DARPA Chief Brags About Their Cognitive Computer Programs

Posted: September 7, 2008 in 2007, Articles
Tags: ,

Noah Schatman | Wired News | Februrary 20, 2007

Tony Tether has headed up the Pentagon’s way-out research arm, Darpa, since 2001. That makes him the longest-serving director in the agency’s nearly 50-year history. He sat down with me for an interview in his office, on the top floor of a blandly menacing Northern Virginia office building, last December. For my story in the March issue of Wired (online next Tuesday), Tether and I talked about everything from bio-terrorists to zombie rodents to thinking machines to the golf courses in Iraq.  Here’s the transcript.

Noah Shachtman: So, again, thanks for doing this. Let’s start with the big picture and talk a little bit about 9/11 — it’?s been five years now -?-? and how, obviously, that has affected defense research hugely. What do you see as Darpa’s big contributions to the war on terror? What do you think has been contributed so far, and what do you think is on the horizon that might be the most valuable?

Tony Tether: We have several efforts in use in Iraq and Afghanistan today. There’s been somewhat of a misunderstanding that when 9/11 unfolded that Darpa suddenly turned totally toward supplying things for the war. Now, of course, the war made us a great deal more interested in trying to find out what the issues and problems were over there so that we could develop programs along that line. And we have. Those programs are long-range and, for the most part, they’?re things that won’t really come to fruition for several years.

On the other hand, Darpa had started many things in the ’90s, because we’ve been looking at this global terrorist war since probably 1994. At that time we called it the transnational threat –?? you know, a threat without a country. At the same time, there was a great push to look at the way our forces were developed and move them from huge divisions, force on force, to small units of action, back to the squad… As it turned out, 2001 came and we went into war in both Afghanistan and Iraq, and, after the major conflict in Iraq, really small units became the way we were orchestrating that war. And it was probably that way from the very beginning in Afghanistan. So the technologies that we have been developing for four or five years, some of them were already ready to go.

NS: Can you give me an example?

TT: One of the major things we knew a small unit would need, especially in a city, was situational awareness. Knowing what’s on the next block — not what’s 10, 15 miles away. So we developed — we already had been developing — a small platform that we call Wasp. It was based on a multifunctional technology approach. This was in the Defense Sciences Office here. The program manager said, We really want things to be electric-driven because they’re very quiet, very efficient. Though, usually, if you put just a battery on an airplane you get only 15 minutes of flight time. But why do we have to do it that way? Why can’t we make the structure of the airplane the battery? And if we do that we should be able to get much better performance. Well, he developed a little thing called the Wasp, and in the initial version the wings were actually the battery. He found that by doing it that way, rather than 15 minutes of flight time, we were able to get an hour and a half to two hours of time. Once that happened we realized we could put sensors on it. Then we put a comlink on it and GPS.

NS: Sure, and that drains battery time.

TT: And that drains battery time. But we were still maintaining well over an hour of flight time. That little Wasp is in use in Iraq today.

NS: Do you have a sense of [how many]?, I mean, is it in use in tens, hundreds of units?

S: Well, I know it’s in use in close to 200 of them. And it’s being used by small units â?? small units that are using it, as they consider it, like a guardian angel.  And the reason they like it is because it’s very quiet. It can fly over them, and it can’t be seen because it’??s so small. And it can’t be heard because it’s electric-driven. That exactly fits into the original concept we had way back in the ’90s.

NS: And how about something that maybe isn’t on the battlefield right this second, but maybe just on the horizon?

TT: Well, we are working hard. One problem is language. We realized that we’re either going to have to teach all of our soldiers 16 different languages or come up with the technology to do so, to help them out. When 2001 came we had already been working on a Phraselator, which is a [simple,] one-way [translation] device. One-way in that it has phrases in it that in any of eight different languages —

NS: Yeah, I’ve tried it out.

TT: You’ve tried it out. Well, that’s good. Did it work for you?

NS: Yeah, it worked.

TT: Sometimes it doesn’t work for me, but people like them. For some people it really works well. But that was just sort of an interim step. What we are developing now is a two-way. And while we want to eventually get to the two-way translator, which is totally generic in all situations, what we’re doing now is coming up with two-way translators that are good for specific functions, like checkpoints. Where at a checkpoint the questions and the answers are somewhat contained. And so you can actually come up with a device that really can be a two-way translator.

