THE DEPT. OF INVENTIONS
How a Pentagon arm spawned Siri, the Roomba, the Googlee car & even tthe ’Net
FOLLOWING the invasions of Afghanistan and Iraq, US troops were embedded in areas where few, if any, of our soldiers spoke the language. The Defense Advanced Research Project Agency (DARPA), the research-and-development branch of the Department of Defense, had a solution: a “Star Trek”-esque device known as the Phraselator.
The handheld device, issued to some divisions in Afghanistan in 2002, wasn’t a real-time translator. Instead, a soldier would hit a button, and the device would say it was going to ask questions in Pashto or Arabic and instruct the person to raise one hand for yes or two hands for no.
Unfortunately, the yes-or-no answers weren’t much help, and Afghan leaders didn’t feel comfortable talking to a machine. A 2009 Army survey universally panned the universal translator: “Took too long to translate the correct phrase,” soldiers said. “Translation wrong more often than not.”
The Phraselator was dropped. DARPA cut off its outside research funding for the project, including its support of a company called SRI International. SRI spun off the technology as a separate entity called Siri, which was then bought by Apple and incorporated into the iPhone.
IT’S a common refrain in Sharon Weinberger’s new history of the agency, “The Imagineers of War: The Untold Story of DARPA, the Pentagon Agency That Changed the World” (Knopf ).
While the agency didn’t always succeed in its primary mission — improving the US military — its influence is everywhere.
“DARPA’s work on natural language processing was not necessarily going to help soldiers talk to Afghans,” Weinberger writes, “but it could help Americans find the closest Starbucks.” Founded in 1958 in response to the Soviet launch of Sputnik, the Advanced Research Projects Agency (the “DD” was added in 1972) was also a pioneer in: GPS: DARPA was in charge of the first communications satellite and first spy satellite. It was involved in Transit, a satellite launched to help submarines navigate — leading to the Global Positioning System.
Driverless cars: Starting in 2004, DARPA sponsored robotic car races across the Mojave Desert. It offered a $1 million prize to the first car that could travel 150 miles across the desert between Barstow, Calif., and Primm, Nev. Unfortunately, the best any car could do that year was 7.32 miles, but one year later, the Stanford Racing Team developed a car (left) that completed a 212-mile course in just under seven hours. The head of that team, Sebastian Thrun, went on to pioneer the development of Google’s self-driving car.
Stealth aircraft: DARPA had been working on invisible jets since the 1970s, starting with a prototype called Have Blue. The agency’s research laid the groundwork for aircraft today that can’t be seen by radar, such as the helicopters used in the Osama bin Laden raid.
Robotics: One of DARPA’s main areas of inquiry has been using robots in the battlefield, such as PackBot, which was used to dispose bombs in Iraq and Afghanistan. The PackBot was developed by the company iRobot using a DARPA grant. iRobot would go on to develop the robotic vacuum cleaner Roomba (right).ht).
BUT DARPA’s most famous contribu-bution was inventingng the Internet. (Sorry,y, Al Gore.) In the 1960s, DARPA sponsored the development of a system of networked comput-ers called the ARPA-NET.
J.C.R. Licklider, a re-research director at ARPA,PA, was fascinated by the ways computer operators’ “command and control” of their machines was evolving, and realized that people would use the devices to interact with one another across great distances. Though the Pentagon initially wanted a way to control its nuclear arsenal, Licklider had a bigger goal in mind, and ARPA gave him the opportunity to steer the research
on networked computing, awarding contracts to scientists at MIT, Stanford and other institutions.
The effort almost fell apart within a year, though, as ARPA’s head, Robert Sproull, was looking for ways to cut $15 million from the agency’s budget. Licklider’s research hadn’t produced any concrete results, so it was a natural targetg — until Licklider invited SproullSpro to visit some of the computerpute labs and convinced him theirthei work would pay off in the long run. It did — ARPANET devveloped many of the protocols
thatth the Internet uses today. F ROM a purely military perspective, DARPA’s biggest success has been ddrones, which the agency has bebeen working on since the Vietnamnam War. WWith Project AGILE, the agenagency spent millions of dollars trying to develop counterinsurgency techniques to help the South Vietnamese government defeat the Viet Cong. One endeavor took a small drone helicopter developed by the Navy to drop nuclear bombs on Soviet submarines, added a TV camera, and installed the remote-control equipment in the back of a jeep. Then, they added a small gun and bombs. Although these drones tended to crash during test flights, and were never used in Vietnam, the project at least established that, in principle, drones could be viable in combat.
By 1980, DARPA had resumed its research into unmanned vehicles that could replace spy planes, working with a company called Leading Systems Incorporated to develop a prototype. This drone, code-named Amber, could be used for reconnaissance missions or as a cruise missile. After 10 years, however, the Pentagon refused to extend the program further. LSI went bankrupt and was sold to a defense contractor called General Atomics — at which point the CIA stepped in. With a few modifications,ns, inincluding a significantly quieter engine, Amber evolved into the Predator drone, which both the CIA and the Air Force would use in Iraq, Afghanistan, and elsewhere.
Project AGILE was also responsible fore the agency’s darkest hour. ARPA’s Operation Ranch Hand tried to develop weapons that would kill people but leave crops and landscapes untouched. The result: Agent Orange, which not only killed Viet Cong, it caused illness and death for thousands of Vietnam veterans. A ND then there were the duds.
DARPA’s attempts to win the Vietnam War bordered on the absurd, like the proposal to relocate entire village populations into “strategic hamlets,” walled communities designed to kkeep out infiltrators. Villagers wwere understandably opposed.
A nuclear-weapon shield during the Cold War proved particularly vexing. One proposal was to sshoot down Soviet ICBMs with a ccharged particle beam, which would be powered by draining the Great Lakes and running all tthat water through a massive generator. That idea got kicked around for at least a decade, though almost nobody ever expected it to work, mainly because it seemed to keep scientists’ creaative juices flowing.
In the 1970s, DARPA psychologist George Lawrence explored the possibilities of biofeedback ttraining: What if combat soldiers ccould be taught to slow their own heartbeats after getting shot, to reduce the odds of bleeding out? What if fighter pilots could willfully maintain low blood pressure in emergency situations? Lawrence eventually decided the training would be too much work for insufficient results, but along tthe way he was asked to look into a CIA-funded project testing Uri Geller, famous in those days for hhis claims of an array of psychic ppowers.
Lawrence flew out to Californnia, showed up at the Stanford Research Institute (SRI) looking heavily hung over, and declared, “OK, show me a f- -king miracle.” But Geller never let the team test him under truly scientific conditions.
As another witness described it, the SRI researchers, who were inclined to believe Geller’s claims, came across as “bumbling idiots.” Lawrence grew increasingly frustrated. At one point, Geller claimed he had moved a compass needle with his mind; Lawrence stomped his foot on the floor and made the needle move much farther.
Much later, asked to give his appraisal of the CIA’s parapsychology research, Lawrence was blunt: “You have been wasting your money. Every damn dime of this is nonsense.nonsense.”