Opinion / Reform

Bad Idea: Relying on Government Software

Competition is much about identifying and exploiting asymmetric advantages—areas of outsized overmatch against an adversary. Secretary of Defense Lloyd Austin, in his recent comments at the Reagan National Defense Forum, identified the U.S. commercial innovation base as one of those key advantages that must be leveraged to compete with China. However, he noted that, “For far too long, it’s been far too hard for innovators and entrepreneurs to work with the Department. And the barriers for entry into this effort to work with us in national security are often too steep—far too steep.” Gaining the best capabilities for national security requires identifying and removing these barriers to the free-market commercial innovation base. In the increasingly important realm of software for national security, the emphasis on government-built and government-owned software forms one of these barriers. Relying on government software without leaving room for commercial solutions is a bad idea.     

The nation that successfully leverages data and software to reach decisions faster than adversaries will possess an asymmetric advantage ahead of future crises and conflicts. This includes software that translates petabytes of disparate digital data sets—open-source, classified, structured, and unstructured—into decision-ready information. Whether curating logistics data for force movement decisions, alerting humans to suspicious changes in adversary maritime posture, or pairing joint force defense assets to incoming enemy missiles during an attack, the best software can provide better awareness and the decision advantage that will be crucial in any future fight. The way that this software is developed, secured, and updated makes all the difference in mission effectiveness.

The 2021 DoD report to Congress on China’s military developments highlights the centrality of data for dominance in “intelligentized” warfare of the future. If data and corresponding software are central to the United States maintaining a competitive advantage over China, the quality of that software becomes a key warfighting variable. Subtle cues in remote sensing data can indicate adversary actions in the gray zone below the threshold of armed conflict—subtle actions the best software can identify and highlight. Raw Artificial Intelligence (AI) detections of 100,000 Chinese maritime militia fishing trawlers in commercial imagery across the western Pacific provide little value to military leaders. That same data, curated by software that combines multiple sources of data to highlight significant changes in vessels at key locations, suddenly become compelling to military analysts and commanders seeking to deter adversary aggression. In conflict, the nation that can observe enemy forces and make decisions to pair those forces with effectors the fastest will be dominant. This decision superiority is enabled by access to data—whether sensors, shooters, or supporting forces—and combining this data enables the possibility of detecting, tracking, identifying, and engaging a threat almost simultaneously. While making sense of the data for a single kill chain decision may be manageable for the human mind, that decision expanded to thousands of near real-time decisions across a vast battlespace like the Pacific becomes quickly overwhelming. Software can manage this data at scale, autonomously presenting options for human decision and allowing a rapidity of actions that outpace adversaries.

However, not all software is created equal. The U.S. national security enterprise is witnessing a software-first revolution, but it is chasing government software solutions to the exclusion of the commercial software innovation base. The recent trend towards government-built software fills an important role of making the government conversant in software and providing an internal capability to quickly write and field code for tactical challenges, such as automating processes that were previously manual. Even so, the increasing push for government software eschews the use of commercial software that can provide the highest quality, scalable capability needed for strategic competition.

Defense software can be broadly categorized in two ways: “who developed it?” and “who owns the intellectual property?” There is an increasing tendency within the U.S. national security establishment to develop software internally using government coders. This approach provides the government with software on the cheap for a defense enterprise predominantly focused on hardware procurement. What’s more, the government then owns the code and can make changes to software without the price tag levied by traditional hardware vendors who sold military equipment but retained the software rights under the hood. While government ownership may reduce the cost of software procurement upfront and satisfy immediate needs, history shows that the intellectual property code, once in the hands of the government, quickly becomes outdated. Rather than leveraging the incentives commercial companies hold to constantly update their subscription software—known as Software as a Service (SaaS)—stagnant government software quickly creates the frustrating experiences ubiquitous for anyone who has used a government system. While seemingly cheaper up front, the long-term costs of government-owned software often prove more expensive, and failure abounds for government-owned IT programs. 

As a result of these lessons, government contracting agencies highlight the need to leverage commercial and subscription software. Congress (10 USC §2377) and the Federal Acquisition Regulations require the use of commercial capability where it exists and even the modification of commercial capability before the government should attempt to write applications itself. Despite these regulatory requirements, government agencies continue to build their own software, creating barriers for commercial companies.

