Q&A: DNI Chief Scientist Eric Haseltine
Eric Haseltine's rapid-fire delivery reminds one of a supercomputer uploading masses of data. The thin, wiry chief scientist for the director of national intelligence has quickly earned a reputation as both a controversial maverick and a source of inspiration within the U.S. intelligence community. Haseltine's official title is associate director for science and technology for the Office of the Director for National Intelligence (ODNI), where he is responsible for ensuring that U.S. intelligence stays on the cutting edge of research and development.
Haseltine bears one of the more colorful résumés at the DNI. A former director of engineering for Hughes Aircraft, he served as head of R&D at Walt Disney Imagineering, where he managed the Virtual Reality Studio and oversaw key technology initiatives for the Disney Co. After Disney, he joined the code-breaking, eavesdropping National Security Agency as R&D chief, then began work for the DNI in June last year. Haseltine has published over a hundred articles in journals such as Brain Research and Neuroscience Proceedings and has been a contributing editor to Discover magazine, where he authored a monthly column on the brain. He has a Ph.D. in physiological psychology and holds 12 patents in the fields of laser projection, optics, animation tools, and special effects.
U.S. News's David E. Kaplan and Kevin Whitelaw sat down with Haseltine on Oct. 12, 2006. The interview took place at his modest, unadorned office at DNI headquarters.
Extended excerpts from that conversation:
What exactly do you do here?
My official title is associate director of national intelligence, science, and technology. You can think of me as the CTO [chief technology officer] of the intelligence community. The reason we need that function is, think about what technology needs to get done and would only get done by somebody being accountable for the whole enterprise. The first things that immediately come to mind are making sure that there isn't any duplication, making sure that when technology is developed in one agency that it connects to that in another, so that you can take chocolate and peanut butter and put it together and make a Reese's Cup.
The classic example of that is, imagine that you were in a dark room and there was some threat in that dark room you wanted to find. You'd want all your senses coherently focused on it. So that if you saw a little flash of light, you'd also want to be able to hear, to see if something came from that direction. In a sense, the intelligence community agencies are like different sensors. We have some that look at imagery, some that look at chemical stuff, some that look at [electronic] signals intelligence. And just like you in that dark room, if there's a very faint signal there, you're looking at how it looks, how it sounds, how it smells. You want all those faint signals to add to each other so that you know whether you got a problem over there or over there. If you don't have somebody at the top trying to figure out how to get all those things to plug together so that they add up upon each other…then you're not truly leveraging the whole thing. It's to do things that no one else would do unless somebody were held accountable.
It's not only pertinent to do the right things, you have to do them the right way, by which I mean fast! We are a lot of things in the intelligence community, but we are not known for our speed of innovation, and there are good reasons for that. Where you're fighting a war, to experiment with radical, new, fast stuff is somewhat difficult, so somebody has to have the ability, outside of the urgency of a war, to try new things and to speed things up. If we only do "ho-hum" stuff, in the aggregate are we really going to surprise anybody? Are we going to be surprised? Can we be as agile as some of our enemies?
What surprised you about the current job here? Was it pretty much as you imagined?
Well, what surprised me, in part, was the astonishing breadth and depth of the technology issues that we face as a community. When you think of the scope of the whole enterprise, it's broad and deep, and the dynamic range or bandwidth that one has to have to understand is jaw-dropping. You have to understand optics, radar, communications technology, nanotechnology, chemistry, biology…it's mind boggling. There's almost no technology out there that isn't hugely relevant to us. I knew conceptually that was the case, but when I started drilling down at each agency and what it was they actually did, that's the first thing that just hits you like a ton of bricks.
Everything is relevant?
Everything is relevant, everything is deep and complicated and moving fast. It's the sheer breathtaking scope of the enterprise and the technology. You know, it's kind of like when I was a kid at Boy Scout camp, they had us get our swimming merit badge in 39-degree water, and you're sitting there going "Uhhh," trying to catch your breath. That's how it kind of felt for the first few months.
Is there something that could illustrate the extraordinary depth and complexity?
Sure. We have a fellows program that we've instituted to identify the very best people and get them together to cooperate. Whenever I get a chance, I call them up and find out what they're doing, because these are the thought leaders. There is one of these individuals who has figured out how to come right up to the edge of what physics allows and do some things that just surprise the heck out of me. I didn't think they were possible. It astonished me that anyone would think to do it.
The one word that best describes it is "audacity." This person was a physicist who said, "You know, if I looked at this, not as an intelligence problem but as a physics problem–what is the absolute limit of what physics will let me do? I'm going to try to come as close to that limit as I possibly can." He has accomplished astonishing feats that are very relevant to the war on terror–relevant to finding terrorists in real time.
I'm not sure what you're talking about.
