The Department of Mad Scientists: How DARPA Is Remaking Our World, from the Internet to Artificial Limbs, by science and technology journalist Michael Belfiore, takes readers behind the scenes at the Defense Advanced Research Projects Agency, which created things like GPS, stealth technology for airplanes, and real-time speech translation now being used by soldiers in Iraq. And oh yeah, the Internet. I spoke with Belfiore about his experience reporting on DARPA and what the agency is up to now.
From reading your book, I got the impression that the folks at DARPA weren't all that happy about you nosing around. But it certainly seems they could do a better job of promoting themselves. Why don't they?
I was told that about 50 percent of everything DARPA does is classified and thus, off limits to reporters. So the agency is used to working in secrecy; it's part of the culture there and almost reflexive in many ways.
On the other hand, since the program managers who run the place turn over at the rate of about 25 percent every year, the agency has an ongoing need to recruit new talent. That coupled with the agency's contracting approach to getting work done (there are no in-house laboratories) means that there is a very real need to get the word out about the agency's work. This provided me with the opening I needed to write my book. The director at the time, Tony Tether, saw my book as an aid to his recruiting efforts, both for program managers and contractors with new ideas.
When you think about some of its most notable inventions -- like the Internet -- it's easy to forget the "D" in DARPA. How constrained is the agency by the need to start its research with a military need?
Everything DARPA does is defined by a potential military need. What makes DARPA particularly interesting to me is that much of the technology it develops finds its way into general civilian use as well. Except for the classified projects, the universities and private companies serving as DARPA contractors are allowed to continue developing the technologies on their own -- in fact are encouraged to do so -- so that the solutions they find can find their way into the marketplace where they can be procured by the armed services.
How much freedom do the scientists and engineers working there have to pursue whatever catches their fancy, no matter how outlandish?
The program managers working at DARPA have a great deal of autonomy, and that's a big part of what makes the agency an exciting place for these folks to work. I met one program manager who had actually become a U.S. citizen (he was from the U.K.) so that he could work at DARPA. I found this all the more impressive given the term limits for program managers; a typical term is four to six years, max.
Program managers generally come into the agency with some burning passion, some project they want to work on that the DARPA director thinks would make a good addition to the agency's portfolio of programs. The scramjet [an experimental airplane design potentially capable of much higher speeds than traditional jets] program I cover in my book is a good example of that. Program manager Stephen Walker was already working on scramjets at the Pentagon when he got hired. DARPA gave him the opportunity to actually start bending metal and flying prototypes rather than simply working on the theoretical underpinnings for the technology.
At the agency, they refer to some problems as "DARPA-hard." What does that mean, and what kinds of things are DARPA-hard?
DARPA-hard means something that other government agencies and even private companies shun because it is viewed as being nearly impossible to pull off. DARPA-hard means taking on big risks for the chance of a big payoff. For example, scramjets have been theorized for decades. NASA got one flying for about 10 seconds back in 2004. DARPA-hard means getting a scramjet to stay lit for as long as the fuel supply holds out. Scramjets could turn out to be as transformative as the jet engine itself. Imagine being able to fly anywhere on Earth in four hours.
Another example is DARPA's Revolutionizing Prosthetics program. Army Col. Geoffrey Ling (one of only about 10 percent of program managers who also serve as active-duty military personnel) wants to restore upper-limb amputees to full function with an artificial arm that will appear, sense, move, and act like a native arm. The challenges to pulling that one off are enormous. DARPA contractors (led by the Applied Physics Lab at Johns Hopkins University) had to perfect direct brain control for robotics and they had to look for power solutions that would allow someone to wear the device all day. The project didn't meet these very ambitious goals by the 2009 deadline, but the $100 million spent greatly advanced the state of the art. One outcome is what Ling calls the "strap-and-go arm" developed by DEKA Research (of Segway fame) and now undergoing trials at the Department of Veterans Affairs.
Free-enterprise advocates would say we don't really need something like DARPA. What's the answer to that?
Free enterprise excels at advancing technology in incremental steps and at bringing that new tech to market. Government works best when it comes to pursuing far-out advances that may not pay off for years, if ever. A company trying to develop scramjets or Ling's prosthetics would quickly go bust without some other very lucrative sources of funding. ARPANET went online in 1969. We didn't get the World Wide Web until the mid-1990s. What company can wait that long for a profitable product?
On the other hand, it makes no sense for a government program to continue to run a technology after its feasibility has been proved.
I saw this firsthand during and after DARPA's Urban Challenge race for autonomous vehicles in 2007. The race proved the naysayers wrong: Production automobiles could indeed be modified to drive themselves without human intervention through city streets, obeying traffic laws and navigating with GPS and onboard sensors to reach destinations in reasonable time frames.
Afterward, Tether said, "We've done it, we've proved that this is a viable technology, and now it's time for the players to take it and run with it." The likes of Volkswagen, General Motors, Continental, and other major auto companies that participated certainly agreed. We're already seeing autonomous vehicle technology in production (self-parking, adaptive cruise control, lane-keeping), and if the social barriers can be overcome, we'll see fully autonomous vehicles on our roads within the next couple of decades.
It's obviously hard to tell what the effects of a particular line of research are -- but in your time at DARPA, was there anything you saw that made you say, "That one could really be big"?
One of the most exciting areas of DARPA's research (though not nearly as flashy as some of the others) is its portfolio of energy programs. DARPA-funded researchers at the Energy and Environmental Research Center, based at the University of North Dakota, have created military-grade jet fuel out of 100 percent vegetable oil. The team used soybean oil, but the process can be applied to any triglyceride. The stuff has been tested by the Air Force as being chemically indistinguishable to JP-8.
DARPA-funded researchers at the University of Delaware are developing solar cells that are 50 percent efficient at converting solar energy into electricity, an unheard of level of efficiency in commercial systems, which top out at about 20 percent. Both technologies, if brought to market, could transform our energy infrastructure. Program manager Douglas Kirkpatrick envisions giving soldiers the ability to generate all the power they need from the environment around them. But imagine a homeowner being able to do so as well!
And a related question: What's the strangest research project DARPA is currently working on?
As I said, I only got a glimpse of perhaps half of what the agency is up to. But certainly one of the most out-there projects I learned about is a project to program actual, living insects into miniature unmanned aerial vehicles. Researchers are doing this by implanting insect larvae with microprocessors and sensors that become imbedded in the adult insects. The researchers can then remotely control the insects with a view toward operating them as mini-spy cams. Kind of gives the phrase "fly on the wall" spying a whole new meaning, don't you think?
At the end of your book, you say that what DARPA needs is "benign neglect." Can you explain?
In many ways, DARPA is an anomaly within the world's biggest bureaucracy, the U.S. Department of Defense. It operates best with minimal oversight on a relative shoestring ($3 billion a year, or about the cost of one and a half B-2 bombers). It is a model of efficiency, somehow embodying the American "can do" spirit in a government setting. It very nearly didn't survive its birth, with the various branches of the military seeing it as a threat to their own autonomy. Even in more recent history, the agency has come under fire for operating too loosely, as with the uproar over the Total Information Awareness program Tether got started after September 11. DARPA has succeeded these last 50 years by doing its thing quietly, poking its head up only when necessary. I say it ain't broke, so let's not fix it.
