What Aircraft Carriers, Fighter Pilots, and Software Teams Know About Campaigns That Political Consultants Don't

Part of The Build — Foundations. Start with the the guide for the full map.

Almost everything written about campaign organization starts from campaigns. It looks at what past campaigns did, sorts the winners from the losers, and tries to extract a pattern. That’s useful and most of what I’ve published so far comes from doing exactly that. But it has a blind spot in that campaigns are a young, chaotic, badly-documented domain to learn management theory from. Other fields have been solving the same question. How do you make good decisions fast, under pressure, with incomplete information when the cost of a mistake is severe?

What I found is that every major failure mode in presidential campaign organization—slow decisions, informal authority overriding expertise, teams asked to own more than they can hold, etc.—has been studied extensively in other high-stakes domains. None of those fields call it campaign management. All of them have better answers than campaign management usually produces.

Before I built anything campaign-specific, I studied five of those fields to see what held up when applied somewhere else. What follows is the part of this project that isn’t about campaigns at all until the end.

Modern military doctrine, particularly the US Army’s, runs on the concept of mission command which I learned is descended from a much older Prussian tradition of Auftragstaktik. The core idea is that a commander states intent, the objective and the reasoning behind it, and then deliberately doesn’t specify exactly how subordinates should achieve it. Subordinates are trusted and expected to exercise initiative and adapt to conditions on the ground that the commander can’t see from further back. The alternative is detailed, centralized orders that try to anticipate every contingency. That approach breaks down the moment reality diverges from the plan. In any complex and adversarial environment, that moment comes almost immediately.

The campaign translation is more direct than it might sound. A field organizer in a county the campaign’s national HQ has never visited is in exactly the position mission command was designed for. That organizer is closer to the ground truth than anyone above them, needing to act faster than a request-and-approval cycle back to headquarters could support. Campaigns that centralize too much and that require field staff to get sign-off on decisions the staff on the ground actually understands better than headquarters does are recreating the exact failure mode mission command doctrine exists to prevent.

The caveat is mission command works because of years of shared training and a deeply internalized doctrine about what “good initiative” looks like. A campaign has far less time and a staff assembled from a dozen different professional backgrounds. The trust and shared judgment mission command assumes has to be built fast and deliberately and not assumed to already exist. This is where a lot of “empower your field staff” advice in campaign circles is under-specified. Empowerment without shared doctrine about what good judgment looks like isn’t mission command. Telling people they’re trusted without ever building the shared vocabulary for what trustworthy judgment means in this specific campaign, this specific state, this specific moment, isn’t delegation.

High-reliability organization theory comes out of studying institutions that operate in unforgiving conditions including aircraft carrier flight decks, nuclear power plants, and air traffic control. These organizations manage to avoid catastrophic failure almost all the time despite enormous complexity and large stakes for error. The research, most associated with Karl Weick and Kathleen Sutcliffe, identifies a specific set of habits these organizations share:

• A preoccupation with failure and treating near-misses as data instead of embarrassments to bury.

• A reluctance to accept simple explanations for why something almost went wrong.

• Sensitivity to actual frontline operations rather than dashboards and reports.

• A commitment to building resilience rather than just preventing every possible failure.

• Deference to expertise over hierarchy in a crisis, meaning the person with the most relevant knowledge in the room gets to drive the decision and not necessarily the person with the most senior title.

That last one is an uncomfortable fit for how most campaigns are run where seniority and proximity to the candidate often override actual domain expertise when something goes wrong fast. Deference to relationship instead of expertise is an HRO failure and it often happens when it matters most.

The caveat is HROs operate the same well-defined task repeatedly, for years, with enormous resources devoted to training for exactly the failure modes they’re guarding against. A campaign is a novel, one-time, rapidly scaling organization with no equivalent runway to build that muscle. The principle transfers. The infrastructure to actually achieve it in a shortened timeline isn’t automatic and it has to be built with intention.

RAPID—Recommend, Agree, Perform, Input, Decide—is a decision-rights framework most associated with Bain & Company and it was built to solve a specific corporate pathology of decisions that stall or get quietly re-litigated because nobody formally agreed on who has authority over what. It forces an organization to explicitly assign who recommends a course of action, whose agreement is required before it moves forward, who’s actually accountable for executing it, who should be consulted for input without having a veto, and who has final decision authority when the group can’t reach consensus.

This maps almost one-to-one onto the shadow-power-center problem that shows up in nearly every campaign in this project’s archive. Informal influence that isn’t captured anywhere in a decision-rights framework becomes untraceable and unaccountable. A campaign that built out an explicit RAPID-style matrix as something senior staff used to resolve disputes would have an answer to exactly the ambiguous-authority failure mode that recurs across the whole 2020 field. None of them seemingly did this in a rigorous and sustained way. It’s one of the more actionable and most underused ideas in this entire foundational layer.

