Military Open Source
by Alex Howard:
"In his remarks, James E. “Hoss” Cartwright, a four-star general who retired from the United States Marine Corps in August 2011, outlined a strategic need to make military technology more modular, based upon open standards and adaptable on the battlegrounds of the future.
Cartwright, the first holder of the Harold Brown Chair in Defense Policy Studies for the Center for Strategic & International Studies, a member of the Defense Policy Board Advisory Committee, and an adviser to several corporate entities in the defense industry, is well placed to have an informed and influential opinion.
Over the course of his talk at the Military Open Source Conference, Cartwright outlined how open source software models could be applied to hardware, making vehicles into adaptable platforms for different missions, not vertically integrated programs that can take a decade or longer to design, build or change.
Given the scope of the Pentagon’s current capabilities and DARPA’s research, potential ethics concerns abound, from drone warfare to sentient robotics to targeted genetic plagues to brain scanning to biohacking.
In that context, Cartwright prioritizing ethical qualms about secrecy, privacy and big data over those raised by biohacking was notable.
The issue that Cartwright said bothered him the most, however, was big data. “There are really no secrets out there,” he said. By exposing data to a larger dataset, it’s possible to correlate real identities. (That’s the so-called “Mosaic Effect.”)
That’s what’s now happening with network intrusions from other countries, he said, which leads to genuine national security headaches. Cartwright noted that while the federal government has huge classification protocols, they’re nearly all discoverable if you know how to correlate the information. Even correlations in anonymized data can lead to the discovery of true identities.
Big data concerns aside, Cartwright highlighted a strategic need for the U.S. Department of Defense to address these risks and develop improved man-machine interfaces, from touch screens for unmanned vehicles and weapons systems to prosthetics for veterans.
One huge challenge that the armed services are facing today, Cartwright said, is adapting code in response to what soldiers are actually encountering in the field.
We can’t send issues back and have people quickly rewrite code, he said, which presents significant problems. To put it another way, the DoD wants the armed forces to be able to “write as they fight.”
Cartwright described a pilot program where contractors and grad students were sent into the field so they could understand the problems they were working against and reduce the time to write code to address it. The results were promising: they didn’t lose any technical staff and turnaround time for patches went down drastically, once they were able to get inside the decision cycle.
Only programmers in the field can teach analytic algorithms to determine the difference between an ambush and a drug deal, said Cartwright.
“You can’t do that unless you know how to dig for data and understand context,” he said. “That’s the turnaround time that we needed to stay inside an adversary’s decision loop.”
That’s particularly relevant for networked warfare. According to Cartwright, new software works in the “cyberfight” in Afghanistan for about 9-14 days before it needs minor changes — but new systems take years to build. Top leadership in the military thinks a problem in the battlefield means that an entire new platform is needed, he said, but you’re looking at 14 years to build a new kind of truck." (http://radar.oreilly.com/2012/12/darpa-big-data-military-open-source-agile.html)
by Alex Howard:
"What needs to change is the incentive structures for the people building and designing the “platforms of record” in the future, said Cartwright. That means designing programs and apps for problems we actually have, versus developing something that doesn’t get into the field for 10-15 years — and if you guess wrong on who an adversary will be, that sends you into a modification cycle of at least three years.
Open source methods, by way of contrast, can give the military the ability to change software in weeks and months, not years, said Cartwright. In that context, he indicated that the Pentagon is looking at how they can move from tightly, singularly integrated programs in the direction of more open platforms and open standards, where war fighters can add or get capabilities with modularity and at a speed measured in weeks and months, not months and years.
During the question and answer period that followed his remarks, Cartwright followed up on his comments on open source. Cartwright said that the Pentagon would like to get to the point where platforms are a conveyance for the needs soldiers have, with infrastructure set up in such a way that things can be switched out.
Notably, he said that in the past few years of the financial crisis, defense technology manufacturers that are agnostic to platform are faring far better. “They’re building code — sensors, activities — and others are not,” he said, “and if one or two programs are canceled, they’re in trouble.”
Cartwright asserted that military service acquisition people have started to understand the value of flexibility of technology that enables soldiers to quickly configure technology for fights.
To scale that across the entire military, he said, they must adopt more common standards across all services. Eventually, that would mean “displays, chipsets, anybody in this room can write code against, depending upon what the customer wants.”
Cartwright said he’d like to see today’s model of open source extended to military software and hardware.
“We’re thinking about a future where everyone’s garage can be a sweat house for the military,” he said, playing to his audience of military open source conferees." (http://radar.oreilly.com/2012/12/darpa-big-data-military-open-source-agile.html)