DARPA to Showcase Latest Cyber, Big Data Ideas

By Amaani Lyle

American Forces Press Service

WASHINGTON, May 19, 2014 – The Defense Advanced Research Projects Agency will display more than 100 projects and 29 programs in the Pentagon’s courtyard May 21 to demonstrate cybersecurity technologies and spark feedback from warfighters, a DARPA scientist said during a May 16 interview with the Pentagon Channel here.

Daniel Kaufman, DARPA’s Information Innovation Office director, said the ever-changing landscape of network warfare spurs the need for ongoing analysis of and dialogue about network security and greater user empowerment for warfighters in even the most remote locales.

“I want them to see the tools we’re building them to fight better and more effectively,” Kaufman said. “We’re technicians, scientists, and engineers and we build things, but there’s nothing like having somebody in uniform who actually has to use your product give you actual feedback.”

Kaufman described network warfare as a “new war,” with some 98 percent of devices having embedded microprocessors of some type.

“Everything in the world today has a computer -- your phone, your television, all our weapons systems,” Kaufman said. “We see huge promise in it and get benefits from network technology, but how are we going to protect these systems?”

But making forces safer yet more lethal to the enemy requires asking tough and unusual questions, Kaufman explained.

“It’s this cross-over between criminal organizations, terrorist organizations, state actors and non-state actors,” he added, “and the question is what tools do we have to create a new map?”

Kaufman said he and his team wondered about the possibility of creating software that was smart enough to contend with the best hackers in the world. He recounted an example of an all-computer chess league scientists developed in the 1970s.

“After seven years,” he said, “we beat a grand master, a year after that we beat [Russian Grand master Garry] Kasparov and from then on humans play for second place.”

Beyond the scope of chess, Kaufman noted, a more “human endeavor” such as the popular game show Jeopardy! even had famed contestant Ken Jennings toppled by IBM supercomputer Watson in 2011.

As such, DARPA will emulate Defcon, an annual hacker tournament in Las Vegas, to pit computers against computers and fine-tune network security.

But changing the game on network protection, Kaufman said, requires recognizing that hackers always have the advantage -- in that one breach can attack millions of machines.

“For the most part, either you have a PC or a Mac; we run almost all the same applications so one attack can hit many of us,” he said. “But what if you can make every computer different, much like our immune system, and now the attacker needs a special attack for every computer?”

And while Kaufman acknowledges some information technology officials are leery of the idea, DARPA scientists think they can accomplish this at a low level where attackers enter a system.

He explained that a computer is the only thing people buy that is “fundamentally broken,” since it always requires updates and patches, which can create network fissures that hackers may exploit.

As a result, DARPA scientists have been developing a computer that is designed with security in mind, and by its very construction prevents hackers. But the balancing act between thwarting hackers and enabling users persists.

“If we’re going to empower the military we need to change that dynamic; we have to make it so that you don’t need a Ph.D. in computer science … that you can still use it and program it for your own uses,” Kaufman said.

In big data, Kaufman noted that while the internet typically only indexes about 1 to 5 percent of all searchable information, national security reasons may call for a deeper probe into the web.

“The first thing we’re going after is the scourge of human trafficking,” he said, “and we’re already showing early results and making a dent in [countering it].”

Kaufman also said DARPA is trying to simplify computer language by taking complex programs such as Java, C+, and Haskell, which require extraneous compiling steps, and instead enabling delivery of actual commands.

“What if you could program a computer by actually telling it what you want to do?” Kaufman posited. “In one sense we should get rid of half the errors because now I don’t have to translate it into some weird language. I’m just talking to it.”

Another simultaneous challenge and promise of big data, Kaufman asserts, is assessing and simplifying the complexities of information gathering.

Big Mechanism software, he said, is a program that could read and digest all articles on a searchable topic and bring that information back to the user.

“Imagine that we could create a piece of software that would go out and read all the articles, break it down into all of its constituent parts, build it back into a model and then hand it back to us … and now that goes to a scientist and fundamentally, if we do it right, it’s a brand-new way of doing science.”