Following through on a campaign promise, president-elect Donald Trump recently appointed Elon Musk and Vivek Ramaswamy to co-lead an effort they are calling the Department of Government Efficiency, or DOGE. While it is not actually a government department, the entity is likely to prove influential within the Trump administration and the Republican-led Congress—at least initially. Its mandate is to cut federal spending, and Musk and Ramaswamy have made clear that the Pentagon is in their sights. “The Pentagon recently failed its seventh consecutive audit,” the co-leaders wrote in a recent Wall Street Journal op-ed, “suggesting that the agency’s leadership has little idea how its annual budget of more than $800 billion is spent.”
Previous DoD efficiency initiatives have tried, and largely failed, to trim spending. This time will be different, one could argue, because Elon Musk is involved. He has done things in industry, such as revolutionizing space launch and electric vehicles, that many thought were impossible. To achieve these unlikely feats, Musk developed a ruthless approach that he calls “The Algorithm.” As recounted in Walter Isaacson’s biography, the Algorithm consists of five sequential steps: 1) question every requirement; 2) delete any part or process you can; 3) simplify and optimize; 4) accelerate cycle time; 5) automate. As Musk tries his hand at public policy, can this approach work?
Musk’s Algorithm has largely, and perhaps exclusively, been applied to high-technology and manufacturing-intensive sectors of industry. While the arm of DOD that develops and procures weapons fits this mold, acquisition consumes only about one-third of the defense budget in any given year. The other two-thirds of the budget go toward labor (military and civilian employees) and operations (training, military exercises, routine peacetime operations, housing, and military construction).
The first step of the algorithm—question every requirement—would be helpful across all parts of the military, from the weapons it acquires to the processes it uses to evaluate people for promotions. Too often, military requirements are passed from generation to generation without re-examining the assumptions behind why things are done the way they are. One of Musk’s trademark moves is to assign a name to every requirement—the specific individual behind it, not a group or department (such as “legal”). In the case of DOD, if the requirement was approved through the joint requirements oversight council, the chair of the JROC or a delegated representative should be the one who defends it. If the requirement is from a combatant command or military service, they should provide a representative to defend it. One of the challenges for DOD is that many of its requirements are external—that is, dictated by Congress or some other part of government, such as the Office of Personnel Management. For such requirements, the lawyer or other individual responsible for interpreting how the requirement applies to DOD should be the one to defend it.
The second step is to delete the parts and processes associated with unnecessary requirements. This sets up a somewhat adversarial environment that can be uncomfortable to many but ultimately productive in the end. Former Army Secretary (and later Defense Secretary) Mark Esper used something along these lines that became known within the Army as “Night Court.” This step is a massive undertaking because inefficiencies in the Defense Department are not always large programs or big pots of funding that can be easily identified and eliminated. (Sorry, there is no billion-dollar budget for diversity programs.) The real inefficiencies are spread thin and wide across a vast bureaucracy, and making the tough calls of what to delete and what to keep will require the active involvement of the next Defense Secretary. This is not something that can be delegated to lower levels if you want decisions to stick.
The main risk with step two is that the effects of deleting a requirement can be complex, opaque (especially for old requirements), and non-linear. A requirement that seems outdated or wasteful may have enormous consequences under certain circumstances. The practitioners of the Musk Algorithm should understand that some inefficiencies in the military are by design: they are strategic choices that maintain options in unlikely but highly consequential scenarios. Musk admits that he errs on the side of deleting too many requirements, arguing that “if you do not end up adding back at least 10 percent of them, then you didn’t delete enough.” How do you know what needs to be added back when failure is not an option? In the military, every deletion could put lives at risk.
The third step is to simplify and optimize, which is where the algorithm starts to break down as it applies to the military. To optimize, one needs an objective function: a metric by which the relative value of different alternatives can be compared. In industry, the objective function could be the time it takes to build each item on a production line, the unit cost of an item, total profit, total revenue, or some other quantifiable outcome that the optimization is intended to improve. Importantly, the validity of an optimum solution is subject to testing—you can try it out and see if it works before committing. The military is different because its ultimate purpose—to protect the nation—is based on value judgements about threats, risks, and sufficiency. How secure does the nation want to be and against what types of threats depends on who you ask and what they value. And while the military can test some changes in operations and processes on a small scale, it has little ability to test major initiatives, such as an alternative strategy, without causing irreversible change. For example, one cannot “test” what pulling back from NATO means for national security without undermining our credibility with allies, which may take years or decades to restore. Even decisions like delaying a weapon system or temporarily reducing peacetime presence missions can have irreversible effects because they create a window of opportunity an adversary can exploit.
Practitioners of Musk’s Algorithm should again be mindful that many inefficiencies in DOD—especially the ones that are government waste at its finest—are often there by design because it creates something else people value. Inefficiencies maintain jobs in someone’s congressional district, which is why they persist and are so difficult to eliminate. Even if a solution can be negotiated among stakeholders to reduce the most egregious inefficiencies, the system does not stay optimized for long because the military operates as a dynamic open system subject to innumerate external factors, including the actions of adversaries and allies and the whims of domestic politics.
Steps four and five of the Algorithm—accelerate cycle time and automate—are highly relevant to the military, but they depend on having simplified and optimized in step three. Otherwise, you will be accelerating and automating processes that remain inefficient. While this may still be a marginal efficiency improvement over the status quo, it could further entrench and legitimize unnecessary processes, making them even more difficult to reform in the future.
Where step three can be accomplished using value judgments negotiated across competing political interests, steps four and five can create important military advantages. The military desperately needs to accelerate the pace of innovation in its weapons programs and the time it takes to make policy, administrative, operational, and tactical decisions. Moreover, the ability to automate processes in the back-office functions of the Pentagon (e.g., filing and reviewing expense reports) and at the tactical edge (e.g., sorting through terabytes of sensor data in real time) frees up people to focus on the cognitive and value judgement elements of warfighting and planning—areas where humans are most valuable. Automation is not a replacement for humans or human judgement; it is an enabling tool that allows humans to be more effective and efficient.
Nearly everyone in Washington recognizes that the Pentagon is in dire need of reform, including many of the people that lead or have led the Pentagon. Defense reform is not and should not be a partisan issue, but it cannot help being a political issue because politics is about the allocation of resources within society: who gets what. The budget is a key tool for reforming defense, but reformers should remember that the budget is ultimately an expression of political values. How much security and what kind of security the nation wants is a value judgement made by elected and appointed leaders that represent the people. If the people elect politicians who talk about peace through strength and eliminating waste and inefficiency but then act in ways that are contrary to these ideas, one can only conclude that the people must not really value these things. In a democracy, the people get the government they deserve, as ineffective and wasteful as that may be.