Let's go to a realm a little safer for us to talk about–astrophysics. If you look at astrophysics over the last 20 years, it's the history of taking today's noise and turning it into tomorrow's signal. How do we find extrasolar planets? We found them by looking at very subtle wobbles in stars, many light-years away, and by looking at the slight wobble in the sun, we were able to infer that there was a planet rotating around it of a certain mass. Once we knew where it was, we were able to use other sensors on it.
Things that look like noise yesterday, by very sophisticated improvements in sensors and computers, we were able to turn into a signal, so a wobble turned into a planet. So where science is going is moving the decimal point farther and farther to the right on signal to noise.
Another example is cosmic background radiation–it used to be that we heard this hiss, and now we have a complete map that tells us about the creation of the universe. He's done that sort of thing, where he's taken what would sound like a hiss in radio, just like the original cosmic background radiation, and he's turned it into a signal that tells us where a terrorist is.
That's fascinating.
To me, the audacity and the vision of someone who could do that was very gratifying. You know, at NSA [the National Security Agency], you see that all the time in the math world. These are Fields Prize-caliber mathematicians, who are inventing new mathematics and doing cutting-edge basic research in math which the very next day may be used in a code-breaking or code-making application. So there's another example of incredible depth, that there is hard-core, cutting-edge science going on in these mission agencies. And that is surprising and encouraging.
What surprised me the most? What surprised me the most is maybe I had believed some of that press about, you know, we're not that sophisticated, we've not kept pace and so forth. Lo and behold, not just at NSA, there's some astonishing work going on that's really cool.
Where is the intelligence community's science and technology brain trust? Is it at the National Labs mostly?
Well, most of it is inside the intelligence community. You know about NSA, but the other agencies all have science and technology groups that are doing research. CIA has a very large activity.
I thought the CIA's S&T section had fallen on hard times.
Well, it's true that they reorganized and did away with what they called the Office of Research & Development, but the fact is, there is still a lot of healthy R&D going on inside CIA. There's an activity there called ITIC–the Intelligence Technology Innovation Center. There's an activity there called the Advanced Technology Programs Office, and let me tell you, there's some really great stuff coming out of there.
Tell us about some of the work being done.
One of the areas we have a lot of progress yet to be made is in technology to enable HUMINT [human intelligence]. I would say that is probably our biggest and most important challenge right now. There are many dimensions in which technology is a huge force multiplier or an obstacle. For example, what is the physical form of HUMINT today? It used to be back in the cloak-and-dagger days of OSS [the World War II Office of Strategic Services] that it was hidden messages or things written on paper or tape recordings or micro-dots. It's the stuff you think of from the old spy movies. Today, it's all bits mostly. It's electronic information because almost all the information that everybody wants to collect is in electronic form today. Can't you see how technology would be of great use in getting to it and once you've gotten it to understand what you've got?
Machine translation, for example–What if we were wildly successful in collecting digital intelligence in HUMINT form. Let's suppose all of that is in a foreign language, which generally it is going to be–How do we pile through all that and see what's useful? As you've probably heard, we have challenges in terms of the number of linguists that we have, so we have to come up with ways of force-multiplying the few linguists that we have.
What other projects are you excited about?
Sometimes our problem isn't that we don't collect information–the problem is the opposite: We collect an incredible amount of information. Then the problem becomes, how do you find the most important stuff and act on it quickly? We have lots of information about terrorists–how do we quickly get to understanding of that information that leads to targeting of them or dealing with them?
What our office has done is go out to the community and find where there were embryonic or nascent development efforts that show promise for attacking that problem. We poured resources and attention on them and have produced some results. I'll give you some examples. There are some people, and I can't say the agency, but they have been doing a lot of work on finding new targets based not on their connection to a known target or a known place but strictly based on their behavior. That is, over a long period of time, when we find bad guys, we then retrospectively go back and say, "How do these people behave?"
If you were a terrorist and you had theories about how good America was at collecting against what you were doing…you would try to avoid them, right? I don't think I'm revealing a whole lot to say that because of leaks and other factors, there is some knowledge about what we can do and what we do do. This does get out there, and people pay attention to it. So when a terrorist tries to avoid us by behaving in ways they believe will get around us, that in and of itself can leave a signature. That's the sort of thing that we've been doing.
Another example is we've been working closely with the military on a number of fronts. I have personally been out to Iraq, to the front three times since I've taken this job, and I work closely with the people fighting the war on terror there. They gave us a top priority with respect to the targeting of insurgents in the way that I'm just talking about.
Is it something like a cop eyeing a drunk driver who's going too slow?
I've got to be careful here in what I say, because we certainly don't want to tip our hand. You know, we were criticized for not having an imagination. And we took that seriously. In a sense with this technology, we're talking about how terrorists tend to behave in certain ways. Oh, by the way, they keep changing their behavior. They're very adaptive. So, anything that you learn today isn't necessarily relevant tomorrow.