NS: Really? And can capture more than just “yes, no” [questions]?

TT: Yes, that’s right. But if you were to ask, How’s your golf score? it wouldn’t be very useful.

NS: How is the golf in Iraq, by the way?

TT: I don’t know –?? they’ve got a lot of sand traps.

NS: [laughs]

TT: So that’s an example of something that will be out soon.

NS: Do you know of anything that Darpa’s working on right now that’s really game changing?

TT: Yes — our cognitive program. The cognitive program’s whole purpose in life is really to increase the tooth-to-tail ratio [military-speak for the number of combat troops to the number of support troops]. Look at how we operate –?? look at a force in Iraq or Afghanistan and count how many people are really fighters and how many people are really supporters â?? and you find that there’s a very large ratio, 20 to 1, or maybe even higher than that.

NS: Still?

TT: Yeah, still. And if you say what are those 20 doing? Well, a lot of them are doing things at the command post like keeping the computer system up, or they’re preparing [charts] or the daily brief.

NS: Yeah, PowerPoint rangers.

TT: Our cognitive program’s whole aim is to have a computer “learn you,” as opposed to you having to learn the computer. We’ve got the technology to the point where we can now apply it in Iraq to a system that we also developed called CPOF, Command Post of the Future. It is a distributed command and control system.

NS: It’s a piece of software that’s being applied to it?

TT: It will be just software that’s applied to it. Let me give you an example of one of the first jobs it might do. It might just keep the Command Post of the Future up, and, if it can do that, those people who are now doing that 24-by-7 won’t be needed. Another task we’re looking at having it do is a shift change â?? at the end of a shift, a person has to take some time to put together some slides, or some kind of a briefing, so that the next person who comes in will know what went on. This transition between shifts must be seamless, and this system could do it in spades. It will be able to put together what happened during the day, and, more important, it will know who’s coming in on the shift and will have learned that person’s preferences. OK? So that, to me, is really a game changer –?? if that all comes to pass.

NS: I mean, I don’t have to tell you that people have been promising cognitive computers [since..]

TT: Oh, forever, forever. Since, I guess, even science fiction back in the ’30s had computers being somewhat cognitive.

NS: People [always] thought it was just around the corner. What makes this iteration any more likely?

TT: Well, a lot of time has passed. If you look at Darpa, we’re almost 50 years old now. And if you look at Darpa’s thrust in computers, and in cognitive computers, although it’s had different names, it really started in the ’60s.  You know, how to make a computer more friendly to a human. The mouse was developed here for that very reason. So we would have spurts. We would take what the technology would allow us to do, and we would go as far as we could and then stop.

The last real attempt — other than this attempt now in the last four or five years — was in the ’80s. We had a program called the Strategic Computing Program. And that Strategic Computing Program showed a pyramid. And in that pyramid were many technologies that had to be developed –?? microelectronics to get things smaller, memories larger, computers faster. But it all was leading toward coming up with a cognitive computer, although at that time we called it artificial intelligence. We did a great job on the component technology, but the architecture for the cognitive part went down a path that was more neural nets, expert systems. And they were OK for what they did. You know, if you built yourself an expert system or a neural net for a specific situation it worked quite well, but it was very fragile. If you got off of that, it crashed. It was back to the two-way translator for the checkpoint –?? don’t ask me what your golf game is like.

Well, in Darpa fashion, we stopped in the late ’80s or early ’90s. Since the ’90s to now, our ability to create algorithms that can reason — can more abstractly reason — about a problem and come up with answers, and also remember what they did using Bayesian techniques and changing values, has really advanced. I mean, it tremendously advanced in the past — from the ’90s to, say, the early 2000s. At the same time, computers became more powerful. We’re on the verge of having computers with densities approaching a monkey’s brain, and it won’t be long before we’ll have a computer with the density of transistors, or equivalent to neurons and almost human. What we’re missing is the architecture. So it seemed like it was time. We had great advances in algorithms for reasoning and in algorithms that learned in general. At the same time, the computers, the actual intrinsic hardware, was really approaching the density of a human brain. And so it seemed like it was time to try again. We’ve had some great success. This cognitive program I told you about is actually showing that it is learning, and it is learning in a very difficult environment. This is the program Stanford Research runs for us.

