Building and Leading Teams
This is a typological theme organising the wiki’s leadership and management cluster around three distinct accounts of what makes leadership work — represented by Kim Scott, Claire Hughes Johnson, Bill Carr, Jerry Colonna, Jeffrey Pfeffer, Keith Rabois, Molly Graham, Carole Robin, Heidi Helfand, Matt Mochary, Howie Liu, Ben Horowitz, Jeetu Patel, and Farhan Thawar. It covers leading others: building, managing, aligning, and restructuring teams and organisations. For personal career and self-authorship, see Designing a Working Life. For the decision processes leaders use, see Deciding Under Uncertainty.
The organising axis
Three accounts compete to explain what makes a leader effective. They are not mutually exclusive — every practitioner in this cluster draws on all three — but they weight differently, and they disagree about where failure typically originates.
Management as operating system holds that dysfunction is structural: the leader who lacks clear mechanisms — written processes, explicit goal-setting, single-threaded ownership, hiring bars — will fail regardless of interpersonal skill or self-awareness. The solution is design. Build the system before you need it.
Management as relationship and candour holds that dysfunction is interpersonal: the leader who cannot give honest feedback, who withholds difficult truths, or who cannot form genuine relationships destroys teams even with excellent processes. The solution is directness grounded in care.
Management as inner work and power holds that dysfunction is psychological: the leader’s own unsorted history, emotional reactivity, and distorted relationship to power corrupt every system and every relationship built on top. The solution is self-examination — and, separately, a clear-eyed understanding of how organisational power actually works.
Management as operating system
Claire Hughes Johnson gives this account its fullest architecture. She frames company-building as three layers: founding documents (mission, long-term goals, operating principles), an operating system (OKRs, review cadences, decision frameworks), and an operating cadence (the rhythm of planning and events). The most expensive mistake is building the system too late. The sign you have delayed too long: new hires repeatedly ask why things are done a certain way, and no one can answer.
Bill Carr’s account of Amazon’s management practices extends the same logic. Single-threaded leadership — permanent teams with clear owners, dedicated resources, and no matrix reporting — is a structural solution to a structural problem: when ownership is diffuse, decisions slow, accountability disperses, and the organisation becomes a referee between competing interests rather than a builder. The PR/FAQ process forces the same discipline earlier: requiring a fictional press release before engineering begins separates ‘deciding what to build’ from ‘deciding how to build it’, and eliminates the failure mode of technology-first thinking in search of a problem (the Fire Phone). Amazon’s input-vs-output metric distinction is the same principle applied to measurement: manage the causal variables (selection breadth, page load, shipping speed), and output metrics follow as an article of faith.
Keith Rabois adds the talent-density claim. The PayPal Mafia was not an accident — it was the product of constrained hiring through first- and second-degree networks, which produced critical mass of exceptional people. Talent compounds: eight extraordinary people and two mediocre ones underperform ten extraordinary people even though they appear larger. The first-order implication is to mine underutilised internal talent before hiring externally. The second-order implication is that public criticism of work (not of the person) raises the shared standard — visibility is the mechanism, not surveillance.
Farhan Thawar’s Shopify practices operationalise intensity without extending hours. Meetingageddon — an annual deletion of all recurring meetings with three or more internal participants, followed by a two-week moratorium — dropped IC meeting load from five or six hours to roughly three. Tobi Lütke’s rule (if you cannot delete and restart a piece of work within one hour, it is not understood) drives the Delete Code Club: code as liability, not asset.
Molly Graham’s Waterline Model is the diagnostic tool for this account. Before blaming a leader’s self-awareness or interpersonal failures, check the upper layers: does the person have the authority and role clarity to succeed? Are the team rituals and communication patterns working? Only once the structural and dynamic layers are clear does it make sense to descend to the interpersonal or intrapersonal. Most coaching fails by going straight to the bottom.