Your colleagues have told us about Argus–this new program you have for detecting disease outbreaks. Can you tell us more about it?
Yes, it came out of ITIC. It started a few years ago. Argus is an example of one of these things that we looked around for, cool things that answered the big problems that we had identified "Day 1." We spent a lot of energy getting them more money, getting them hooked up with customers, accelerating. And through our efforts, we were able to go from a small part of the world that it was aimed at to the entire world in just four months.
How does it work?
What we do is train the computer to harvest open-source material, unclassified material. It looks at the statistics of what is normal in a particular dimension, in this case references to disease and the things that disease cause in societies, that cause social disruption. When there is a change in that rate for that particular type of information, the computer takes note of it. We have been very successful with this technology in predicting well ahead of when the World Health Organization knows about things. We know that something has happened before it's reported in the public press. NIH [National Institutes of Health], in fact, makes great use of this information.
When we found it, Argus was a program in ITIC that had been going on for a number of years on biodefense and warning against biothreats. Bio is a very tough problem and by bio, I mean not only biowarfare but also nature. In a sense, nature is conspiring to get us, too, with things like SARS and avian flu, and at the end of the day our job it to keep America safe, whether the threat is nature or a human being.
So, the leaders of this R&D effort had a central insight about the Worldwide Web and the explosion of media–that the global information network is, in a sense, the world's EKG. People through blogs, through websites, through electronic newspapers, through magazines that find their way onto electronic media and digital libraries–all open source–information flows from the world into this electronic information gathering and then it is distributed back out again. We now have a nervous system for the globe, and it's called "the global information network." Essentially, their insight was to go out there on this nervous system and see what it's telling them about what's happening in the world.
So they tried it and it worked. You know, in just a few months we got them the extra money, and then a few months after that they put points on the board and they made this thing go worldwide. I sleep a little easier at night knowing that Argus is out there.
But how does monitoring spikes or anomalies in the world's information grid serve as an indicator of a bioevent?
Well, if you think about a disease outbreak and how it would impact a local community, if a lot of people were sick, would you have reports of absenteeism? If a lot of the medications were no longer on the shelf because everyone was buying antihistamines? It turns out that human societies are like a pond where an event will cause ripples. If you think about your neighborhood and if 30 percent of the people there got sick, do you think it wouldn't show up in some electronic forum, like MySpace or a blog or a website or a local newspaper in electronic edition? Those things show up… and it's immediate. What we found was the coming events cast their cybershadows.
How has Argus been used?
It was used on avian flu, but it's designed to look at social disruption. Initially, we are looking at disease. But I think you can see from my description that anything that disrupts the social fabric is potentially going to be….
Argus was threatened when you found it?
Argus for whatever reason was threatened. So we came in and protected it, stopped it from being cut, and then we started promoting it and getting it more resources.
I want to come back to how Argus is an example of our top priority, which is "speed." Instead of just focusing on fixing what is broken, you focus on nourishing what is already healthy. You get farther faster by doing that. So that's part of our strategy. The two other pillars of our strategy are surprise and synergy. So, three S's.
Surprise means, it really is important for us to do things that are going to surprise our adversaries, that they're not going to expect. I don't think I'm going to reveal any great surprise that we would like to know more about WMD [weapons of mass destruction] proliferation than, in fact, we do. One of the reasons we have trouble is, there's so much known about what we can do that we're not necessarily surprising people as much as we need to. We've got to fix that.
By the way, we have to avoid being surprised. You know, that is Job 1 in the intelligence community. Technology has a huge say in whether or not we get surprised.
What can you do that you couldn't while at NSA?
We're building teams. We have come up with four or five mission-focused technology challenge teams, and I call them the "Take that hill" teams. We have one on IEDs [improvised explosive devices], we have one on counterterrorism, and we have one each for big nation-state problem actors that, let's say, are at the top of the headlines.
And we've gone out to the science and technology labs and we said, "Hey, can we do some experimental new things against real mission problems today?" And, of course, we've had some successes in the IED front, in the war on terror
Do you have enough resources to do the job? Are we playing catch-up?
Well, we have pointed out that there could be a lot of "bang for the buck" at the margin for new investments in S&T, but you're not going to get me to complain about resources at this point. I think the first year has really been about, "Let's make the most of what we've already got." The fact of the matter is, I came from private industry–the Walt Disney Co.–and I know how much money I had there, versus how much money I can influence here. … [Laughter] I think that we can make a tremendous amount of headway just being more intelligent and using what we've already got.
You talked about criticism that the intelligence community lacks imagination. Do you still encounter that as a problem? Are we on the cusp where we need to make some paradigm shifts in how we even conceive of intelligence, how we think about collection?