NS: Which program is this?

TT: It’s PAL [Perceptive Assistant that Learns]. And we have other related programs. One major research issue has to do with learning. If you and I learn something, like baseball, and then we go play another sport, say golf, we somehow transfer that –?? we are able to transfer some of what we learned in baseball to golf. That’s what makes humans very resilient and flexible. We have some research programs trying to come up with the same technique –?? that if you had something tackle a problem and then gave it another problem, it would do better on that second problem than if it had not had the previous experience. This technology is actually going to avoid the thing that, in the 80s and ??90s, caused AI to stop –?? stop because it was too fragile to accept anything but the problem it was working on. This is what gives me confidence. I mean, we are seeing outputs now. But why did we start it in the first place? Well, it’s because of what I just said. During the ’90s, there were a lot of great advances in reasoning and Bayesian [probability analysis] processes that computer themselves had come up, and it just seemed like it was time to try it again. Now it may have y not come up with anything. But this is Darpa, right? It’s OK for us to fail.

NS: One thing that I’ve heard you talk about before, and certainly heard Donald Rumsfeld talk about, is preventing strategic surprise, preventing technological surprise. But is that really the situation we’re in now? I mean, an IED [improved explosive device] isn’t a technological surprise. Even a North Korean nuke isn’t a technological surprise. It’s a low-tech surprise.

TT: Yeah it is. And maybe that’s a technological surprise in itself. Just because something is a technological surprise doesn’t mean that it’s a high-tech surprise.

NS: OK, fair enough.

TT: And the IED is a good example of a low technological surprise when you have very innovative, smart people who can iterate very quickly. It’s not a weapon –?? it’s really mind warfare.

NS: Does Darpa’s mission change at all when it’s dealing with a low technological surprise as opposed to a high technological surprise?

TT: No. A lot of people think that, when we look at an effort that, unless it’s going to take us 20 years to do it we’re not interested. When we look at ideas and efforts, we look to see what the impact would be if something could be done. And if it takes 20 years, that’s fine. But if it takes a year, that’s fine too. So we evaluate more by the impact of the idea than we do by the length of time it happens to take to do it.

NS: Right.

S: One area that we really are concerned with — quite frankly, I’m a little uncertain about it, so I won’t go into any details –?? is quantum computing. Quantum computing is where you create a computer that uses the fact that you can have photons or something coherently coupled —

NS: Sure, encryptions.

TT: You can get great, great parallel processing. That is something that, if somebody else got it before us, would be a great technological surprise. And so we’re looking into that.

NS: And that concerns you more than biological [weapons] development or –??

TT: No, no. It’s equal. The biological, I think, is a little bit more worrisome because it’s more potentially near-term. But the impact of the quantum computer, if it can be done, will be really, really revolutionary.

NS: But isn’t it a little bit ironic that Darpa is funding BBN [Technologies — one of the original developers of the Internet’s precursor, Arpanet] to do quantum computing? So, aren’t you in some senses bringing about the thing that you’re scared of?

TT: I know, that’s always a worry, isn’t it. And, in some cases, obviously when we are worried about a technology that we don’t want to teach the world how to do, as we’re learning how to do it, well, we put controls on it.

NS: But doesn’?t that lead to a conundrum? I’m sure you’d agree that the best science is done out in the open, right?

TT: Not always.

NS: Not always?

TT: No. I mean, I think that’s the legend. But I have not found that to necessarily be the case. The best science is done when you get the best people together. That doesn’t have to be in the open. What you do have to do is gather a large enough population of people with different disciplines in order to make progress. And whether you do that open source or by having a very tightly knit project, I’ve not seen –?? I can give you a fairly near-term example, and that is stealth. Stealth was very closed, a very secure program. Great advances were made. Lots of science. Materials science actually made great leaps and bounds. And it was a very, very closed discipline for a long time. Actually, it still is closed.

NS: Right, parts of it.

TT: Parts of it are still closed. We deal with industry, and industry has learned over the years that it could have a very secure project, and, at the same time, at least on a one-way diode, get information from outside the program in the disciplines necessary to solve the project.

