Software Eating Defense

Mehdi Alhassani

Source Link

In 2011, venture capitalist Marc Andreessen argued that software was taking over every industry, including ones that would not necessarily be intuitive. With the proliferation of the internet and the enablement of cloud computing, building a software-based business became the natural beneficiary because access to software driven products became much easier. Now, if one wants to watch a movie at home, they no longer need to drive to Blockbuster, rent a video, and play it at home. Now they can watch it on their video streaming device with the click of a button. They don’t go to the record store to buy CDs and cassettes, instead, they listen to Spotify, Pandora, or Apple music. Taken further, a lot of the novelty of Tesla is that it is a high-quality electric car, but also that when it wants to improve the range or other functionality, it sends a software update.

In Defense, however, software was slower to break through. Defense discourse has predominantly focused on big expensive hardware programs like the Super Hornet jets or the Black Hawk Helicopters. Less attention is paid to IT and software programs or their failures. As such, software at the DoD has not kept up with its commercial peers.

Earlier this year, Naval officer Artem Sherbinin published an open letter asking senior DoD leaders to “fix our software.” He warned that the “next war will be network and data centric,” but many U.S. intelligence systems do not share information or budgets or even talk to end users.

“It isn’t just our data that’s siloed, it’s our organizations too,” he added, with the concerns that failing to act “will lead to decline in American military superiority.”

Companies have noticed the potential for neglect and lack of accountability with software. One Defense executive even told me about his startup: “We are really a software company, but slap hardware on top because DOD doesn’t pay for software.” In other words, the government commoditizes software and doesn’t put a premium on the quality of the product (let alone the product working).

There is good news, however, that indicates the tide may be turning. In the FY23 National Defense Authorization Act, software receives more attention than ever. For example, the legislation calls for an independent review to identify the most critical software and IT challenges that result in lost working hours and the increased costs of failed software systems. The Under Secretary, in consultation with the Chief Information Officer and the newly established Chief Digital and Artificial Intelligence Officer, will be required to submit annual reports addressing the Department’s software development process and progress. Congress’s increased attention and oversight of software systems within the DoD are both welcomed and badly needed.

Things are slowly turning within DoD in taking more software-centric approaches to acquisitions. The Army TITAN program, an AI/ML- enabled ground station intelligence platform, is currently being competed for by both traditional Defense primes as well as software companies in partnership with hardware focused companies. The Air Force awarded Northrop Grumman the prime contract for its Sentinel program to replace the antiquated intercontinental ballistic missile system. The service is taking a “digital engineering” approach. Both these examples put software first to enable systems that still include hardware.

But for the U.S. to retain military superiority, more is needed. First, we must fully leverage commercial software to build operational advantage and stop building technology where solutions already exist. Although it may finally be turning a corner, roadblocks like the years-long delay in technology upgrade for its IPPS-A personnel system, which has not optimally leveraged existing commercially available technologies, should never happen again. Part of the problem is that there isn't clear instruction to leverage commercial software, which is already built, proven in the marketplace, and ready to deploy, and instead the government often tries to build bespoke solutions that are costly, slow, and often fail.

Second, we must invest. Project Maven is the most advanced and successful AI initiative the DoD has ever executed. But it has been several years since that work kicked off. Where is the next Project Maven that will drive our capabilities forward?

Finally, we cannot continue to think all software engineers are fungible. Like in any profession, from sports to science, talent matters. Ask yourself which companies employ the best software engineers in the world. Then look at the list of top 100 Defense companies by revenue. We need America’s best talent working on these complex technical problems.

To be sure, hardware will never be unimportant to American defense. The traditional defense primes and other industry leaders have been crown jewels in American industry and innovation, building some of the most advanced rockets, planes, and machines the world has ever known.But in today’s technological era, software is equally important, and we must treat it as such. The latest NDAA is a strong step forward.