When it poses a scientific problem to the world, the Pentagon’s research laboratory for spies never asks respondents to surrender their intellectual property because doing so scares off innovative startup businesses

“We leave it to the inventor,” said Jason Matheny, director of the Intelligence Advanced Research Projects Activity, often referred to as the Defense Advanced Research Projects Agency (DARPA) for spies.

Jason Matheny, director of the Intelligence Advanced Research Projects Activity
Jason Matheny, director of the Intelligence Advanced Research Projects Activity

“We do not ask the inventor to give up their intellectual property, for a couple reasons,” Matheny told reporters March 14 during a Defense Writers Group breakfast in Washington, D.C. “We think, ultimately, the technology market benefits from inventors holding their own IP. Second, we find when we ask for the IP it has a chilling effect on participation rates, particularly on the people we really want to attract, who are the entrepreneurs and small startups.”

While DARPA focuses on scientific endeavors to support the military, IARPA aims for leap-ahead science-and-technology advances that boost the effectiveness of U.S. intelligence agencies, though it sometimes transitions programs to the military services. Its portfolio runs the gamut of basic scientific research from computer science to physics, cognitive psychology, sociology, linguistics and neuroscience.”

Most of IARPA’s funded work is not classified, though its budget is. It is funded as a subset of the Defense Department’s Military Intelligence Program (MIP) budget, which came in at $21.2 billion for fiscal 2019.

Unlike the National Science Foundation or the National Institutes of Health, IARPA runs research competitions that award whichever team can solve a complex problem most effectively in the shortest time, he said. To attract the largest talent pool as possible, IARPA strives to keep its programs unclassified, Matheny said.

Many of the requests for information the agency releases propose a challenge that is analogous to a classified capability gap. Instead of asking for software systems that can identify terrorist activity, IARPA will ask for technologies that can identify videos with break dancing in them, Matheny said.

“We work a lot with industry, particularly small businesses who aren’t the usual suspects in, say, defense or intelligence contracts, to ensure that the best and brightest are able to work on our problems,” he said. “We avoid, to the greatest possible extent, running classified programs. … We try to find an analogy for an intelligence problem that is de-classified and can be kept open.”

IARPA engages researchers in high-stakes tournaments in which multiple teams compete toward a common set of technical goals with “highly ambitious milestones associated with them,” Matheny said. The agency is interested in achieving at least a 10-fold improvement in the state of the art capability in whatever field is being tested, he said. Many times, solutions yield 1,000- or 10,000-fold improvements, he said.

When a suitable solution to a specific problem is identified through research competition, the technology is made available to various intelligence agencies, which can then decide to buy the item from the inventor or develop its own based on the scientific principals established through IARPA’s process.  

A large part of IARPA’s portfolio is chasing artificial intelligence and machine learning systems that can make sense of the massive flood of raw data collected by intelligence, surveillance and reconnaissance platforms daily. Much of its video analytics programs has been sponsored or adopted by the Air Force as a solution to its data-flood problems, Matheny said.

“There is too much data for human eyeballs to look at,” he said.

IARPA partnered with Carnegie Mellon University, IBM’s [IBM]Watson Research Center, Raytheon [RTN] and others on the Aladdin Video Program, which seeks to combine the state-of-the-art in video extraction, audio extraction, knowledge representation, and search technologies in a revolutionary way to create a fast, accurate, robust, and extensible technology that supports the multimedia analytic needs of the future. That program is complete and has transitioned to use by the Air Force.

Aside from its high-stakes research tournaments, IARPA crowd-sources problems that are open to anyone. A challenge it posed to forecast basic world events brought in 3 million responses from more than 40,000 participants, Matheny said. Such programs attract non-traditional contractors and businesses that are either unable or unwilling to work with the government, he said.

“For those who don’t want to go through the whole federal contracting process, which is itself a war of attrition, we run public prize challenges in which we just award prize money to anyone who can solve a problem,” he said. “They don’t have to have a federally approved accounting system. They don’t have to have a contractor officer go through a six-month process in order to award money. If they can solve our problem, we award them cash.”