NS: But a one-way capture.

TT: It’s typically a one-way, yeah. It has to be almost a one-way. But you set up your project in the beginning to take all of that into account. That’s why I have not seen a real problem [somewhere].

NS: Because, I mean, I’m sure this is not going to come as news to you, but there are definitely people in the defense research community that are like, “Oh, my God, Darpa –?? I tried to get information out of them, and it’s very hard.”

TT: It’s very hard. It’s very hard. And, quite frankly, much of the problem is that they tend to use it in a way that they either overstate it or understate it.

NS: What do you mean?

TT: Either they overstate what we’re doing in the sense of “??my God.” We say, hey, we’re doing this. And then they try to extrapolate what we could be doing with it. Or, we say, we’re doing this, and we’re having a problem, or something like that. And then they focus on the problems. It’s just not necessary.

NS: Let’s talk about the biology. What I’ve heard is that, in a lot of ways, under your tenure is when Darpa really started to take a big interest in biology.

TT: Not really.

NS: Not really?

TT: No. The biology program was going strong when I got here. It started at least four years before. The transnational threat – which I mentioned earlier -? became the biological threat. Just think, what is something that people who don’t have a big infrastructure could do that would be very dramatic. Well, biologic threats come to mind. You don’t need a big factory, right? And so Darpa started programs in the ’90s.

The most famous one in the ’90s was “??One Drug, Many Bugs.”? Basically, that was started here?? because we had a problem. The bugs we were worried about, such as Ebola and some of the others, were ones that the major pharmaceuticals were never going to tackle, because there just wasn’t a business there. However, if we could find a technology with which we could create a drug to take care of not only Ebola but also the common cold, then we could have our cake and eat it, too, because the big pharmaceuticals would go make that drug, and then it would be available for the ones that our forces might see. That has transitioned. Back when we first started thinking “??One Drug, Many Bugs”? everybody said absolutely not. But now if you get an antibiotic, doctors will prescribe broad-spectrum antibiotics. That’s one drug, many bugs. And that really started back then.

Now, on the other hand, when we saw we were moving into using small units of action, we realized that the medical ability wasn’t going to be there. If you had a squad of six people going in, you couldn’t have a couple of medics with them, you know, with all of the trappings –?? so we started developing things that would allow people to take care of themselves.

NS: OK, I understand that there may have been some biology programs going on before you got here. But certainly there’s been many more —

TT: Oh, yeah. I kept it going. I kept it going, yeah.

NS: And that it’s really expanded. There’s been a fairly big expansion in terms of soldier self-care, like you were mentioning.

TT: I think it was natural. As time went on we found more things we could do. Yes, I agree that there’s been an expansion.

NS And I guess I was hoping you could comment just a little on how it is different. Is there a different philosophy or a different approach when you’re dealing in biology and living systems, rather than the sorts of gadgets and computers and algorithms that Darpa is a little more used to?

TT: When you’re dealing with things that will have to be tested on living things — animals and eventually humans — you’re much more cautious. We spend a lot of money on creating IRBs. Institutional review boards are boards made of people who are independent of that institution or company who really review these things to make sure that humans aren’t being — you know, that it’s safe. And they go over the protocols of what is to be done. And then people look at that. And you can only do those things. They approve only certain things. So, when you deal with biology, you just don’t take the gun out in the back and shoot it, you know.

NS: You think that slows things up?

TT: Yes, it does. But it’s OK. You need to have that. It’s a good check. It all came from violations, if you will, from back in the ’30s and the ’40s when volunteers were not really volunteers. You’re going to volunteer for this experiment.

NS: Sure, [the] Tuskegee [Syphillis Experiment], sure. Also, is there a needle that has to be threaded in terms of, we are trying to improve soldiers’ capabilities and improve biological defenses, and “Oh my God, you’re creating Frankenstein!”

TT: Yeah, I know there’s a worry about that.  But, quite frankly, we’re not doing that. There’s probably more hype on that. You know the old Army saying, “be all you can be”? That’s really what we’re doing. We’re making it possible for people to be all that they can be, not making them be better than they can be.

In the [armed] services that train people, they take these young kids’?? bodies, and their bodies become extraordinary in strength and endurance. We’ve got the best training in the world. But it’s not any better than their body can be. What we try to do is come up with techniques that allow them to maintain that level of proficiency when they go into situations that are not as clinical as the training itself.

