Robert X. Cringely buried one of the sharpest organizational insights of the twentieth century inside a book about the personal computer industry. In Accidental Empires (1992), the journalist and former Apple employee described a pattern he'd watched destroy company after company in Silicon Valley: every venture requires three distinct types of people at three distinct stages, the three types despise each other, and the company that fails to swap groups at the right moment either stalls or dies.
Commandos go first. They're the small, obsessive team that takes the beach — the handful of people who thrive in chaos, build under impossible constraints, and ship the first version of something that shouldn't exist yet. Commandos don't care about scalability, documentation, or org charts. They care about proving the thing can work at all. They code at 2 a.m., make decisions in hallways, and hold the entire system in their heads because there is no other record of it. The original Macintosh team — twenty engineers in a building with a pirate flag — was a commando unit. So was the team that built the first version of PayPal, WhatsApp's 55 engineers serving 450 million users, and the two Collison brothers who shipped Stripe's payment API from a single apartment.
Infantry arrives second. Once the commandos have taken the beach, a larger force is needed to occupy and consolidate the territory. Infantry builds the systems that commandos refused to build: hiring processes, quality assurance, documentation, repeatable sales motions, customer support infrastructure. Infantry takes the commando prototype and turns it into a product that works for millions of people who don't share the commando's tolerance for rough edges. The infantry doesn't take the hill — they build the supply chain, the field hospital, and the road network that makes holding the hill possible.
Police arrive last. Once the territory is occupied and the systems are built, you need people who maintain order, optimize what exists, and prevent the organization from regressing into chaos. Police write the compliance manuals, enforce the processes, and extract efficiency from the systems the infantry built. They're the operations leaders, the finance professionals, the regulatory experts, the HR architects who make a company run at scale. At 10,000 employees, you need police more than you need commandos. The Fortune 500 runs on police.
The tragedy Cringely identified is structural, not personal. Commandos look at infantry and see bureaucrats who can't ship. Infantry looks at commandos and sees reckless cowboys who leave messes. Police look at both and see liabilities. Each group is right about the others' weaknesses and blind to their own limitations. The commando who was indispensable at 5 people becomes a liability at 500 — not because they changed, but because the company did. The police officer who optimizes a mature organization would suffocate a startup. The infantry leader who builds scalable systems would never have taken the beach in the first place.
Most companies fail at the transitions. The founding commandos refuse to yield control — they built this thing, and they'll be damned if some process-obsessed MBA is going to bureaucratize it. Or the board, terrified by the chaos of the commando phase, brings in police before the infantry has finished building. Or the infantry, comfortable with their authority, resists the police who threaten their autonomy. Each transition requires the people who were heroes in the previous phase to accept that their skills are no longer the binding constraint — and human ego rarely accommodates that gracefully.
Section 2
How to See It
The model reveals itself whenever a company's culture visibly clashes with its current stage. The tell is not performance metrics — it's friction. When the behaviors that made the company successful start creating problems, you're watching a phase transition in progress. The question is whether leadership recognizes the transition or fights it.
You're seeing Commandos vs Infantry vs Police when the people who built the original product are being asked to do work that bores them — documentation, process design, stakeholder management — and they're either doing it badly or refusing entirely.
Technology
You're seeing Commandos vs Infantry vs Police when a startup's founding engineers resist hiring a VP of Engineering because "we don't need management." At ten engineers, they're right — the commando team communicates through osmosis. At fifty engineers, the absence of infantry leadership means no code review standards, no incident response protocol, and no onboarding process. New hires spend three weeks figuring out where the documentation should be. The commandos' refusal to accept infantry isn't strength. It's the organizational equivalent of refusing to build a road because you enjoy bushwhacking.
Business
You're seeing Commandos vs Infantry vs Police when a fast-growing company hires a COO from a Fortune 500 company and the culture rebels. The COO introduces quarterly planning cycles, approval workflows, and performance review frameworks — police-phase tools deployed during the infantry phase. The founders complain that "we've lost our startup energy." The COO complains that "these people have no discipline." Both are correct. The failure is timing, not talent.
Investing
You're seeing Commandos vs Infantry vs Police when a startup's Series A deck features a founding team of five brilliant engineers and zero operational leaders. The product is remarkable. The question every investor should ask: does this team understand they'll need infantry within twelve months, or do they believe commandos can scale forever? The answer determines whether the investment thesis depends on a team transition that the founders may actively resist.