Team Topologies comes out of software engineering, from Matthew Skelton and Manuel Pais, and addresses a problem that looks technical but isn’t really: how do you draw boundaries between teams so that each team can actually get its work done without constantly waiting on, or stepping on, another team’s work? Their framework identifies four team types, stream-aligned teams doing the core ongoing work, enabling teams that help other teams build capability, complicated-subsystem teams that own a piece of deep specialized complexity, and platform teams that provide shared infrastructure everyone else builds on, plus a strong emphasis on cognitive load: a team given more to own than it can hold in its head will underperform regardless of talent or effort.

Campaigns build their functional structure—political, communications, field, digital, finance, data, etc.—almost entirely by convention rather than by asking whether the boundary between any two functions is reducing coordination cost or creating it. The cognitive-load concept in particular is one I don’t think campaign organizational thinking has fully absorbed. A field department asked to own both traditional canvassing and a from-scratch distributed-organizing build simultaneously is being handed two different disciplines under one roof. The resulting strain might show up as underperformance in both. The direct translation isn’t “adopt this exact team taxonomy” because campaigns aren’t software organizations and stream-aligned/platform/enabling doesn’t map cleanly onto political functions. What transfers is the underlying discipline of drawing functional boundaries around what a team can hold in its head at once and treating every function you bolt onto an existing team as a cost and not a free addition.

The OODA loop—Observe, Orient, Decide, Act—comes from fighter pilot combat theory, developed by the military strategist John Boyd. Boyd’s central insight wasn’t really about any single stage of the loop so much as it was that victory in a fast-moving adversarial environment goes to whoever can repeatedly cycle through the full loop faster than their opponent and force the opponent to keep reacting to an outdated picture of reality.

This is the most directly load-bearing import in the whole foundational layer for the specific campaign phenomenon that is rapid response. A campaign’s ability to observe an attack or a fast-moving news event, orient it correctly (not just fast, but correctly, Boyd was explicit that a fast wrong orientation is worse than a slower correct one), decide on a response, and act faster than the story or the opposing campaign’s narrative can harden is close to a direct OODA-loop problem. The compact decision room this project keeps returning to as close to invariant is both good for strategic planning and a structural requirement for a fast OODA loop. Every additional person or approval layer in the decision path adds latency to the cycle.

The caveat is Boyd’s loop assumes an adversarial, zero-sum framing that fits crisis response well and fits a campaign’s day-to-day operations less well. Not every campaign decision is a dogfight and treating routine strategic choices with wartime urgency is its own failure mode. Decision fatigue and burnout is where the next field comes in.

The Job Demands-Resources model, from organizational psychologists Arnold Bakker and Evangelia Demerouti, offers a simple but rigorously studied framework. Burnout results when the demands placed on someone consistently outstrip the resources, time, support, autonomy, and recovery available to meet them and the relationship isn’t linear. Past a certain point, additional tires people out and produces the kind of disengagement and error-proneness that actively degrades organizational performance.

This is the field that campaign organizational research has thought about least and it shows. Campaign staff—especially field organizers—routinely operate under exactly the conditions JD-R research identifies as producing burnout at scale. High demand, compressed timelines, and minimal structural recovery built into the role. None of the source material I worked from treats staff wellbeing as a structural design variable rather than a culture-and-individual-resilience question. I think that’s a gap and probably the most underexplored question this foundational layer points toward.

“Burnout is bad” undersells the mechanism. A field organizer who churns out mid-cycle leaves a staffing hole and takes a network of volunteer relationships with them that the campaign spent months building and can’t reassign with ease to a replacement. The trust an organizer builds door by door, call by call, doesn’t transfer with a new hire’s start date. JD-R research is specific about why this happens. Rather than the simple view that people run out of willpower, it’s moreso that sustained high demand without matching resources produces a measurable decline in judgment quality and error rate before it produces outright quitting. That means the cost shows up as degraded performance well before anyone notices the turnover. A campaign treating staff sustainability purely as a culture question—be nice, offer snacks, thank people publicly, etc.— is treating a structural design problem as a morale problem.

The two require different fixes. The structural fix looks like what JD-R research actually recommends with real autonomy over how work gets done and recovery periods built into the calendar instead of treated as a luxury for whenever things calm down. On a campaign, there is no calm. Staffing ratios set with burnout thresholds in mind is also something to think about. None of the campaign-organization writing I worked from treats any of this as an org-design lever on par with decision rights or team boundaries. But it should.

Not everything here transfers cleanly and I’d rather say so directly than pretend five outside fields map perfectly onto a domain none of them were built for. Mission command and OODA both assume forms of shared training and adversarial urgency that a campaign has to manufacture in months rather than inherit over years. Team Topologies’ specific taxonomy doesn’t translate literally. What does transfer across all five is a shared discipline. Treat organizational design as something you test and interrogate instead of something you assume. Favor explicit decision rights over informal deference. And name the actual cost—in coordination, in latency, and in human sustainability—of every structural choice a campaign makes. Org design isn’t free. Most campaigns treat it like it is.

Everything else in this series sits on top of this layer.