NS: Are any of those programs looking really promising?

TT: One in particular, but it’??s one that might surprise you. Have you heard about our cooling glove?

NS: I have, yes.

TT: Quite frankly, it’??s a simple idea. When you’??re training, the reason you can’t get everything your body has, is because you come up against the thermal wall. So, the cooling glove allows somebody, when they’re training and really, really in shape, to train longer and thereby get better. But it wasn’t making them more than they can be. It was really making them what they could be. It wasn’t changing any genes or anything like that.

It has really worked out, in fact, to the point where the San Francisco 49ers are using them. But it’s in Iraq now — not because people are training over there, but because in Iraq heat exhaustion is a real thing. Nature, over our evolution, has made it so that the pores of our hands and the pores on the soles of our feet are really the faucets for getting heat out and heat in. And the simple little idea of drawing a little bit of negative pressure on your hand to keep these pores open, is really tremendous. Now it’s both ways. The soldiers in Iraq are using it to cool their body down, and the SEALs are using it to get heat into their body.

NS: The cooling glove is fairly near-term. Is there anything a little longer term that’s game-changing?

TT: There’s one that’s really ? over the horizon, the fact that it goes back to the testing part, too. There’s always been this notion of the “??golden hour.”? If you can get somebody to good medical care, where they can be given blood and all of that stuff within an hour, the survival is greatly enhanced. Well, when you have these small units out there, and especially in places like Afghanistan and Iraq, it’s really six hours before someone can be evacuated. So a person with great blood loss, which is what happens in wars, will die.

We had a guy with an idea. He said, why is it that people die with 60% of their blood gone? Is there any reason for that? What happens, what is going on? So we started an effort. We went out in our normal way and challenged the community. Look, here’s the problem: Is there any way we can keep a person alive with 60% blood loss for six hours as opposed to an hour? And we have found two amazing techniques.

NS: Tell me about them.

TT: These are techniques that could be given by a medical on the spot. One is giving somebody a shot of estrogen. I know that’s sort of strange, isn’t it? And it’s the typical Darpa fashion. Somebody said, “You know, women in general survive blood loss better than men.” And what is one of the chemical differences between a male and a female?  Estrogen. Think about what had to happen in order for us to have survived for all this time and evolve. Childbirth is a very bloody process.  If a woman had a baby 5,000 years ago, and if she couldn’t handle that blood loss and take that baby and go to a safe spot and all the rest of it, we would have gone away. And if you look at what happens with a woman at childbirth, she gets a big shot of estrogen from her own body. So we tried it. By the way, the one difference between Darpa and the National Institutes of Health, for example, is that Darpa will take a bet on an idea to go get the data to see if the idea is worthwhile, whereas at the NIH you typically need to have the data before they give you the effort to get the data. I’m sure you’ve heard those stories.

NS: Yeah.

TT: So we went out and gave this fellow a contract. And, by God, with a shot of estrogen — the control group is made of rats, mind you —

NS: Yeah, I’m assuming.

TT: The IRB on rats is a little bit easier. [laughs] First we bled all of them 60 percent. The rats that didn’t have the estrogen all died within three hours. Of those that were given a shot of estrogen â?? which, by the way, is very safe â?? right after 60 percent blood loss, 75 percent of them were living after six hours. Now that is not as good as the next technique. One fellow had another idea. When you have severe blood loss, your cells pass electrons. And when you have full blood, this electron passage is somewhat tampered. The doctors will call these the ions. You will take vitamins to try to get rid of —

NS: Yeah, free radicals.

TT: Yeah, get rid of these free radicals. Well, these electrons cause these free radicals. But when you get 60 percent blood loss, these electron passages still go up. And, he said, you know, it’s probably the electrons that are causing the people to die because they are really destroying tissue, and there’s nothing there to stop them, because you’ve got only 40 percent of the blood you had. He found that if you gave a small amount of hydrogen sulfide — this is a poison –??

NS: Oh, right, this is [Fred Hutchinson Cancer Research Center biochemist] Mark Roth, right?

