Admiral Yamamoto Isoroku, the architect of Japan’s attack on Pearl Harbor, predicted the pivotal rolethat industrial overmatch would play in the outcome of World War II. Proving him right, America produced a staggering 300,000 aircraft during the war compared to Japan’s meager 76,000.
America won World War II not through tactical prowess or strategic genius, but on the back of its unparalleled industrial capacity. As the U.S. military reorients itself toward great power competition, it can learn a lot from that experience.
Japan sought to overcome its industrial shortcomings with higher-quality personnel and equipment. The U.S. Navy found itself completely outmatched in the early Guadalcanal Campaign, and the ships and aircraft Japan started the war with were often superior. Those advantages won many early victories. But early victories were not enough to overcome the productive and innovative American economic machine.
Fast forward 70 years, and worries about America’s modern industrial capacity are commonplace. Those fears are real, but the problem is manageable. In the manufacturing sectors that matter most for modern warfare, America is still a global leader: think aerospace or energy.
What America cannot do quickly and at scale is expand and replace the human capital necessary to operate the output of its economic engine. It took the United States about nine months to train a qualified pilot during World War II. Today, it takes about 4-5 years. World War II only lasted around 4 and a half years. World War II-era submarine officers only spent about 3 months training before joining the fleet. Modern-day submarine officers take about 14 months from commissioning to joining the fleet if they have no breaks, though this includes nuclear training. To affect a marked force buildup and replace losses in a great power war, the U.S. military would need to find a way to dramatically reduce this time.
America is relying on superior equipment and laboriously trained personnel to win a large conflict. As we’ve seen in both Gulf Wars, the initial invasion of Afghanistan and the campaign against ISIL, complex weapon systems, manned by professionals who are costly to train and retain, will win battles and small wars. But they are not likely to win the kind of war that alters great power balances. In these conflicts, small, unscalable forces that are elite at the tactical level will be less important than sustainable processes at the strategic level that ensure that effective conventional forces can be continually fielded.
While U.S. naval and air forces are not as scalable as they once were, there is a clear route to restoring that capability: Make the job easier with software automation that empowers operators through a “mission command” model.
Lt. Gen. George Kenney remarked in 1943 that Japanese forces could “hold their own in any league, but … simply cannot train Airmen to compare with ours in a hurry. [Japan’s] original highly trained crews were superb, but they are dead.” The side that can keep fielding effective forces after suffering losses will have the advantage in a full-scale war.
Reducing Cognitive Complexity
When facing difficult situations, business executives can choose from two decision-making models. They can lock themselves in a room, brainstorm a number of possible solutions, analyze them laboriously, and then decide on the best one. Or they can direct their staffs to analyze the problem, cultivate solutions, and then present the solutions to the executive so they can make an informed decision. Successful executives choose the latter model, which allows them to make consistent, rapid decisions about topics over which they don’t have complete mastery.
At the operational and strategic levels of warfare, the military operates in a similar fashion. Buton the opposite end of the spectrum, the U.S. military still expects its tactical-level operators to act as staffs and executives wrapped into one. It was possible for inexperienced naval aviators in World War II to do this when they had only to master their simple flight controls and weapons to fight against similarly simple adversaries. There are only so many ways to attack a ship with a dumb bomb. Today, however, naval aviators must operate long-range sensors to cue dizzyingly complex weapons in contested electromagnetic environments against stealthy adversaries. It takes a great deal more time to train someone to make decisions and execute tactics successfully within this complexity.
Enabling software to handle the burdensome task of formulating courses of action and then presenting the operator with the best ones would help restore simplicity to complex modern warfighting tasks. This would make training new operators much quicker and allow them to maintain their skills with much less practice. To do this, the Department of Defense should change the way it develops software. There is a place for existing purpose-built acquisition organizations, such as NAVAIR, DIUx and others, which is to handle the complex, scientifically cutting-edge projects that involve hardware or machine learning. But the vast majority of software development that’s required to build an effective force doesn’t fall into that category. Skilled warfighters can and should handle the more basic levels of software development. They understand best what’s needed. Delegating to second-party entities when not absolutely necessary often results in solutions that don’t solve problems.
Most of the needed improvements are not overly complex. A single warfighter with just rudimentary programming skills is fully capable of developing tactical decision aids that solve many of a community’s tactical problems. This isn’t hypothetical. The Maritime Patrol and Reconnaissance community, which flies the Navy’s new P-8 Poseidon anti-submarine warfare aircraft, has been successfully experimenting over the last two years with this new paradigm.
I’ve led a group of fleet operators in building a community-grown piece of software, called iLoc, that solves dozens of tactical problems for P-8 operators. We train fleet operators with a computer science background to develop software applications for iLoc that can then be used by anybody on one of the U.S. Navy’s P-8s. This group of trained developers can develop an idea for a new application, build a prototype, and deploy it against a real target in the course of a day. Continuous experimentation, feedback, and iteration results in a steady stream of useful products developed at incredible speed.