History
You're seeing Commandos vs Infantry vs Police when a revolutionary movement that was brilliant at overthrowing the old regime proves catastrophically incompetent at governing. The commandos who won the revolution — skilled at disruption, inspiration, and rapid improvisation — are precisely the wrong people to build institutions. The American Revolution succeeded in part because its commandos (the revolutionary generals and pamphleteers) yielded to its infantry (the constitutional framers) within a decade. Revolutions that never manage the handoff — where commandos govern indefinitely — tend toward authoritarianism or collapse.
Section 3
How to Use It
The model's value is diagnostic. It tells you which phase your organization occupies and whether the current team matches the current need. The hardest part is not the diagnosis — it's the willingness to act on it.
Decision filter
"Ask: is the primary constraint right now taking new territory, consolidating existing territory, or optimizing a stable system? The answer tells you which group should be running the operation — and whether the current leaders belong to that group."
As a founder
Accept — early and honestly — that you are probably a commando. You built the company because you could operate in chaos, take risks that salaried managers wouldn't, and ship without permission. Those traits got you here. They will not get you to the next stage alone. The founders who build enduring companies either evolve their own skillset or step aside for the people whose skills match the new phase. Steve Jobs was fired from Apple in 1985 because the board decided a commando couldn't run an infantry-phase company. They were partially right — and the decade Jobs spent at NeXT and Pixar taught him the infantry and police skills he lacked the first time. His return in 1997 was the rare case of a commando who learned to command all three phases.
The practical move: hire your infantry leaders before you need them. If you wait until the commando chaos is causing visible damage — lost customers, missed deadlines, departing engineers — you've waited too long. The best time to bring in your first VP of Engineering, your first head of operations, or your first finance leader is when the commandos are still succeeding but the cracks in scalability are becoming visible.
As an investor
Map every portfolio company to its current phase and ask whether leadership matches. The most common failure mode in venture-backed companies isn't product-market fit — it's the commando team that can't transition to infantry. The founder who can't hire, can't delegate, and can't build systems will hit a ceiling that no amount of capital can raise. During diligence, probe the founder's self-awareness about their own limitations. The founders who say "I need to hire people better than me at operations" are investable. The founders who say "we'll figure it out — we always have" are telling you they'll resist the transition until it's too late.
The highest-leverage board intervention is often the simplest: helping a commando founder hire their first infantry leader without framing it as a demotion.
As a decision-maker
Audit your team composition against your company's current phase. If you're in the infantry phase and your leadership team is still entirely commandos, you have a structural mismatch that will express itself as chaos at scale — missed deadlines, inconsistent quality, and an inability to onboard new people effectively. If you're in the commando phase and your leadership team is police, you have the opposite mismatch — excessive process choking the speed that early-stage survival requires.
The most dangerous configuration: police running a company that still needs infantry. This happens frequently in turnarounds, where the board hires an operator from a mature company to "professionalize" a growing startup. The operator installs police-phase systems before the infantry-phase work is complete, and the company ossifies before it has finished scaling.
Common misapplication: Treating the three phases as sequential and irreversible. Large companies frequently need commando units for new product lines even while the core business runs on police. Amazon operates this way deliberately — small, autonomous two-pizza teams (commandos) launch new products within a structure managed by infantry and police. The phases coexist within a single organization; they're not a one-way progression.
Second misapplication: Assuming commandos are inherently more valuable than infantry or police. The commando myth — the cult of the scrappy founder — is Silicon Valley's favorite self-delusion. Commandos take the beach. Infantry holds it. Police makes it profitable. The commando who can't yield the territory to the people who will make it productive has destroyed as many companies as the MBA who never should have been put in charge of innovation.
Third misapplication: Firing commandos instead of redeploying them. The best organizations don't discard commandos after the beach is taken — they point them at the next beach. Apple kept its commando engineers by continuously opening new product fronts: iPod, iPhone, iPad, Watch. The commando who would suffocate in an optimization role thrives when aimed at the next unsolved problem.
Section 4
The Mechanism
Section 5
Founders & Leaders in Action
The pattern is clearest in companies where the phase transitions were public — where you can watch the personnel changes, the cultural shifts, and the organizational growing pains in real time. Both cases illustrate that the model's value is not in identifying the phases but in managing the transitions between them.
What separates these leaders is not that they avoided the tension between commandos, infantry, and police. It's that they recognized the tension as structural rather than personal — and designed their organizations to navigate the transitions instead of pretending they wouldn't happen.
Steve JobsCo-founder & CEO, Apple, 1976–1985 / 1997–2011
Jobs is the model's most dramatic case study because he lived both sides of the failure mode. In the early 1980s, Jobs was Apple's commando-in-chief — the visionary who drove the Macintosh team to build a product that redefined personal computing. But as Apple grew past a thousand employees and needed infantry-phase leadership, Jobs's commando instincts became liabilities. He feuded with the professional managers the board had hired, undermined the organizational structures that scaling required, and created chaos the company could no longer absorb. In 1985, the board sided with CEO John Sculley and removed Jobs from operational authority. The commando was extracted from a company that needed infantry.