I think we have a lot of room for improvement in that arena. The way I imagine the historical record of the intelligence community–back in the early '60s, we were very fast, we put the U-2 together in under a year, Corona went up very quickly, the Glomar Explorer. … these programs were measured in months rather than in decades. I do believe that we've become somewhat addicted to those big-ticket, decade-long, acquisition programs. What worries me most is the lack of agility in the system. I wouldn't say "imagination" as much as I would say "novelty," because some of the big-ticket items that take a decade and many billions of dollars are quite ingenious.
But we don't get points for being ingenious. We get points for finding terrorists and keeping America safe. And here is my vision, unfortunately, of what is more the rule than the exception, that we have this big rusty gear of innovation that creaks along, where we have a much faster gear winding it a zillion RPM next to it, and the teeth of the gears are not meshing. We are very slow, cumbersome, and take a long time to get stuff out there, which ultimately doesn't necessarily surprise anybody. So, the main transformation I would like to see is for us to get as agile or more agile than our adversaries. You know, when we do things to counter IEDs, they react very quickly. We've got to get out of that cycle. You know, we are not anymore faced only with big nation-states that are big bureaucracies like we are. We're faced with people who, if they want to go out and get technology, they just go buy it.
Let me give you an example of something that we're toying with–it's the idea of disposable technology. One of the reasons things take so long is we have to plan for their supportability: Who's going to operate it, who's going to maintain it, what about spares, what about upgrades, how are we going to get things fixed out in the field…? And what happens is that creates a lot of inertia in the system.
It's good inertia in the sense that it gives you stability once you have a weapons system or a collection system, but it's bad in that, what if technology has passed you up? Moore's Law says that technology is going to double every 18 months. Does the capability of the intelligence community double every 18 months? No! And that's the problem. Even worse, some of our adversaries are on that much-faster curve. The terrorists are extremely innovative in the use of the Internet, not only for the way they communicate and coordinate but also how they get out their information messages, their propaganda. Like one of the generals in Iraq said–he viewed IEDs as a media event. They take pictures of them, they get it out there and it helps recruit, it helps sustain their momentum, and they understand that.
So if I have a single concern, it comes back to my overwhelming focus on speed–we must, must, must, get much more agile! And I'm not here to tell you that we're anywhere close to being where we need to be, and that's why I keep harping on that as an issue.
What is disposable technology?
Instead of having to plan for something's life cycle over 15 years or 20 years or 30 years, you say "No!" we're going to adopt a technology and we're going to get rid of it in two years. If you want to innovate fast, you have to obsolete fast. Because everything that you have on the shelf occupies space that something else can't take, so you have to constantly get rid of stuff.
I think that's an interesting metaphor for what's happened with the terrorists. These are bright, tech-savvy, young men, who really understand this stuff, doing very smart things with technology. They don't have to hold on to technology for 15 years. When there's something cooler, they toss out what they have and then get the next best thing.
I think there are ways, if we get really creative, at doing this. What I noticed over in Iraq when I went out on conveys is people used some stuff they just went and bought in the PX. They used these push-to-talk radios, they did all kinds of stuff they weren't supposed to do. Well, you know, I think we need to take a hard look at that.
We have to have lots of arrows in the quiver. We need to have things that surprise our adversaries. And sometimes surprise is only achieved by speed and agility.
What are you doing about information overload in the community?
At every level of the intelligence enterprise, whether it's collection or analysis or distribution or consumption of intelligence, at every stage of that pipeline, you have way more data than any stage can deal with. We inundate our customers with way more stuff than they can digest. How do we give them stuff in a way that they know what the good stuff is? That probably is a problem that will be with us forever. The more successful we are at collecting against hard targets, the more we're going to have that problem of what do we now do with the stuff we've collected.
There has been some progress made in this area, like at NSA. Their Knowledge Discovery Group has worked on what they call intelligence value estimation–trying to get computers to have deep semantic understanding of what's inside a document. It's almost as if a human were looking at it and said, "What does this thing really mean, and does it matter for this particular query?" That's where some of the "out there" research is going.
What about the growing use of wikis and blogs in the intelligence community? These seem like tailor-made tools.
We are using wikis, we are using blogs, we are using chat, we're using instant messaging, we're using VTC [video teleconferencing]. … But our problem is vastly more complicated than anyone else's, so we have to be very creative in coming up with better stuff and not following but leading.
These are not characteristics I associate with a $40 billion bureaucracy. How's the reception you're getting? Are people listening?
I have gotten nothing but magnificent support from top to bottom (at DNI) for the kinds of turbulence I've been creating. I think that's a very salient example of what's different about DNI and what our value-added is–that we are not wedded to the old way. We're held accountable for reforming the place, which means breaking some eggs. However, we don't want to irritate people just to do it. We pick our battles carefully.