Date Published: September 19, 2010
In writing The Department of Mad Scientists: How DARPA is Remaking Our World, from the Internet to Artificial Limbs, Michael Belfiore received rare cooperation from the secretive avant-garde technology and science wing of the US military. DARPA (Defense Advanced Research Projects) is noted for its occasional forays into unintentional surrealism — or at least into scientific territories that we journalists can have some fun with.
Belfiore's informative book focuses largely on DARPA's main projects, which are not so much odd as they are promising — artificial arms with implants that allow for user control; robots that perform lifesaving emergency operations; and, of course, those self-driving cars we've been waiting for. More than that, Belfiore gives us the fascinating history of how the birth of the internet (ARPANET) wound up in the laps of ARPA (the earlier version of DARPA).
The internet, of course, was an orphan project that nobody else wanted. We now know that ARPA may have done more than any other institution on earth to change the course of human history.
I interviewed Belfiore via email.
H+: DARPA's beginnings with ARPA — and the start of ARPANET — are legendary, but the legend is generally vague. You explore the roots of all this. It started as a project that nobody wanted! Can you say a bit about that?
MICHAEL BELFIORE: DARPA started life as ARPA (without the "D" for Defense) in response to the Soviet Union's 1957 launch of Sputnik. It was conceived as America's first space agency. But when NASA came online just a few months later, ARPA nearly didn't survive. It survived by branching out into other areas besides space, including information technology. And it survived by becoming a sort of dumping ground for projects that the armed services either didn't want or didn't know what to do with. That was how ARPA got its first computer, a surplus 250-ton, room-filling machine made to analyze radar data for the Air Force, along with the support staff needed to operate it. From there ARPA began its experiments in time sharing, interactive computing, and the ARPANET that was the genesis for today's Internet.H+: J.C.R. Licklider, who figures prominently in the ARPANET story, may have been the first "singularitarian." Can you tell us a bit about his history with ARPA and how visionary his work was?
MB: J.C.R. Licklider was a psychology professor at MIT when ARPA tapped him in 1962 to lead its Information Processing Techniques Office, or IPTO. This was, and still is, ARPA's information technology office. The man was out there, even by today's standards. He laid out his ideas for Man-Computer Symbiosis in a 1960 paper of the same name. You can read it on an MIT website.
Licklider was visionary in his understanding that in order to be truly useful, computers had to evolve from the glorified calculators of his day, where scientists input programs, walked away, and returned hours later for their results, into devices that people could interact with in real time, using such then-exotic technologies as speech recognition. In his words, from the paper: "The hope is that, in not too many years, human brains and computing machines will be coupled together very tightly, and that the resulting partnership will think as no human brain has ever thought and process data in a way not approached by the information-handling machines we know today." Under his direction, ARPA began funding the early research that led to the computing environments and user interfaces we take for granted today. Arguably, we have not yet developed all of the technologies needed to fully realize Licklider's dream.
H+: What is the current DARPA project that's most likely to be as world changing as ARPANET?
MB: My main criterion for choosing the DARPA projects I cover in my book, The Department of Mad Scientists, was that each has the potential to change the way we live, much like the ARPANET, the DARPA-funded project that became the Internet.
To pick my favorite example, autonomous vehicle technology could change our relationship to our cars. DARPA has been funding the technologies that will allow cars to drive themselves, most notably through its series of challenges. [http://www.darpa.mil/grandchallenge/index.asp] The most recent challenge, the Urban Challenge, was an auto race that required cars to navigate city streets while obeying all traffic laws and avoiding human-driven cars — all without human intervention. The race, completed in 2007, successfully demonstrated autonomous city driving for the first time, and greatly advanced the state of the art. Autonomous vehicle technologies, mostly in the form of safety systems, are now appearing with greater rapidity in production cars. Already we can buy cars that keep themselves in their lanes on highways and automatically keep a safe distance from cars ahead. In 10 to 20 years our cars will be able to actively avoid accidents, greatly reducing the number of people who die on the roads — currently 30,000 to 40,000 people in the U.S. every year.
H+: DARPA has a reputation for engaging in a lot of odd, sort of surreal projects... I think you chose to focus on their bigger, more "mainstream" projects, but did you bump into any stuff they're working on that struck you as very — shall we say — avant-garde?
MB: Two of the most "out there" projects I encountered were projects for modifying insects to act as spy bots, and one for programming matter to dynamically change shape, color, or just about any other property.
The insect project involves implanting larvae with electronics such as electrodes and radio transceivers. The bug develops around the devices, which become an integral part of it. The result is a cyborg insect, and the hope is that these bugs can act as tiny unmanned air vehicles that can be remote controlled and send back data to a human operator. Amazingly, researchers have actually succeeded in remote controlling moths with implanted electronics. Next step is adding spy gear such as cameras to the package.
