Want an Agile Pentagon? Don’t Go Chasing ‘Waterfalls’

In the absence of something better, the Pentagon churns out strategies every four years, releases already-outdated guidance and concepts, and issues five-year spending programs that would be familiar to Soviet planners. It is no way to keep ahead of China’s military modernization or technological disruption. There is a better way to manage complex processes like these, and some Pentagon and Congressional staff know it — if only they would just use it.

It’s called agile development — a concept from the software industry, where it has largely replaced the traditional “waterfall” method. The Biden administration should apply the same principle for developing strategy and translating it into concepts and capabilities. 

In the waterfall method, software developers collect requirements in spreadsheets, hand them off to engineers who did not always understand the root problems their products were solving, and then delivered technical solutions that sometimes were outdated by the time they released the software. This should sound unsettlingly familiar to anyone in the Defense Department’s requirements and acquisition processes.

In contrast, the agile model — and the rise of product management as a discipline in technology — focuses on delivering what users actually need, not what gets recorded in requirements spreadsheets. To do this, software developers collaborate closely with the product’s end user to iterate rapidly and deliver better products, faster. This change revolutionized the software industry and cemented Silicon Valley’s status as the global leader in software innovation. The Defense Department would be well-suited to learn the same lesson. 

It’s not new practice to apply business management ideas that improve the weapons-buying process. In 1961, Defense Secretary Robert McNamara sought to rationalize the chaotic defense budgeting process when he adapted then-popular business management techniques to create the planning, programming, budgeting, and execution system. With minor modifications, that industrial-age system is still in use today and is woefully out of step with the current strategic environment, according to a new report by AEI’s Bill Greenwalt, former deputy under secretary of defense for industrial policy, and Dan Patt, DARPA’s former deputy director for the Strategic Technologies Office. They suggest ways to make budgeting and acquisition more agile as a way of developing and acquiring new weapons on tighter, more relevant timeframes. 

This is an excellent idea, but let’s go further. The department updates its strategy every four years, per congressional mandates. The time-consuming, linear process usually takes at least one year, then the strategy sits untouched for four years. As a result, the new strategies are always too late to influence a new administration’s first budgets, and many are quickly overcome by events. Take the 2001 and 2014 Quadrennial Defense Reviews. The 2001 review was prescient about the emergence of military competition with China, but President George W. Bush’s Pentagon released it mere weeks after the shock of the 9/11 attacks upended its assumptions. Likewise, the explosion of ISIS and Russia’s invasion of Ukraine invalidated the 2014 review’s assessment of the strategic environment within months. 

Why not issue something quickly  — call them “minimum viable strategies” — comprising critical priorities and tradeoffs early in an administration, then iterate on these strategies over time? At one level below strategy, Pentagon leadership issues critical guidance documents such as the Defense Planning Guidance on a yearly basis, but these are always too late to actually guide development of service programs and budgets. What if the guidance were standing and iterated quarterly in collaboration with the services instead? 

Likewise, on the uniformed side, the Joint Warfighting Concept is meant to galvanize the department around a future vision of warfare and guide service investments. Unsurprisingly, the concept is late and will likely need to be updated immediately. Why not issue a “minimum viable concept” six to 12 months ago, before the transition to a new administration, and update it quickly based on feedback from the department?

As it so often is, the answer to these “Why not?” questions is, likely, “because that’s not how we do it.” Experienced bureaucrats might point to any number of policies or processes that inhibit agility, such as the need to coordinate every policy decision with relevant stakeholders. Hill staffers rightfully will note the annual nature of the defense authorization bill as a constraint. But we should never forget that processes are a means to deliver good outcomes. If the processes fail to deliver good outcomes, we should change or abandon them.

Agile development is a much better way to deliver complex outcomes than current, Byzantine Pentagon processes. It does away with overly-literal and unresponsive requirements-gathering in favor of embedding with stakeholders and understanding their challenges. Adopting this model for strategy and guidance could make coordination between all parties a far less onerous affair, as each review only would pertain to that iteration, rather than a gargantuan final document. Moreover, policies and guidance could be updated quickly based on new information, making each review less of a “life or death” struggle for stakeholders, thereby reducing bureaucratic resistance. 

Across the Potomac River, Congress should use the National Defense Authorization Act to move away from rote, sequential processes toward more agile ones by working closely with the Pentagon to focus and iterate on issues of critical strategic importance, like joint command-and-control or posture in the Indo-Pacific theater. 

The Biden administration has started strong by issuing an interim National Security Strategy quickly, rather than releasing a full strategy after a year of review, as is the norm. Short-term Pentagon reviews of key issues like China, Afghanistan, and the global laydown of forces also are welcome. Even better would be for the Biden administration to develop and release an interim National Defense Strategy and supporting guidance soon, instead of waiting a year for “perfect” and likely outdated documents. 

Processes are necessary to manage complex tasks and deliver good outcomes, whether in software or in national security policymaking. But we must always remember that they are a means to an end, not the end in themselves. Just because we’ve always done it this way doesn’t mean that’s how we should always do it. Pentagon and Congressional leaders need to adopt a better, more agile model quickly, even if doing so is uncomfortable or disruptive in the short term. Otherwise, clinging to familiar, outdated processes will provide little comfort when China surpasses the United States as the world’s foremost military power.