloud.

Well, that means that every agency in the US government has to evaluate cloud computing before they award any contract, as an alternative, as an option.

As a mandate, because we would zero those out in the budget.

Well, but then they would have to come sit down with me at the budget meetings, and we would start zeroing them out.

It meant that you would consider, based on your security requirements, if you're in the national security space, they had data that had to be in government-owned, government-operated systems, that there will be a government-run cloud. The DoD has a government-run cloud. So instead of having various agencies going out and building their own, but if you are, for example, looking at collaboration technologies, there's no need to go out there and build a government-owned, operated solution. There are a number of third parties in the private side that were doing a much better job.

This started actually in September of 2009, and I still remember, this is at the NASA Ames Research Center is where we started.

Well, every agency has moved about three systems they're moving to the cloud itself. Sorry, I mentioned USDA, GSA actually put out a report of the state of public sector cloud computing, and you can look online if you go to CIO.gov, which systems moved by agencies.

Well, I identified about 20 billion dollars I could move to the cloud. I'd have to check what the latest data is, but of course there's pushback because you have the old guard that wants to preserve the current status.

I think it's about right. You have government folks who are afraid of change, and they have held on to these beliefs because they grew up in an IT environment where it was all about owning, managing, running your own systems. And you've got a set of vendors, frankly, the way they make money is through a land-and-expand strategy, which is that they start off by putting a consultant on board, and the consultant's job is to go out there and multiply as many consultants as possible in their client to make sure that they're billing at hundreds of dollars an hour. And guess what happens? The government gets IT projects like the one at the Department of Defense, after ten years and 850 million dollars in taxpayer money, we had to kill it because it didn't work.

No, no, I said it while I was in the White House.

So one of the things that I have talked about at length and have proposed at the World Economic Forum in January, that we get together, governments, to actually talk about the data sovereignty issues, which is a real issue. What I mean by data sovereignty is that if you're a government agency, you have to figure out where that data resides, whether it's in the continental United States or globally. But a bigger problem that I see in terms of the future of commerce is that if every country in the world starts saying that the data must only reside in our country, all of a sudden it's going to have a chilling effect on commerce. Also, we've got to recognize that the US government isn't just in the United States. The State Department is all over the world, the Department of Defense is all over the world.

If you think about it today, already there are over 4,700 systems that are not in the hands of the federal government. They've been basically being run by private contractors. If you look at our voice systems right now, 70% of the telco infrastructure that the US government uses, that's being run by Verizon, AT&T. So it's actually not that different. I think this comes down to the point where folks who want to prevent this move to the cloud, because it has the ability to save the government billions of dollars and disrupt the current business models, will raise this point. But the reality is that government is already engaged in those types of contracts.

Well, I think it's going to depend system by system, right? And you're going to have to weigh what are the consequences of the data that the government has on those systems. So in the case of national security, you would make a decision that it would have to be in a government-owned and operated cloud. In a case where you actually have a lot of sensitive data, you may do a hybrid approach. But if you look at something as simple as, you know, the example I used on email and so forth, absolutely makes no sense.

I think it's both of them, depending on the talent that's actually running those systems. So the argument that I've made is, in many cases, the government would be better off by moving to some of these commercial providers because it can't compete for the talent when it comes to security.

I mean, that would be the biggest mistake, because what we've been able to prove is that by shining light, by building these platforms, by engaging an ecosystem of third parties, that not only is it good government, but it saves billions and billions of dollars. And I'm watching closely as these battles go back and forth.

Well, I moved, when I first came on, I had talked to the President and I committed to two years. So two and a half years later, one of the things that, you know, as I reflected moving forward, I was coming in at 4:30 in the morning, leaving around 10:00 p.m., have one day after the other after the other. And it was probably the best job I've ever had in my whole life. But I also recognized, you know, that I was moving forward, that it was time to move on.

Great question, thank you, Rich. Well, from my perspective, in terms of WikiLeaks, it was classified information that was taken from the government, and it has had very negative implications. So I don't necessarily view this as, you know, it's obviously not an area where I've spent a lot of energy, but I mean, it was classified information that was taken from the government, it had massive negative ramifications to our national security.

Well, it's very, very difficult in any ecosystem to try to predict, you know, what one individual will end up doing. I think there are inherent risks. That's why if you look at our national security apparatus, they go through extensive sort of background checks and so forth. But if you look at the WikiLeaks incident, it shouldn't have happened, right? In terms of from a technology perspective, that there needed to be more controls in place that were just not there.

Right. And that's not, for a classified system, that's unacceptable.

Well, no, I mean, you can engineer systems that would prevent that.

I think that if you look at not just the data.gov initiative, but also if you look at what happened with the White House Secret Service logs that have been released, I think it will be very difficult for someone to make an argument to go back when it comes to transparency. But crazier things have happened in politics. So, but my view is that it's a one-way street. And I believe that if this community, the more it's used as a platform, the more likely that it won't be terminated.

It's absolutely not partisan. What I can tell you is in the number of meetings that I've even had with chairmen, I saw on the Oversight Committee that he's been very supportive and helped fight actually for the funding both on the House side and also on the Senate side. Senator Tom Coburn has been a huge proponent of a more transparent and open government.

Well, the President made the decision to make public the Secret Service logs in terms of visitors who are coming to the White House. So the idea is it would be fairly awkward for any future administration to come in and say we're not going to make any of this information available. And they've been used, by the way, to write a number of stories on what's happening within the White House.

I think they're apples and oranges.

And that's exactly the point I was making around this focus on data sovereignty. The government case, just so I'm clear, one of the things that we did is we made sure as these contracts were being signed, whether it was a move to Google or Amazon, that they actually stepped forward and built a gov solution so that data would reside only in the United States, because a lot of the data sovereignty issues haven't been solved.

Oh, absolutely. So a lot of these vendors, and look, why wouldn't they be? They recognize there's an 80 billion dollar opportunity to fundamentally disrupt what's going on in federal IT.

But I think that's the same problem you have right now, which is a majority of the systems are being run by third parties when it comes to federal IT. I think the bigger question for us is as we think about cyber warfare and as nation-states build and militarize sort of capabilities on the offense side, and the issues of what privacy means when we talk about information sharing and data.

Well, my parting words would be, you know, to form sort of a more perfect republic, that the work that you do is so important. And that when, whether it's institutions like the United States government or cities and states put out that data, use it, help build this future, and hold your government accountable for how it operates. And also, you are at such an interesting point in history in terms of processing power and storage that's so much cheaper than it's ever been before, that you can deliver information to people that was just structurally impossible before. So take advantage of where you are and the access that you have to a lot of this information.

Thank you very much, Tom. Thank you.