Programmable matter technology is in the very early stages, but the potential here is to create objects on the fly in response to user input or environmental conditions. Think the evil terminator in the second Terminator movie. Or imagine fully interactive 3D displays that let you interact by touch with the objects "displayed," or, rather, created. How about ordering some device, toy, or other product from Amazon.com, and having it emailed to you instead of sent by UPS? The product would assemble itself from a slurry of raw matter in your 3D printer to conform to a blueprint sent in response to your payment. Perhaps you could even rent such products. On expiration, they would collapse back into their raw state. This technology would utterly change the way we interact with our data, and could prove every bit as revolutionary as the Internet itself. I, for one, would love to be able to store my possessions digitally instead of in bulky material form.
H+: What is DARPA doing in the energy field and why is that part of their mission?
MB: DARPA sees energy as a national security issue because, as program manager Douglas Kirkpatrick put it to me, energy is the most important limiting factor in any conflict. A huge part of every supply convoy going to the front lines, even the individual load of each soldier, is some form of energy, whether it's oil or batteries. Cut off that flow of energy, and you stop an army.
DARPA wants to give the armed forces the ability to generate their own energy in situ, wherever they find themselves. No energy supply convoy, no vulnerability. Accordingly, DARPA programs are working toward very efficient solar cells, which should allow each piece of gear a soldier carries to be battery-free; they're creating biofuels — fuels harvested from plants, algae, and other renewable feedstocks — that meet the specs for the Air Force's most demanding applications (i.e., jet fuel); and they're working to improve the efficiency of lighting systems, electronics, and more, so that they use less energy.
These programs don't have the sex appeal of novel forms of weaponry, but I believe this thrust has the potential to have the greatest impact on our lives as civilians. Imagine decentralized power production, where every house, every business, every building, generates all of its own power from the environment around it. Imagine the U.S. economy free of Middle Eastern oil (and the future conflicts we can avoid). Imagine the enormous fortunes that will be made in producing the solar panels, biofuels, and the rest that will make it happen — at a cost that is competitive with our current energy supplies. The effect will be truly transformative.
H+: About seven years ago, John Markoff told me that Tony Tether is a pretty conservative guy and that DARPA would probably stick to more traditional projects than they had in previous years. Was that your impression and did you catch wind of any dissatisfaction during your research?
MB: If you mean by conservative the political variety, then, yes, Tony Tether is a Republican, and proud of it. But his tenure was marked by boldness and risk-taking in the best sense. In fact his management was rather radical, and it did ruffle feathers among, especially, some of the academic centers DARPA contracts to realize its mad schemes.
Tether emphasized fast results. Under Tether, DARPA created programs aimed at very specific outcomes with specific, usually 2-4 year, deadlines. For instance, Revolutionizing Prosthetics had a 2-year component and a 4-year component. The 2-year component aimed to create a prosthetic arm that would be more functional than any previous prosthetic. It succeeded. The 4-year component had as its goal creating an arm that restores an amputee to full functioning with, essentially, an arm-sized robot controlled by the wearer's brain. Researchers on that project have made enormous advances, and they are continuing the work that DARPA helped them start.
But some academics that I have spoken with were very unhappy with Tether's management. One even told me that he felt Tether had actually eviscerated the country's R&D infrastructure by refusing to fund projects that had uncertain outcomes, i.e., pure research that may or may not one day lead to practical applications. (Tether strenuously refutes this, by the way.) My source felt that this was too big a departure from DARPA's traditional approach, and that it made it extremely difficult for academic researchers to commit resources (including graduate students) to research they had spent years pursuing. Tether's successor, Regina Dugan, has been busy repairing damaged relationships at academic research centers.
My take on this is that DARPA's mission never has been to fund research for its own sake. DARPA's mission has always been to prevent our country's adversaries from gaining technological advantages over us (i.e., to prevent technological surprises like Sputnik), and to help the military generate surprises of its own. The fact that DARPA has become so important to our national R&D infrastructure is a measure of its success in attracting the best minds to work on its programs. That its work has lead to so many generally useful technologies is a wonderful side benefit, and DARPA should be a model for other agencies that follow its lead in the pursuit of non-defense technologies. But it can't be all things to all people.
H+: Do you ever worry about the military having the most extreme technologies first? Will the first SuperAI be a soldier?
MB: The first super AI may well be a soldier, or at least one that works for the military. It has been said that DARPA programs have led to something like 3/4 of all the advances to date in the fields of computer science and artificial intelligence. It is, to me, an uncomfortable fact that the U.S. defense budget eats fully half of our annual national budget. I think that's too much and is ultimately unsustainable. Look what happened to the Soviet Union. Sooner or later we will have to spend less on our military and more on our people at home. And it needn't be at the expense of DARPA, which consumes only one-half of one percent of the overall defense budget. DARPA has already served as a model for the Department of Energy's ARPA-E, with great result. Hopefully more such civilian-focused efforts will follow.
See Also
DARPA takes on Suspended Animation: Zombie Pigs, Squirrels, and Hypersleep
DARPA Funding Nano-UAV Hummingbird
DARPA Plans for Immortal Synthetic Life
No comments:
Post a Comment