CIA needs nimble tech to keep pace

The need for increased computing scalability and agility is important to most federal chief technology officers — and especially the CIA's CTO, Gus Hunt.

Hunt says cloud computing is not merely an opportunity to improve efficiency and lower costs. It can tackle computing problems in the intelligence community that the laws of physics and constraints of existing systems can’t always handle.

Hunt also said he sees the evolution of intelligence widgets and sophisticated data algorithms bringing new agility to analysts throughout the intelligence community.

He spoke about those and other developments with Editor-at-Large Wyatt Kash and will speak about them further Oct. 27 at the Virtualization, Cloud Computing and Green IT Conference, sponsored by Federal Computer Week's parent company, 1105 Government Information Group.The interview was edited for style, clarity and space.

FCW: Cloud computing has become a strategic issue for most CIOs and CTOs. How are you looking at cloud based on the CIA’s needs?

Gus Hunt: Two points, [and] we need to do both to support our mission: the massive amounts of information we have to be able to understand and deal with to do our job effectively, combined with the economic pressures that are on us from Congress and other places to be more efficient in IT. The external commercial sector has reached a curious tipping point where this advent of cloud — because it drives very, very high efficiencies in terms of managing infrastructure with high automation — also enables the capabilities we want to be able to apply to enable our mission for the future.

FCW: When and where will cloud computing services really become a functioning part of the CIA’s IT operation?

Hunt: We have an early cloud, I will use quotations on the term “pilot” that we are working with the community for the [Director of National Intelligence (DNI)]…known as I2. And we have stood up in a large-scale computing environment the necessary capabilities to share both technology and information more effectively as part of a joint activity. That environment is not truly cloud-wide because it doesn’t yet contain the dynamic in elastic components you need for real-time provisioning and the ability to take the resources that are underutilized and apply them in other places to get that job done. But it is serving as our framework for us to push into that space.

FCW: How is being part of the DNI office and the intelligence community as a whole helping or hindering what you need to accomplish at the CIA?

Hunt: Our working relationship with DNI and the IC has vastly improved. For example, our widget framework is based on iGoogle, it’s called Ozone. It’s the framework that NSA developed that we have adopted to build our new applications and the user interface functions in this framework. This enables us to share our widgets across the community.

FCW: There appears to be a general sentiment that enterprise architecture is not as important as it once was. What’s your take?

Hunt: Enterprise architecture is still very important, but I think the way in which we have done it needs to change to accommodate the way of the future. So speed and agility are now essentially everything in this new world. One of the beauties of cloud computing is that it drives very rigorous efficiencies and effectiveness measures because of the scale of what you can manage now. Think of what Amazon or Google manages with very few people running huge infrastructures to great effect.

FCW: How do you view the administration’s push to adopt social media tools, especially given the tools the intelligence community already has with A-Space and Intellipedia?

Hunt: In fact, we actually started a long time ago embedding these sorts of social sharing capabilities into the systems that we build here. These are part of the tools and services capabilities that we will be sharing out with the intelligence community through the I2 pilot.

For example, we have a thing we called common folder service. This is a mechanism that allows an analyst to create a folder on any subject — say on Iraqi weapons of mass destruction — and it allows the analyst or other users to deposit objects that they find in their searches and share that folder. Part of our folder service, though, is that it invokes security on the basis of you as an individual, not on the basis of the folder at large, so that your view of the folder and my view of the folder may be different based upon what our permissions are.

We have embedded this concept of sharing in almost everything we have done. Our analytic tool, which we called Trident, and our current work, which is known as Next Generation Trident, will take advantage of this widget-based framework.

FCW: Can you give an example of the new kinds of capabilities that you expect Trident to offer?

Hunt: We have a very cool monitor feature. It simply puts up a graph that says that over the last window of time, however you specify it, these are the number of objects that have come in matching your profile or a query that you have asked for. You can see it at a glance now that, "Oh my, something happened," because I got a spike right here.

FCW: What are the top technology issues that you find yourself needing to study?

Hunt: Clearly going to where we started back to the cloud. It has hit the hype cycle quicker than anything — it is the single most overhyped term in the history of IT, eclipsing [service-oriented architecture] by probably an order of magnitude.

I am also very interested in network virtualization. That’s a space that I think needs more maturing.

A third area is storage. I wonder whether tape is dead. When you look at the fact that I need four tapes to back up a disk drive, what does it mean, and do I have the space to do this?

And then the final technology issue is what I called the algorithm development. Algorithms that detect very weak signals in a sea of noise have matured extraordinarily thanks to companies like Google and others who are using it to target selling new things. They are just looking for you on the bases of patterns, they don’t know who you are, right? We need similar algorithms to find people on the basis of patterns or behaviors for making sure that the next Christmas Day bomber can be stopped before he can get on a plane or some other things like that.

Having storage is nice, having data is nice, having computer power is nice, but algorithm development is also essential to intelligence value.

FCW: It would seem that a critical need for the CIA is language interpretation technology.

Hunt: Machine translation is at best mediocre, and we know this is a huge gap. Things keep getting better, but they are only better on the margin. There hasn’t been a major breakthrough in this space for a long time. English is fast becoming the second language of the Internet, but you have to be careful here. Just look at the Chinese population, look at the Indian population; they don’t want to communicate in English, they want to communicate in their native languages.

I am going to be honest: I thought machine translation and…voice recognition were technologies that would have been well solved by today. I thought that many years ago, [and] these are two areas where either I was overly optimistic or way off base. But today, I don’t see a good, high-performance, high-accuracy solution emerging anytime in the next five years.