TT: Yeah. If you give a small amount of hydrogen sulfide, it inhibits the electron generation. So his rats, OK, you bleed them 60 percent. The control group all dead within three hours. The group that you just give a little bit of hydrogen sulfide to, after 10 hours, 90 percent of them are still living. And, by the way, in both of these cases, to resuscitate them you just give them water. I mean you don’t have to even give them blood. You just give them water.

NS: It’s amazing. And also they’re not hibernating — I mean these aren’t hibernators.

TT: They’re not, no.

NS: It is really amazing.

TT: Yes, it is amazing. That’s a typical Darpa type of thing. Now, did we make somebody more than they can be because we saved somebody from dying? Of course, in the estrogen, who knows what the impact of giving a guy a shot of estrogen is, but you know.

NS: Who did the estrogen research?

TT: You know, I don’t remember. [Turns out it’s University of Alabama at Birmingham surgery professor Irshad Chaudry.] We had six or seven efforts. These two are really panning out. But we have a long way to go. We’re going to go into pigs. Oh, the IRB. But, quite frankly, in something like this for which you can’t ever really get a human, you say, OK, get the control group. [laughs]

NS: OK, bleed about 60 percent?  Yeah, I won’t volunteer for that one.

Let’s change gears a little bit and talk about the challenges.

TT: Challenges?

NS: You know, the prizes [– like Darpa’s $2 million all-robot rally, the Grand Challenge.].

TT: Oh yeah, OK.  [Under Secretary of Defense for Acquisition, Technology and Logistics] Ken Krieg gave us approval just a couple of days ago to have the cash prizes on the [next Grand] Challenge to be $2 million for first place, $1 million for second, and $500,000 for third place. Before, we had a winner-take-all. Now we’re going to have three places. But these are for the people who do 60 miles in six hours, so the necessary condition is you’ve got to do that, and for those people who do do it, the fastest one will get $2 million.

NS: Got it. So these are for the people who got some money up front [some of the current Challenge contestants got seed money from Darpa] or —

TT: You would think so. But, you know what, one person we offered a contract to turned us down.

NS: Is that right?

TT: Yeah, they turned us down because they got a big aerospace firm as a sponsor. And the big aerospace firm didn’t want to have anything to do with the money.

NS: Really?

TT: Yeah. [laughs] I said, are you guys sure?

NS: This just in: “Aerospace firm turns down Defense Department contract.”? [laughs]

TT: Well, the company that won the money was a fairly small company. But they got a big sponsor who didn’t want to have a contract. So I was [laughs] OK.

NS: Do you see prizes in any way replacing the traditional Darpa grant awards? Do you see it augmenting it? Is this just something for fun? Is it something crucial, or is it central?

TT: It’s hard to find  a topic. You know, even here the only reason we had â?? that you couldn’t give contracts for. The original Grand Challenge [robot rally] was really motivated by a couple things. One, having autonomous vehicles is a valuable military capability. But we worry a lot about the [feedstock] — the fact that kids today don’t really see engineering and science as a career. And I believe that’??s because we haven’t come up with any challenges or problems to excite them to go into those areas.

NS: Nothing like the moon shot.

TT: Nothing like the moon shot, no. And the autonomous vehicle thing, we could have just given contracts probably in the beginning and not have gotten as far. But a big motivation was that everybody in this country owns a car. You can go buy these computers, you know. The sensors you can buy, and even the actuators are in the handicapped market. So anybody in the country could participate. And the only thing they needed to have was the imagination to create the secret sauce that took the inputs from the sensors and converted it to controls for the actuators and close that loop. That’s really why we did it. It was that little extra dimension that allowed it to be over the edge of just having contracts. And it was fantastic, absolutely fantastic. The response has been incredible, absolutely incredible, worldwide.

NS: Do you see that the x=”” prize=”” foundation=””>, for example, is going from space to many other areas?

TT: Yeah, they’re actually going to make a business out of it.

NS: And NASA is, of course, doing prizes now, too. Do you see Darpa doing more prizes, or do you think it’s pretty limited?

TT: I think we’ll do more prizes. But, again, there’d have to be very special reasons for doing it. Even the X Prize and the NASA thing –?? really what they’re doing is maybe getting something out of it technologically that they couldn’t have gotten out of it by just giving contracts. They are getting a whole bunch of people interested.

