DEEP READING · READ 914
A leadership experiment aboard a nuclear submarine · L. David Marquet · 2012
A submarine captain noticed that the sharper and more decisive his orders were, the more completely his crew outsourced their brains to him — so he did something close to unthinkable in the military: "I will never give another order." Instead he made everyone aboard say "I intend to…", and turned the worst-performing boat in the US fleet into its best-rated one — and into a machine that kept producing captains for a decade afterwards. The book isn't platitudes about empowerment; it is empowerment broken down into concrete moves you can copy tomorrow morning.
Marquet is a US Navy officer who in January 1999 took command of USS Santa Fe, a Los Angeles-class fast-attack nuclear submarine then sitting at the bottom of the fleet in both retention and readiness. The pivotal accident of his career: he had spent a full year preparing to command a different boat and was reassigned to Santa Fe at short notice, meaning he knew this particular submarine's technical detail far less well than his own crew did. That predicament — "I'm not the one who knows" — is what forced out the entire method. Published in 2012 with a foreword by Stephen Covey, it has since become a leadership manual quoted equally in Silicon Valley and in the armed services, for one reason: other books hand you a philosophy; this one hands you mechanisms.
Leader-follower is the default setting: one person thinks, everyone else does; orders go down, get repeated back, executed, reported on. It isn't stupid — in the age of rowing, hauling and firing muskets in ranks it was optimal, because the work was physical and standardisable, and one brain could easily direct a hundred pairs of hands. The trouble starts when the work becomes cognitive: the scarce resource shifts from hands to judgement, and this structure switches ninety-nine people's judgement off.
It carries two hidden costs, both of which Marquet watched happen on his boat. First, capability atrophies structurally: a person who never has to decide anything never learns to decide, so when he is promoted he too will simply wait for instructions. Second, performance is welded to the leader: the excellent results a brilliant captain produces evaporate the day he transfers out. Marquet treats that as the real test of leadership — not how well the ship runs while you're aboard, but how well it runs after you leave.
Leader-leader starts from the opposite assumption: people already have the capacity to lead; the structure is what switches it off. Your job is not to grant them leadership, it is to remove the things preventing them from using it. So all of Marquet's moves are subtractions — remove the approval step, remove the report-up, remove "wait until I say so." His one-line summary is the keeper: "Don't move information to authority; move authority to the information."
This is the book's most famous and most copyable move. The turning point was a drill: Marquet ordered "ahead two-thirds," the officer of the deck dutifully repeated it — and nothing happened, because this boat had no two-thirds setting at all. It emerged that the officer had known the order was wrong and passed it along anyway, on the grounds that the captain had said so. In that moment Marquet saw it: as long as he was giving orders, the smartest people aboard would automatically switch off their own judgement. He decided on the spot never to give another one.
But if you don't give orders, how does anything move? By converting requests into declarations. Before: "Captain, should we come to periscope depth?" — the decision belongs to the captain. After, mandatorily: "Captain, I intend to come to periscope depth, because…" — the decision belongs to the speaker, and the captain merely holds a veto. It looks like grammar. It relocates three things at once: the duty to think, the ownership of the decision, and who owns the post-mortem. Anyone who says "I intend to" has had to work the plan, the reasoning and the risks through first, because he knows what follows the sentence is his own action.
Marquet arranges this into a ladder of leadership: for the same problem, a person can open his mouth in about seven different ways, and every rung up hands him more of the initiative. Your job is not to answer, it is to ask "so what do you intend to do?" and push him up one rung.
| How he opens | Who is thinking | What you say back |
| "Tell me what to do." | Only you | Bite your tongue: "What do you think?" |
| "What should we do?" | Still you | "What options do you see?" |
| "I think we should…" | He's started | "Why? What's the risk?" |
| "I'd like to…" | He's done thinking, awaiting you | "Do you need my approval?" |
| "I intend to…" | He is — and he owns it | "Very well. Proceed." |
| "I've done it; here's what happened." | Fully his | Just be informed, and debrief |
The ladder of leadership (schematic): same person, same problem — move the phrasing up a rung and the initiative moves down a level.
The hard part isn't your people; it's you. You have to resist the urge to provide solutions — because you probably do know the answer, and you can produce it faster than they can. But every time you supply it, they slide back down a rung.
This is the skeleton, and the reason the book is more serious than the usual empowerment sloganeering. Control means pushing decision rights down. But Marquet keeps sounding the warning: the moment control moves down, the other two pillars must grow with it, or you have simply abandoned the wheel.
An easy way to hold the three together: control is the accelerator, competence is the brakes and tyres, clarity is the navigation. An organisation that only presses the accelerator isn't agile, it's out of control. Nearly every failed "empowerment" initiative dies right here — the boss devolves authority without training and without stated priorities, something breaks, the authority is yanked back, and the whole organisation learns one lesson: better to ask permission next time.
An incident aboard: a crew member, running on muscle memory, put a switch in a position it should not have been in. The traditional response is punishment, more checks, another supervisory signature. Marquet's response was to ask: how do you want to stop this happening again? The crew's own answer became the single most portable mechanism in the book — deliberate action: before operating any control, vocalise what you are about to do, hold your hand still for a beat, then do it.
The mechanism isn't "so others can supervise you" (though bystanders do get a chance to shout stop). It is to let the brain catch up with the hand: speaking and pausing drag an automatic, unconscious motion back into the conscious channel. Most low-level errors don't come from not knowing; they come from knowing, but the hand moved first. Which is why it transplants into any industry: the surgical pause, the pilot's spoken checklist, reading the change aloud before you hit enter on a deploy, saying the recipient's name out loud before you send the email. It adds no process. It adds half a second of awareness.