Management as relationship and candour
Kim Scott‘s Radical Candor framework maps the failure modes of feedback on two axes: care personally and challenge directly. Each is necessary; neither is sufficient. The dominant failure is not aggression but Ruinous Empathy — withholding feedback because the conversation would be uncomfortable. Scott estimates this accounts for roughly 90% of management failures. A manager who lets a direct report walk into a performance review surprised at month twelve, after months of withheld feedback, has protected their own comfort at the direct report’s expense.
The overcorrection trap compounds this. Managers who recognise they have been obnoxiously aggressive typically swing not to Radical Candour but to Manipulative Insincerity — they become politically vague, withholding both honest challenge and genuine care. The corrective is not less challenge but more care.
Carole Robin’s three-realities model operationalises the same insight structurally. Any exchange has three distinct realities: your intent, your observable behaviour (the only reality both parties share), and the impact on the other person. ‘Stay on your side of the net’ means confining feedback to what you actually know — your own intent and the observed behaviour — and never attributing a motive to the other person’s inner state. ‘You don’t care’ is an over-the-net attribution; ‘when you do X, I feel Y’ is not. The 15% rule governs disclosure: relationships deepen through incremental vulnerability, not sudden openness. Step 15% outside your comfort zone; the other person reciprocates; a slightly larger comfort zone is established.
Alisa Cohn on Scripts for Difficult Conversations and Leadership adds the practitioner’s detail: scripts for performance feedback, promotion denial, and pre-termination conversations. The pre-termination conversation is the most commonly skipped. Managers believe they have been clear; they have not. A termination that surprises the employee is a leader’s failure, not the employee’s.
Ben Horowitz frames the interpersonal dimension in terms of hesitation. In leadership, both options are usually bad — the test is which is slightly less bad. Hesitation is worse than either: it signals indecision to senior staff, creates a political vacuum, and locks the organisation. The psychological capacity needed is not optimism but the ability to look into the abyss, pick the less-bad option, and run at it. Managerial leverage follows: when a CEO has to prompt a functional leader for ideas, the CEO has lost the leverage that justified the hire.
Jeetu Patel’s ‘debate in public, build trust in private’ principle addresses candour at organisational scale. Healthy organisations surface disagreement openly in the room. Consensus manufactured outside the room — decisions already made before the meeting — poisons candour and removes the signal that disagreement provides. At 90,000 employees, the challenge is ‘storytelling without packet loss’: stripping complexity until only what matters at each organisational layer survives transmission.
Management as inner work and power
Jerry Colonna‘s Leadership Equation states the inner-work case formally:
practical skills + radical self-inquiry + shared experiences = enhanced leadership + greater resilience
The circled term is resilience, not leadership. Colonna’s premise is that teams fail from unsorted baggage, not lack of talent. Unconscious patterns — imported from founders’ histories, from childhood, from the unexamined drive to prove something — corrupt team dynamics before any process or feedback framework can operate. Until a pattern becomes conscious, it directs the leader and they call the result bad luck or bad people. The leader with the most power is the highest-leverage point for change.
Radical Self-Inquiry is the practice. The signature question — ‘How have I been complicit in creating the conditions I say I don’t want?’ — is deliberate on two counts. ‘Complicit’ evokes agency without blame: you are driving the getaway car, not sticking up the bank teller. ‘I say I don’t want’ names the gap between stated wish and actual behaviour. The questions that startle and slightly frighten are ‘where the gold is.’
Jeffrey Pfeffer’s account of power sits in this cluster but diverges sharply from Colonna’s. Where Colonna treats the leader’s psychology as the source of dysfunction, Pfeffer treats it as the source of self-sabotage. His Seven Rules of Power are descriptive: get out of your own way (self-limiting beliefs kill trajectories before external obstacles do); appear powerful (perception precedes and often creates reality); network relentlessly (lead with generosity, pursue weak ties, aim to broker the unconnected); use power (power unused atrophies). The governing claim is that power is a learnable skill, not a personality trait — and most people stay powerless because acquiring it makes them uncomfortable.