The decade that followed was Jobs's infantry education. At NeXT, he learned the pain of building systems, managing supply chains, and operating within constraints. At Pixar, he watched Ed Catmull build one of the most disciplined creative organizations in history — a company that combined commando-level creativity with infantry-grade process. When Jobs returned to Apple in 1997, he was a different operator. He still had the commando's instinct for product and the willingness to bet the company on radical ideas. But he also had the infantry commander's ability to build teams, manage supply chains, and create structures that scaled. The iPhone team of 200 engineers operated with commando intensity inside an infantry structure that Jobs had learned to build. He kept Apple's commando culture alive by continuously opening new fronts — iPod, iTunes, iPhone, iPad — while building the operational infrastructure that let the company execute at global scale.
Bezos solved the transition problem by refusing to let it happen sequentially. Instead of cycling through commandos, infantry, and police at the company level, he built an organizational structure that ran all three phases simultaneously. The two-pizza team — a unit small enough to be fed by two pizzas — was Amazon's mechanism for preserving commando dynamics inside a company that eventually employed over 1.5 million people. Each two-pizza team operated with startup-level autonomy: they owned a problem, made decisions without waiting for organizational permission, and shipped independently.
The infrastructure surrounding those teams was pure infantry and police. AWS's operational rigor, Amazon's logistics network, the leadership principles that governed hiring and promotion — these were the systems that let commando-scale teams operate without commando-scale chaos. Bezos understood that the phases don't have to be sequential. They can be structural. You run commandos at the team level, infantry at the business-unit level, and police at the corporate level — all simultaneously, within one company. The key was interfaces: the teams were autonomous, but the APIs, the data formats, and the operational standards were non-negotiable. Bezos encoded the lesson in a mandate that became legendary: all teams must expose their functionality through service interfaces. No exceptions. The commando teams could build however they wanted — as long as the output conformed to the infantry's specifications.
Section 6
Visual Explanation
The three boxes represent the three phases, but the real content of the diagram is the bottom half — the danger zone. The transitions are where companies die. The commando-to-infantry transition fails when founders refuse to cede control to the system-builders. The infantry-to-police transition fails when the organization's growth culture calcifies into its optimization culture before the growth work is done. Both transitions require the currently dominant group to voluntarily reduce its own authority — which is why both transitions are so reliably botched. The companies that navigate both — Apple under Jobs's second tenure, Amazon under Bezos — do so because their leaders understood the model well enough to design the handoffs in advance rather than fighting them in real time.
Section 7
Connected Models
The commandos-infantry-police framework intersects with models that explain why early-stage behavior works, why it stops working, and what accumulates when the transition is mismanaged. The reinforcing connections show how commando-phase principles align with startup orthodoxy. The tension connection reveals the pathology that emerges when commandos refuse to leave. The leads-to connections trace the downstream consequences of failed or delayed transitions.
Reinforces
Do Things That Don't Scale
Paul Graham's advice to early-stage founders is a commando manifesto: recruit users one at a time, do manual work that won't scale, solve problems with effort rather than systems. The reinforcement is direct — the commando phase is the phase where unscalable behavior is correct. The Airbnb founders personally photographing every listing, the Stripe founders manually integrating each early customer — these are commando tactics that would be organizational malpractice at infantry scale. The model explains why Graham's advice has an expiration date: what doesn't scale is exactly right for commandos and exactly wrong for infantry.
Tension
Founder's Syndrome
Founder's Syndrome is what happens when the commando refuses to leave the beach. The founder who centralizes decisions, resists delegation, and treats every process as an affront to the original vision is a commando operating past their phase. The tension is painful because the founder's instincts were right — during the commando phase. The syndrome emerges when those same instincts persist into the infantry phase, where they create bottlenecks, drive away operational talent, and prevent the organization from developing the systems it needs to scale. The model reframes Founder's Syndrome not as a personal failing but as a phase mismatch.
Reinforces
10x Individuals/Teams
Commandos are 10x teams by definition — small groups producing outsized output through talent density and minimal coordination overhead. The 10x model explains why commandos outperform during the early phase: fewer people, faster decisions, deeper context per person. The reinforcement runs both directions. The 10x framework explains why commandos work; the commandos-infantry-police framework explains why 10x dynamics have an organizational shelf life. A 10x commando team that refuses to build infantry systems will eventually be outperformed by a well-organized infantry that trades peak individual output for reliable collective throughput.
