How Leaders Think
Leadership thinking is judgment — a call made with incomplete information, on a clock, on behalf of people who will live with the result — and intellectual honesty is the character operation that makes it trustworthy.
From the Founder
Let me tell you about a decision I got wrong, because it was a thinking failure before it was anything else. We had two employees and not enough in the budget to keep both at full salary. I did not want to let anybody go. So I asked them to take half, they agreed, and I felt good about it — everybody stayed, nobody got hurt. Except we had taken funding under a program whose terms did not permit what I had just done, and we never alerted them. We paid back over twenty thousand dollars in a couple of months. I paid it out of my own pocket. Here is what I actually got wrong. I reasoned backward from the outcome I wanted, and I never asked the one question that could have cost me that outcome: am I allowed to do this? Trying to keep everybody is not kindness. It is a decision, and somebody pays for it.
Executive Summary
Leadership does not require rare intelligence; it requires a kind of thinking most intelligent people never develop. Puzzles have answers. Judgment calls have only trade-offs, arrive with information missing, run on a deadline, and land on people who did not make them. This lesson names that work as the Judgment Load — incomplete information, irreversible consequence, entrusted stakes, and the clock — then asks the harder question: when can a leader trust his own gut? The honest answer comes from Kahneman and Klein's joint paper, which locates skilled intuition in regular environments with real feedback and finds it unreliable elsewhere. You will learn to tell your domains apart, and to notice that deciding not to decide is a decision with a price.
Learning Objectives
- Distinguish a puzzle from a judgment call, and explain why leadership consists mostly of the second
- Apply the Judgment Load — incomplete information, irreversible consequence, entrusted stakes, and the clock — to a decision you currently face
- State the two conditions under which expert intuition is trustworthy, and identify which of your own domains meet them
- Explain why intellectual honesty is a character operation and not merely a cognitive skill
Teaching Manuscript
The Character Operation You Perform With Your Mind
Module 2 ended with you alone in a room. The capstone came down to one question — who are you when exposure is impossible, when there is no audience to perform for and no consequence attached to cutting the corner? I told you the answer is who you actually are. Now I want to show you the door in that room that almost nobody checks.
There is a second thing you do in private that no one can audit, and it is not your conduct. It is your reasoning. Nobody watches you decide whether an argument actually worked. Nobody sees the moment the evidence started pointing away from a position you had already announced, and you kept announcing it anyway. Nobody is present when you round an estimate toward the number you were hoping for, or when you assign a colleague's objection to his motives so you do not have to engage its content. Every one of those happens in an empty room. Every one of them is a character operation performed with the mind.
This is why Module 3 comes third instead of first. If clear thinking were a matter of intelligence, we could have started here and saved a lot of time. But the leaders you have watched fail were not stupid. Look at the wreckage of the last twenty years — the collapsed banks, the collapsed churches, the collapsed campaigns — and count how many were run by people who could not think. Almost none. They were run by people who could think extremely well in the direction they had already chosen.
So let me name what this module actually trains. Not intelligence, and not information. Judgment — which is a different animal from the thing schools measured in you.
Here is the distinction that runs through everything after this paragraph. A puzzle has an answer. It may be hard, requiring real skill, but it exists, it is discoverable, and when you find it the puzzle is over. Calculating a payroll tax is a puzzle. So is diagnosing why the server crashed. Most of what earned you your credentials was puzzles, which is why credentials predict leadership performance so poorly.
A judgment call is not a puzzle. It has no answer waiting to be found, only options with different distributions of pain. The information you would need to be sure is unavailable, and some of it permanently. The people who will bear the consequence are mostly not in the room. And there is a clock running, so the option of waiting until you are certain is not neutral — it is a choice with its own costs, quietly accumulating while you tell yourself you are being careful.
Should we lay off eleven people now or risk laying off thirty in June? Do we tell the congregation what we know, when half of what we know will change? Do we close the intersection while the study is pending, knowing that closing it moves the traffic onto somebody else's block? Nobody solves those. Somebody decides them. That is the work.
The Judgment Load
I want to name it precisely, because named things can be trained and unnamed things just feel like pressure. Call it the Judgment Load: the four weights that turn ordinary thinking into leadership thinking. Every one is present on the decisions that matter, and each distorts reasoning in its own direction.
The first weight is incomplete information. You will never have the study, the audit, the full picture. What you have is a partial signal, and — this is the part leaders miss — the missing information is not randomly distributed. It is systematically what is hardest, most expensive, or most embarrassing to obtain, which is exactly what is most likely to be load-bearing. The absence of bad news is not the presence of good news. A leader who treats a quiet dashboard as evidence of a quiet operation has confused his instruments with reality.