An analogy may be instructive to demonstrate the deleterious effect of a solely government-built approach to software development. Imagine a military aviation unit suffering from a major pilot shortage (as they often do). There are not enough men and women to accomplish the highly specialized technical skill of piloting combat aircraft and integrating within the Joint Force. The unit devises an expedient solution: survey the unit’s support staff for anyone interested in flying. Whether the staff dabbled in aviation in the past or simply have the itch to fly, they can sign up to attend a 12-week crash course in piloting. These newly trained pilots would be augmented by additional pilots-for-hire who would also fly for the government—but the performance of these pilots would ride on the shoulders of the government, not the reputation of a commercial company. Such a program would quickly fill capacity shortfalls and ensure a unit is ready for combat deployment—right?

This solution would likely be absurd to any military pilot who has endured the two or more years of exhaustive, requisite training for all professional combat-ready aviators. From the nuances of hazardous weather phenomena to the subtle cues of spatial disorientation and the complexities of combat employment, effective flying is a highly specialized skill developed with years of both education and experience. Developing and implementing effective software is no less challenging and demands similar experience and expertise, the absolute best of which exists in the commercial Software as a Service realm.

A key advantage the United States maintains over autocratic strategic competitors is its free-market commercial innovation base. The engine of competition for commercial market share drives companies to constantly improve their products and assess their approach. Without non-stop innovation, firms become stodgy, products become outdated, and an upstart company outcompetes incumbents. This commercial environment not only drives exceptional product development, but also necessitates that companies constantly update products to ensure top-line performance and security. While government software developers certainly hold an interest in providing suitable software for national security, the incentive structures are just not as powerful. Government software that is good enough at 5 PM on a Friday will likely remain in that state. However, for a commercial company, good enough is the death knell for private-sector market share.   

There is certainly a role for government developers and code, not the least of which is to make the government more conversant in the challenges of software and data and to better enable them to set requirements and manage commercial software programs. Several government software groups have achieved success solving data problems at the tactical level. However, policymakers make a tenuous logical leap when they assume these efforts can scale to encompass what they envisioned with Joint All-Domain Command and Control and deemed necessary to dominate in peer-conflict of the future. It is far easier to write software to alleviate the frustration of a daily task, such as automating a workflow, than to write code that integrates across a military service, connects to cloud-based applications, and translates with the Joint Force—not to mention being constantly updated to protect from emerging security threats and to incorporate new features enabled by other systems and applications available on the network. The main challenge for national security software is not in tackling simple problems in a vacuum, it is in making those software solutions relevant and compatible with a constantly changing Joint Operating Environment—this is what the best commercial software companies do every day at scale.

In addition to challenges of quality and scale, government software can create barriers to commercial companies by removing incentives for private sector investment. The commercial startup software base, with a few exceptions, relies on venture capital investment to research, develop, and aggressively scale to compete for and win market share. These investors bet on startups that have a chance of breaking massively into an existing market or opening a previously unknown or underdeveloped market. Government software effectively shrinks or even removes the already small commercial market for national security software. When the investors don’t see a market in the government, the startup software company either looks for private sector problems to solve or sells its intellectual property to the government, stymieing further commercial investment, and effectively killing the market. Government agencies should ensure a portion of the national security market is available for commercial software developers with direct transition paths to Programs of Record, in order to incentivize investors and leverage the commercial innovation base.

If strategic competition is much about analyzing information for decisions faster than an adversary, software will be at the heart of that competition. Government agencies will often happily spend tens of billions of dollars on underperforming hardware procurements but will balk at spending one-tenth of one percent of that budget on quality commercial software—software that can drive U.S. strategic competitive advantage. The national security enterprise should engage the flywheel of innovation and free-market competition in the commercial software sector for national defense. This is a strategic advantage over our competitors and one that could prove decisive. 

The views expressed are those of the author and do not reflect the official position of the U.S. Air Force or Department of Defense.

(Photo Credit: DoD photo by Senior Airman Franklin R. Ramos, U.S. Air Force/Released)