Section 8
One Key Quote
"Whether we're talking about a startup or an army, the weights have to shift. Commandos take the beach. Infantry hold the territory. Police run the government. And each group has nothing but contempt for the others."
— Robert X. Cringely, Accidental Empires (1992)
Cringely delivered the framework as an aside — a few pages in a 300-page book about IBM, Apple, and Microsoft. The book is largely forgotten. The framework endures because every founder who has lived through a scaling crisis recognizes the pattern instantly. The commando-phase team that built the product resists the infantry-phase leaders who arrive to build the systems. The infantry-phase leaders who built the systems resent the police-phase operators who arrive to optimize them. The contempt isn't irrational. Each group genuinely lacks the skills the others possess and genuinely cannot understand why those skills matter.
The deepest insight is the one Cringely left implicit: the company's survival depends on managing a transition between groups that don't respect each other. This is not a problem that vision statements, team-building exercises, or All Hands meetings can solve. It's a structural problem that requires structural solutions — clear role transitions, intentional team composition, and leaders who understand which phase the company is in and which type of person should be running it. The founders who navigate this build companies that last. The founders who don't build companies that either stall at commando scale or collapse under the weight of premature bureaucracy.
Section 9
Analyst's Take
Faster Than Normal — Editorial View
The commandos-infantry-police model is the best diagnostic I know for organizational dysfunction. When a company is struggling — shipping slowly, losing talent, growing revenue but losing culture — the first question is always: which phase is this company in, and does the leadership team belong to that phase? The answer is wrong more often than any other diagnosis I run. The commando founder running a 500-person company. The police-phase CFO installed as CEO of a company that still needs infantry. The infantry VP of Engineering trying to manage a six-person commando team with Jira and sprint planning. The mismatch explains the dysfunction more reliably than any individual performance issue.
The hardest conversation in startups is telling a founder they've become the bottleneck. The commando who took the beach, who hired the first ten people, who wrote the original code — telling that person their skills are no longer the binding constraint feels like ingratitude. It's not. It's the most important transition the company will make, and the board that avoids the conversation out of loyalty to the founder's feelings is choosing sentiment over survival. The best founders understand this. Jobs understood it — eventually. Bezos understood it from the start and built the structure to avoid ever having to choose between commando energy and organizational maturity.
The pattern I see most in the current market: AI startups stuck in the commando phase. Brilliant founding teams of three to five ML engineers who have built remarkable technology and cannot hire, cannot build go-to-market operations, cannot create the customer success infrastructure that turns a research demo into a business. The commando phase produced something extraordinary. The infantry phase is where it will either become a company or remain a project. The founders who recognize the transition and hire for it will win. The founders who believe commando intensity is a substitute for organizational capability will stall.
One underappreciated version of the model: the reverse transition. Mature companies that need to innovate must reintroduce commandos into a police-phase organization — and the antibodies are fierce. Every established process, every compliance requirement, every approval workflow is optimized to prevent exactly the kind of rule-breaking that commandos require. The companies that manage this — Lockheed's Skunk Works, Google X, Amazon's new product teams — do it by physically and organizationally separating the commando unit from the police structure. The commando team operates under different rules, different reporting lines, and different success metrics. Without that separation, the police will kill the commandos before they take the beach.
Section 10
Test Yourself
The scenarios below test whether you can diagnose which phase a company occupies and whether its current leadership matches that phase. The key skill is resisting the instinct to judge the individuals involved. The model is not about good leaders and bad leaders — it's about the fit between a leader's strengths and the organization's current needs.
The most common analytical error: assuming the commando is always the hero and the police officer is always the villain. Silicon Valley culture reinforces this bias. The reality is that more companies die from commando excess — refusing to build systems, resisting process, treating every hire who isn't a founding engineer as a second-class citizen — than from premature bureaucracy.
Which phase — and does the team match?
Scenario 1
A startup with 12 engineers has strong product-market fit and is growing revenue 20% month over month. The founding CTO still reviews every pull request, makes every architecture decision, and personally handles production incidents at 3 a.m. Three senior engineers have quit in the past six months, each citing 'no room to grow' in their exit interviews. The CEO says the CTO is 'the most important person in the company.'
Scenario 2
A Series B company with 200 employees hires a VP of Operations from a Fortune 100 company. Within three months, the VP has introduced weekly executive reviews, a formal procurement process, mandatory quarterly OKRs for every team, and a 12-step approval workflow for new feature launches. Engineering velocity drops 40%. The VP argues that 'discipline is what separates real companies from science projects.'
