In 1957, the Soviet Union caught the US completely off-guard. Its military launched Sputnik -- the world's first artificial satellite -- heralding the dawn of the space age.
President Eisenhower's response was to create the Advance Research Projects Agency (ARPA) with a clear mission: "prevent technological surprise". Eisenhower hoped that the agency would produce revolutionary technologies and thus guarantee that never again would the US military be caught with its technological trousers down.
Now in its 50th year, the Defence Advance Research Projects Agency has an impressive list of accomplishments behind it.
After playing an integral role in the fledgling US space programme, DARPA gave us the satellite-based global positioning system (GPS), stealth aircraft and the precursor to the internet. So it is hard to argue that the agency hasn't lived up to Eisenhower's early dream.
But DARPA has produced its fair share of clangers too. Over the years it has been widely criticised for investing millions of dollars in some pretty harebrained research schemes from futures markets aimed at predicting assassination attempts, to mechanical elephants that could barge through jungle terrain unsuitable for wheeled vehicles.
Read about some of DARPA's most spectacular successes and failures
Those supportive of the agency's unique approach argue that such failures are important to the culture that has made DARPA so successful.
Tony Tether, DARPA's director, says it is a "freedom to fail" that lets his staff discover truly revolutionary new technologies.
"And fail we do," he told an audience of 3000 potential recruits at the DARPATech Symposium last year.
"But that's OK -- failure sometimes happens when you are bringing new capabilities into reality," he said. "You only really fail if you don't learn what happened and stop trying to succeed -- you have to try again, and again, and again."
This attitude undoubtedly sets DARPA apart from other research agencies. Indeed DARPA has no laboratories or scientists of its own. Nor does it use any kind of peer review for assessing the viability of a project or programme.
Instead, the agency employs "program managers" who fund universities and businesses to carry out research that might otherwise be too risky for research agencies to back. Currently 140 programme managers disseminate some $2.9 billion of funding each year.
Program managers act as judge, jury, and, if the research doesn't go well, executioner. This simple set-up lets DARPA enter new areas quickly, and pull out just as fast if the research turns out to be going nowhere.
Ephrahim Garcia, a mechanical and aerospace engineer at Cornell University, Ithaca, New York, has worked on DARPA-funded research, and also served as a programme manager for DARPA between 1998 and 2002.
During his time at the agency, Garcia ran five programmes on research such as morphing aircraft and exoskeletons, managing a budget of $20 to 30 million a year.
Garcia says the limit to what gets funding is the imagination of each programme manager and their ability to convince the director that an idea could improve US military capabilities.