The Knowing-Doing Gap is Pfeffer’s complement: the primary obstacle to building influence is not lack of knowledge but failure to act on it. Almost everyone understands what to do. Almost none do it. The mechanism is social discomfort — the actions that build power feel awkward, so people prefer being right over being effective.
The tension between Colonna and Pfeffer is real. Colonna argues that the leader’s relationship to power — the drive to accumulate it, the fear of losing it — is often the source of organisational dysfunction. Pfeffer argues that refusing to engage with power is naïve and leaves good people unable to accomplish what they want. Both claims are compatible: the leader who has done the inner work Colonna describes may be better placed to exercise the power Pfeffer advocates without the distortions that make power corrosive.
Matt Mochary occupies the overlap. Fear and anger give bad advice: when a leader acts from either state, the decisions that follow are almost always wrong. Naming the emotion before making a consequential decision is not soft psychology — it is a structural prerequisite for good judgement. His energy audit (four zones: incompetence, competence, excellence, genius) is both an inner-work tool and a leverage instrument: move as much time as possible toward zone-of-genius activities and delegate the rest.
Where the accounts agree
Three points cut across all three schools.
Talent density matters most. Keith Rabois is the most explicit, but Claire Hughes Johnson‘s Bar Raiser equivalent at Stripe and Bill Carr‘s Amazon Bar Raiser process make the same bet: the hiring bar is the single highest-leverage control in the system. A company of exceptional people in a functional system beats a company of adequate people in a great system.
Clarity is the mechanism. Scott requires feedback to be specific and observable. Carr requires ownership to be explicit and single-threaded. Graham requires goals to be few, shared, measurable, binary, ambitious, and time-bound. Colonna requires leaders to make the unconscious conscious. Pfeffer requires leaders to name what they actually want. Every account collapses dysfunction to ambiguity in some form.
Teams change, and change must be managed. Heidi Helfand’s Dynamic Reteaming is the structural account of this. Team composition changes are not a disruption to be minimised — they are a constant to be managed through five patterns: one by one, grow and split, merging, isolation, and switching. The failure modes are structural (the percentage anti-pattern, spreading high performers) and communicative (‘poof, they’re gone’). The IC CEO model from Howie Liu applies the same insight at the leadership level: as AI collapses the distance between deciding what to build and building it, the leader who maintains hands-on product participation — ‘tasting the soup’ rather than reviewing finished output — sustains the judgment that delegation alone cannot.
Where they diverge
On feedback culture: Rabois argues explicitly against psychological safety as an organising principle for high-performance teams. Public criticism of work — not the person — is the mechanism that lifts the shared standard. Scott and Robin argue, with equal force, that feedback received in a context of evident contempt or distrust cannot be processed: the physiological response to perceived threat prevents the cognitive work of hearing and integrating criticism. This is not a terminological disagreement. The two camps are betting on different models of how people change.
On power: Pfeffer’s descriptive account of how power is actually acquired sits uncomfortably alongside the conventional leadership literature’s preference for authenticity, humility, and servant leadership. Pfeffer’s reply — that the rules are descriptive, not normative, and that more good people having power is the point — does not fully resolve the tension, which he acknowledges: the Seven Rules of Power underweight the price of power (scrutiny, lost autonomy, family cost). Colonna’s Radical Self-Inquiry asks a question Pfeffer’s framework does not: what has acquiring power cost you, and is the cost visible to you?
On org design: Dynamic Reteaming‘s embrace of continuous team change sits against the Team Topologies orthodoxy (stable, cognitive-load-optimised teams) and against Rabois’s talent-density view, which implies that team chemistry, once achieved, should be protected rather than redistributed. Organisational Kayfabe — Alex Komoroske‘s structural account of why status reports are always green until a project is cancelled — suggests that the disagreement may be downstream of a deeper asymmetry: you cannot make your boss look dumb, and every org-design intervention must reckon with this thermodynamic before any cultural intervention can take hold.