NS: And for not a lot of money.

TT: Yes, but the thing is, we now have tens of thousands of people interested in autonomous robots — your AI thing again — that weren’t. And now they’re thinking about solutions. And those solutions are going to apply to areas bigger than autonomous vehicles. And that was really what it was about. And, in fact, the Grand Challenge is a good example of how trying to do cognitive or AI things again is not a bad thing.

NS: Right. Again changing gears a little bit: I remember being at a DarpaTech [the agency’s conference] a couple of years ago, and you said something to the effect of how nobody has a career at Darpa; “We hire ideas, not careers.” You’re now the —

TT: I’m the longest living — not the longest living — I’m the longest-sitting Darpa director ever.

NS: You’re the last of the Mohicans, right? So how do you keep the ideas fresh? Do you think there’s a tension between those two things?

TT: Well, first of all, it’s the program managers at Darpa who are the important people. And, quite frankly, that makes us different than anyplace else in the world. People come to me from all over the world, and they look at our track record and what we’re doing, and they want to know how they can make an organization like ours. I tell them it’s simple. You just have to make sure the people don’t stay there very long.

NS: [laughs]

TT: And they don’t know how to handle that. Because it’s not like we detail. People are detailed here from the department, and then they go back there. These people come from industry and universities and give up everything, because they do become government people. And then, with no help as to where the hell they’re going and, in fact, with all of the ethics laws that come up, sometimes it’s not clear they can get a job. So these are strange people, and I’m sure you’ve met them. But they’re all motivated by the fact that they have an idea that they can’t get done anywhere else. And if that idea lines up with DOD objectives, sometimes we hire them and bring them in.

They get in here, and they know the clock is ticking. They know they probably can’t get their idea completed in the time they’re here, but they can get started. The other neat thing is, when they leave, they have started these [good] programs, but a new program manager comes in, and while he’s trying to get his idea through the process of getting money, he gets these other programs. But he has the full right to take those programs, review them, and trim the tree. So even the performers never get comfortable. And that’s just great. That’s what makes this place what it is. And he may look at that previous program, and say, hey, there’s an objective other than the original one that we can now use this for. Because he might have knowledge of something. We can do it over here, we can use it for this. That might even be more important than what the original objective was.

NS: Right. OK. All true. But you’ve been here a long time now, so how do you square that?

TT: I don’t. I never expected to be here this long. I will probably hang on until the end of the administration. The one great thing about Darpa is that nobody is there long enough to really screw it up. And there have been people who would say you know, Tether, you’re getting there, you know. [laughs] But I really would like to be here for the 50th. We are going to be 50 years old in February of 2008, and we have a whole thing starting next year all leading toward that. And I do want to be here for that. That’s, quite frankly, my reason.

NS: You’ve done a bunch [over your career]. You were here at Darpa before, you’ve done industry jobs, intel jobs. Any of those particularly prep you for this gig? Is there one you draw on a lot, or one that’s more analogous?

TT: I think the industrial experience is almost a prerequisite, because when I do things here I have a feeling for the other side. But probably the most valuable experience I had was, after I graduated from Stanford in ’69, helping start a company. There were six or seven of us in Palo Alto. We were all technical guys â?? in fact, my background is control theory. We were control theory guys, who believe that every problem can be solved by putting a loop around it. And we went out to prove that that is true. And we did it. I mean, we did it for the commercial world, and process control did it for the military world. We were lucky â?? the timing couldn’t have been better. Computers were just getting to the point where you could think in real-time control. And it really was that experience that set me up and gave me the right background for this kind of a job.

NS: I assume you’re talking about the entrepreneurial aspect?

TT: Yeah, and the fact that we were always selling something that people didn’t want. They just didn’t know that they didn’t want it. I mean, they didn’t know that they wanted it. So we had to figure out how to convince them that they wanted it. Which is what this job is all about.

NS: How is selling to people what they don’t know they want part of this job?

TT: Because that’s what we do here. We develop technologies and try to transition it to people who don’t know that they really want it. They just haven’t really thought about what they could do with this new capability. And so we have to be clever –?? I don’t mean that in a bad way –?? to be able to communicate in their language. And I think we do that quite well. Yeah, we really do that quite well.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s