In August 2023, U.S. Deputy Defense Secretary Kathleen Hicks announced the Replicator effort, catching many off guard. It aims to field thousands of autonomous drones within two years to compete with China’s massive capabilities in a region fraught with tension. While audacious, this announcement was met with significant skepticism—much of it warranted given the Defense Department’s track record with similar efforts.

The difficulty is that Replicator requires disruptive innovation — innovation that rapidly introduces new concepts and/or technologies, and significantly changes the operational level of warfare. This is exceptionally hard and should not be confused with the incremental evolutionary innovation that the DoD historically exploits in peacetime. Critically, emerging technology and creative warfighters can rapidly disrupt traditionally dominant forces, such as in the spectacular example playing out in Ukraine.

The good news is that the DoD has recent successes.

In 2022, the Navy began a two-year experiment known as the Unmanned Task Force. While much of the UTF effort was classified, it operationalized and fielded multiple disruptive capabilities to the naval and joint forces. This success is being institutionalized as the Navy’s new Disruptive Capabilities Office and will directly support Replicator.

Key to the UTF’s success was unblinking adherence to four principles:

1. Solve problems, don’t meet requirements. Starting with requirements constrains thinking and immediately eliminates options because it begins with a solution rather than trying to solve the problem itself. It also takes many years, involves organizational politics and includes individuals far removed from today’s problems.

The UTF focused on problems identified by four-star combatant commanders and further refined their detail with the operational community. While the warfighters are the experts in the operational problems, they rarely have bandwidth to dive deep into every facet of every problem they face or to optimize their problem articulation for the innovation ecosystem. This was a critical functionality performed by the UTF.

2. Protect, incentivize and embed the innovators. An organization’s primary innovation activities (i.e., its evolutionary innovation activities) will destroy all attempts to innovate disruptively. This isn’t because of stodginess or bad behavior — it’s by design.

Because evolutionary innovation seeks to improve the status quo while disruption seeks to overthrow it, the bulk of an organization will see disruption as misaligned, attempt to kill it and assimilate its resources. A common reaction to this tendency is to isolate the innovation group. This is a mistake. Organizational separation will give the innovation group speed and agility, and build barriers ensuring their fruits are impossible to leverage.

To this end, the UTF did two rare, seemingly counterintuitive things. First, it remained physically located inside the Pentagon and vehemently fought all attempts to relocate to more so-called innovation-friendly settings. Second, it remained administratively located in the Navy’s resourcing and requirements organization. While these choices throttled the UTF’s tactical speed and agility, the strategic gains from always having a seat at the table were a cornerstone of its successes.

3. Experiment early, incrementally and only against actual hypotheses. Experimentation is too often confused with testing. Cost-performance-schedule cultures typically conduct tests (often mistakenly called experiments). In a test, failure is an undesired result, and those championing failure lose credibility and resources. In an experiment, an undesired result is a learning experience, and organizations that learn fastest typically prevail (not those that fail fastest).

A gold standard for early, incremental, hypotheses-based experimentation is the NASA lunar landing program that conducted many launches from 1961 to 1972. All launches were against explicit learning objectives, and many produced arrays of unexpected lessons that shaped the next experiments. The UTF emulated this approach by jumping to experimentation with users, existing technologies and explicit hypotheses before any development was initiated. This was contrary to traditional procedures.

4. Optimize for discovery and speed, not for efficiency or scale. Clayton Christensen defined the difference between crux evolutionary innovation and disruption as seeking to improve the current business model versus searching for a new one. Improving a business model is the realm of process improvement such as Lean Six Sigma or the Toyota Production System where finding and scaling efficiencies is paramount.

Searching for a new business model requires nearly the opposite behavior. The key insight is that disruption has an unknown end state. Thus, it involves an iterative search for true customer pain points and rapidly iterating potential solutions. This requires different processes, risk tolerances, organizational configurations and cultures.

To this end, the UTF rejected a one-size-fits-all innovation approach. In one example, the UTF’s primary contributions were shepherding and funding incremental experimentation to eventually hand off to others. In another, it was distilling insight from an operational problem, finding relevant solutions in other operational communities, and shepherding the matchmaking process. In yet another, it was a year of daily, small-scale skirmishes against political and organizational antibodies to allow the disruption sufficient time to prove itself on its own merits.

While these activities were directly driven by the UTF’s standardized innovation process, the tactical execution was adapted to the unique nature of each operational problem and user base.

Disruptive innovation inside any large organization is extremely hard. Few organizations have shown they can sustain disruptive innovation over time. Most militaries excel at disruption during wartime but struggle in absence of an existential threat.

Nevertheless, the DoD has recent successes to leverage for Replicator. The question remains: Which path will the DoD choose?

Jason Stack is the chief technology officer and co-founder of a dual-use maritime logistics startup and a senior adviser at the consultancy BMNT. He co-founded the U.S. Navy’s Unmanned Task Force and previously served as the deputy director.

More In Opinion