Scenario 3
A 10,000-person enterprise software company creates an 'innovation lab' of eight engineers tasked with building new products. The lab reports to the Chief Innovation Officer, who reports to the CEO. After 18 months, the lab has produced three prototypes — all technically impressive, none launched. Each prototype stalled during the company's standard product launch review: legal compliance, security audit, accessibility review, pricing committee, sales enablement, and marketing approval. The CIO blames 'corporate bureaucracy.' The CPO blames 'a lack of rigor in the innovation lab.'
Section 11
Top Resources
The commandos-infantry-police framework sits at the intersection of organizational behavior, startup strategy, and military theory. Start with Cringely for the original model, read Horowitz for the operational reality of managing phase transitions, and study McChrystal for the military solution to scaling commando dynamics within large organizations. The academic literature on structural inertia and organizational ecology provides the theoretical foundation for why the transitions are so difficult.
The source text. Cringely's insider account of the personal computer revolution contains the original commandos-infantry-police framework — a few pages that have outlived the rest of the book. The model emerges from Cringely's observation of Apple, IBM, and Microsoft, but its explanatory power extends to every technology company founded since. Read for the framework. Stay for the portrait of an industry that was already repeating the pattern it couldn't see.
Horowitz's memoir of running Opsware is the best account of what the commando-to-infantry transition feels like from inside the CEO chair. The chapters on hiring executives, firing friends, and managing organizational change are case studies in the transitions that Cringely's model predicts. Horowitz doesn't use Cringely's terminology, but every chapter describes the pain of replacing one type of leader with another at the moment the company's phase demands it.
McChrystal's framework for scaling commando dynamics within a large military organization is the structural solution to the phase-transition problem. By connecting small, autonomous teams through shared consciousness while preserving empowered execution, McChrystal built an organization that operated with commando speed at infantry scale. The book provides the organizational design blueprint for companies that want to run all three phases simultaneously rather than sequentially.
The definitive case study of a commando unit preserved inside a police-phase organization. Kelly Johnson's fourteen operating rules for the Skunk Works — minimal reporting, small teams, direct authority, exemption from corporate bureaucracy — are a design manual for maintaining commando energy within mature companies. Rich's account of building the SR-71 and the F-117 demonstrates that the commando phase can be an organizational structure, not just a developmental stage.
Grove's management framework at Intel is the infantry commander's bible. His concept of high-leverage activities — identifying the actions that produce disproportionate organizational output — is the operational methodology that infantry-phase leaders need to build systems, manage teams, and create the scalable processes that commandos can't or won't build. Read alongside Cringely to understand what the infantry phase demands of its leaders.
Three phases of organizational growth — each requiring a different type of leader and a painful transition between them.
Leads-to
[Team of Teams](/mental-models/team-of-teams)
General Stanley McChrystal's framework for scaling special operations in Iraq is the military resolution of the commando-infantry tension. McChrystal's innovation was connecting commando-scale units (small, autonomous teams) through shared consciousness (infantry-grade information systems) while preserving empowered execution (commando-style decision authority). The result was an organization that operated with commando speed at infantry scale — the exact configuration that Bezos built at Amazon with two-pizza teams and service-oriented architecture. The model leads naturally to Team of Teams as the structural answer to "how do you preserve commando energy inside a large organization?"
Leads-to
Organisational Debt
Every delayed phase transition accumulates organizational debt. Commandos who stay too long leave behind systems held together with tape and tribal knowledge — no documentation, no tests, no onboarding paths. Infantry that resists police leaves behind inefficiencies that compound as the organization grows. Organisational debt is the residue of phase mismatches: the processes that should have been built during infantry but weren't, the controls that should have been implemented during the police phase but were resisted. The commandos-infantry-police model predicts where organisational debt will accumulate based on which transitions were botched.
Reinforces
[Pivot](/mental-models/pivot)
Pivoting is a commando skill. It requires the willingness to abandon a position, move fast, and rebuild under uncertainty — exactly the traits that define the commando phase. Infantry organizations resist pivots because they've invested in systems optimized for the current position. Police organizations are structurally incapable of pivoting — the compliance frameworks, the governance structures, and the operational playbooks all assume stability. The model explains why startups can pivot and large companies can't: pivoting requires commandos, and large companies have systematically replaced their commandos with infantry and police.
The meta-lesson: the right person at the wrong time is the wrong person. This is the model's sharpest edge and the one most leaders resist. It's not about good people and bad people. It's about the right people for the right phase. The commando is not inferior to the police officer. The police officer is not inferior to the commando. Each is essential — and each is essential at a specific moment. The leader who internalizes this stops taking transitions personally and starts treating them as engineering problems: identify the phase, staff for the phase, plan the next transition before it arrives.