After analyzing a problem mathematically, iLoc presents the operator with the best courses of action available. One example of a problem it solves is determining the optimal pattern of sonobuoys – or floating microphones that listen to submarines – the aircraft should deploy to maintain track of a submarine. The program accounts for the ability of the P-8 to maneuver in position to deploy sonobuoys, the time it takes for sonobuoys to become operational, the detection ranges of the sonobuoys, and the speed of the target to recommend patterns that will maintain contact with the target while utilizing no more buoys than available. P-8 operators used to solve these problems with imprecise mental math or hand-drawn, slide-rule-esque charts, generating solutions that were often adequate, but never optimal. Now, iLoc can solve them with the press of a button and recommend a truly optimal set of solutions to the decision-maker. And the software is continuously updated by actual fleet operators as they learn lessons and innovate new tactics.
While a good first step, this software runs in a web browser that is available on the primary software interface on the aircraft but is not fully integrated within it. In the sonobuoys example, this means the user must manually type in values for the number of buoys available, detection ranges, and target speeds. This makes a completely seamless user experience difficult to achieve. The next step is to develop a scripting capability — or the ability to create functionality in another software environment — that warfighters can access in the P-8 and other platforms. This would allow user-built tools to automatically access data generated by the main mission system, unleashing unprecedented tactical advancement and easing training difficulties. Instead of training repetitively to execute onerous tasks in stressful environments, talented operators could train their systems to handle much of the work themselves. This would dramatically reduce initial training times and make it easier to retain skills with less practice.
Scripting Tactical Mission Command
The German army of World War II revolutionized modern warfare with its relentless application of a philosophy called mission command. Rather than tell subordinate commanders how to do their jobs, their superiors would simply tell them what end state they required. It was up to the subordinate at the critical point of action to decide how best to use their local resources. The sum of these small, distributed decisions was a far more agile force than would have existed had decision-making been centralized under a small staff. Within this paradigm, high-level decision-makers do not have to be experts in the tactical application of each specialized unit under their command. Nor need they develop extravagant plans that account for every contingency. They need only be experts at developing simple plans of action and communicating intent.
The tactical level of naval and air warfare is due for a similar revolution. While software can offer a menu of tactical responses, other measures can make it easier for the operator to enact those responses. Think of it as mission command on a micro-scale. Currently, if an operator wanted to attack an adversary’s naval platform, iLoc could recommend courses of action to best accomplish that. But the operator would still have to:
- Choose to attack
- Present the situation to the tactical decision aid software
- Choose a course of action
- Analyze information gathered by the system
- Properly pre-set a dozen settings in the weapon
- Select an appropriate aimpoint
- Ensure their own platform is at the correct altitude and airspeed
- Continue to track the target
- Prepare for a response from the target
On the P-8, each of these actions requires extensive training and practice. But if any operator could write scripts and draw data from the underlying system via an Application Programming Interface (API), they could automate steps 2-9. All that would be left for subsequent operators to do is:
- Decide to attack
- Select a general course of action from a menu of options
- Monitor the automation so they can intervene if something is awry
Instead of micro-managing their weapons platform, the operator can convey their intent and then go back to decision-making.
It’s important to note that realizing these gains doesn’t require machine learning or data analytics. This is just allowing fleet operators to do the fairly simple programming that, for example, a regular employee at an investment bank would be expected to do in Excel on a daily basis. It is simply translating the job a human already does into the computer’s language.
Automating execution of cutting-edge tactics will also allow a large force of warfighters to rapidly acquire new skills as soon as others develop them. Right now, when someone develops a new tactic or technique, experienced operators must seek out this knowledge and train themselves . They often choose not to learn the new methods because it costs too much in time and resources. Since they are the ones training their less experienced colleagues, innovation can be quickly stifled. However, if the new tactic or technique were accessible through a single button click in the software, operators could train themselves by reading a vetted article explaining the theory and where to access the functionality. This would lead to a renaissance of tactical innovation across the military.
The best part of this solution is its cost: near zero. Talented coders are already serving in front-line units across the U.S. military. Their skills are just not being utilized because, unlike hardware, software is viewed by the military as too complex for its own people to build and manage. By investing in software flexibility up front, the military can unleash the creative capacity of American forces to build the most usable and effective combat platforms possible. Continually improving user interfaces will increase the U.S. military’s capacity to train quickly and retain readiness longer.
The first concrete step is to require platforms’ software interfaces to have scripting capabilities and APIs. While it’s true that the firms that build these systems may object to increased involvement of operators at lower levels, my team’s experience with the P-8 shows that it’s possible to get buy-in for a more decentralized approach. The P-8 is a sensible first test platform as it is already transitioning to an application-based architecture for its existing interface and has a community of developers ready to take advantage of software flexibility. With demonstrated success, alumni of the program can help spread the paradigm across the rest of the fleet.
The second step is to build the infrastructure to manage this process. Warfare development centers like TOPGUN and the Maritime Patrol and Reconnaissance Weapons School can manage the teams that create and curate these scripts. They can vet application effectiveness and publish articles detailing their theory and use. To compensate for their time managing these programs, warfare development centers would need to spend much less time training front-line units on new tactics and techniques. Like today, unit commanders could always elect to forbid their subordinates from using a tactic they see as risky or ineffective.
It’s a good bet that 21st century great power conflict will be different from the world wars of the 20th century. But there is no doubt that force-structure scalability will remain critical. Enabling U.S. naval and air forces to build ease, simplicity, and effectiveness into their platforms will restore that scalability and allow for continual tactical improvement well below the cost curves of their adversaries.