The military briefs everything: the person in charge runs through the plan, everyone else nods. Marquet spotted the flaw in the ritual — it is passive. Listeners can drift the entire way through, still not be able to say what they personally must do, and responsibility has quietly transferred to the person doing the talking.
He replaced it with certification: instead of me telling and you listening, I ask and you answer; the operation is authorised only when every person can demonstrate that they understand the task, the risks and their own part in it — and if they can't, the evolution is cancelled on the spot. Briefing establishes that information was sent. Certification establishes that understanding happened. This one drops straight into any team's design review or pre-deploy meeting: swap "let me walk you through it" for "walk me through it," and you will discover in minutes how much of "everyone knows this" was never true.
The first practical problem after devolving authority: someone grinds for three weeks, brings you the finished thing, and the direction is entirely wrong. Do you overrule it? Overrule, and "I intend to" was theatre. Don't, and the mistake costs real money. Marquet's answer is short, early conversations: don't wait for the deliverable; spend two or three minutes on direction while the person is still five minutes into the idea.
Why it works is worth spelling out. A two-minute conversation at the start costs essentially nothing to correct, and because nothing has been built yet, nobody feels their work was rejected — the initiative stays in their hands. Wait three weeks and you have only bad options: overrule them (they learn "the boss decides in the end anyway") or swallow it (the organisation learns "wrong direction goes unremarked"). Frequent, short, early and light is the only way empowerment and quality can both be true at once. People assume delegating means communicating less. It requires communicating more, in smaller pieces, much sooner.
Its partner rule is specify goals, not methods: state what must be achieved and what must not be touched, and leave the how entirely alone. Because the moment you also specify the method, responsibility silently moves back to you — when it fails, the answer is "I did what you told me," and once that sentence exists, ownership is already dead.
This is the substance of the clarity pillar and the most counter-intuitive item in the book. What grows in an organisation whose goal is "don't let anything go wrong"? Layers of sign-off, people who won't move, and a hunt for the culprit after each incident — because under a "make no mistakes" objective, the optimal strategy is always to do less. Marquet rewrote the ship's goal as "we're going to be the best," then bent every mechanism toward it.
One elegant reversal follows: embrace the inspectors. Fleet inspection teams are normally treated as the enemy — you manage them, you hide things. Marquet had the crew treat them as free outside expertise instead: show them the problems, chase them for "where else could we be better?" Change the goal from "don't get caught" to "get stronger," and inspection converts from a threat into a resource. A companion rule: use immediate recognition — don't save it for the annual review; say it when you see it, because the shorter the gap between behaviour and feedback, the faster the learning.
Submarine officers are trained to appear certain at all times; hesitation reads as weakness. Marquet required the opposite of everyone including himself: speak your assessments, guesses and worries aloud — "that current's bothering me a little," "my guess is they'll turn north."
Two effects. First, it externalises the invisible reasoning inside an expert's head, so juniors absorb not just the conclusion but the route to it — something no training material can supply. Second, it leaves a door open for correction: a spoken judgement carrying visible uncertainty invites someone to say "but…", whereas nobody talks back to a flat, confident order. Which is the "ahead two-thirds" lesson again, from the other side: the more certain the leader sounds, the quieter the organisation gets.
The reported results: Santa Fe went from bottom of the fleet to one of its highest-rated boats, and from the worst retention numbers to the best. But the metric Marquet himself cares about is a different one — that over the following decade the boat produced far more submarine commanders and department heads than its peers (his account). That is the real acceptance test for leader-leader: not your scorecard while in post, but whether the thing keeps growing once you are gone.
Its soft companion mechanism is begin with the end in mind — get every person, including the newest sailor, to articulate what they want this tour to have given them by the time they leave, then work backwards to what that means today. It sounds like a poster. But in an organisation where people are mostly counting the days until transfer, it is precisely the switch that reconnects "why am I here at all."
The book is one causal chain, not a bag of tips: I don't know this boat (so I can't give orders) → let the people who do know decide (devolve control) → but they must genuinely know, and must know what we're for (competence and clarity, in lockstep) → the structure turns from one brain plus a hundred pairs of hands into a hundred leaders → and the test is whether it still runs after I leave.
1. The two-hundred-year-old leader-follower model uses one brain to direct a hundred pairs of hands: superb for physical work, and the single biggest waste in cognitive work.
2. Don't move information to authority; move authority to the information — the people who know sit at the bottom, the right to decide sits at the top, and the whole book is a repair of that mismatch.
3. The turning point was a wrong order relayed by a man who knew it was wrong: while you are still giving orders, even your sharpest people will switch their judgement off.
4. Change "should we…?" to "I intend to…, because…", and one sentence relocates the duty to think, the ownership of the decision, and the ownership of the post-mortem.
5. The hardest part is you: resist the urge to provide solutions — every answer you supply knocks someone one rung down the ladder.
6. Control, competence and clarity must grow together: control is the accelerator, competence the brakes and tyres, clarity the navigation — accelerator alone isn't agility, it's a crash.
7. Say it aloud and pause before you act (deliberate action) — not so others can police you, but to let the brain catch up with the hand; most errors aren't ignorance, they're the hand moving first.
8. Stop briefing and start certifying: swap "let me walk you through it" for "walk me through it" and watch how much of "everyone knows" evaporates.
9. In an organisation aimed at "don't let anything go wrong," the optimal move is always to do less; aim it at "be the best" and even the inspectors turn into free outside expertise.
10. The real test of leadership is not how well the ship runs while you're aboard, but how well it runs after you leave — whatever evaporates on your transfer date was never leadership, only dependence.