The second weight is irreversible consequence. Some decisions are doors you can walk back through. Others are not. You cannot un-fire a person, un-say a thing to a reporter, or un-ring the bell on a public accusation. Reversibility is the most useful variable a leader can learn to read, because it tells you how much deliberation a decision deserves. Reversible decisions should be made fast; the cost of slowness exceeds the cost of a wrong turn you can correct. Irreversible ones deserve every hour the clock allows. Most leaders get this exactly backward — they agonize over the reversible and improvise on the permanent, because the reversible ones are usually the ones people are watching.
The third weight is entrusted stakes. This is the one that makes leadership judgment a moral category and not just a cognitive one. When you decide for yourself, you are wagering your own money, your own time, your own risk tolerance. When you decide as a leader, you are wagering theirs. The teacher whose school you consolidate did not vote on it. The family in the shelter did not get a briefing. Your appetite for risk is not their appetite for risk, and you are spending it anyway. That is why the tone of a leader's private reasoning should be closer to a fiduciary's than a gambler's. You are not being brave with your money.
The fourth weight is the clock. Every consequential decision has a deadline, and the deadline is usually not announced. It simply arrives, and the option set shrinks as it approaches. Which brings me to the sentence I want you to carry out of this lesson: deciding not to decide is a decision. It is not a neutral holding pattern. It is an active choice to accept whatever the default produces, and it is the most common way leaders fail without ever appearing to do anything wrong. Nobody writes a case study about the executive who studied the problem for nine months. The default just quietly happened, and everyone agreed it was unfortunate.
Now the executive corollary, and it is worth more than the framework. You can delegate the analysis — the modeling, the research, the options memo, the brief that steelmans the position you dislike. You should. What you cannot delegate is the load. Accountability stays where the authority is, and every attempt to distribute it is theater. Committees do not bear weight. The report that recommended the decision will not be in the room when the decision is bad. You will.
So when you find yourself assembling one more working group on a question you can already answer, ask honestly which of the four weights you are trying to set down. Usually it is the third. Usually the leader is arranging things so that if it goes badly, it will not have been his.
Solomon Asked for the Right Thing
Scripture puts the whole matter in one scene, and it is worth sitting in: a young king is offered anything he wants and picks the thing this lesson is about.
At Gibeon, God tells Solomon to ask for whatever he wishes. He does not ask for wealth, long life, or the death of his enemies — the text is explicit that he asked for none of those. He asks this: "So give Your servant an understanding heart to judge Your people to discern between good and evil. For who is able to judge this great people of Yours?" (1 Kings 3:9, NASB). The Hebrew phrase behind "understanding heart" is literally a hearing heart — a heart that listens. And God's answer names the thing directly: "I have given you a wise and discerning heart" (1 Kings 3:12, NASB).
Notice what he asked for and what he did not. Not information, and not certainty. He asked for the capacity to judge a people — to discern, on their behalf, with the weights we just named pressing down on him. And he stated his reason, which is the most honest sentence in the passage: who is able to judge this great people of Yours? That is a man who has correctly estimated the size of the job. The leaders who scare me are the ones who have not.
Then the narrative does something clever. It immediately hands him a case with no evidence. Two women, one living child, no witnesses, no documents, and two accounts that cannot both be true. This is a judgment call in its purest laboratory form: incomplete information, irreversible consequence, entrusted stakes, and a room full of people waiting. And Solomon does not deduce the answer. He designs a test that makes the parties produce the evidence — call for a sword, propose dividing the child, and watch which woman would rather lose the case than lose the child. He did not find the truth. He built a situation in which the truth had to reveal itself.
Hold that, because it is a transferable method and not a Bible story. When the data will not tell you, design a small test that makes reality answer. Ship to one district before all five. Ask the two vendors for a paid pilot rather than a proposal. Judgment is not only choosing among the options you were handed; often it is inventing a cheaper way for the world to show you which one is real.
The wisdom literature keeps the same posture. "The naive believes everything, but the sensible man considers his steps" (Proverbs 14:15, NASB). The naive man's failure is accepting input without testing it — the briefing, the number, the confident colleague, the thing everyone knows. The sensible man's discipline is deliberation before movement. Note that Scripture does not say he believes nothing. Corrosive suspicion is not wisdom either; it is a different way of not thinking. He considers.
And then the honest footnote, because this academy does not clean up its examples. The same book that records the wisest king records his end. 1 Kings 11 tells us his heart was turned away in his old age, and the kingdom fractured under his son. Wisdom given is not wisdom kept. Module 12 takes that up in full. For now, take the smaller lesson: the man who asked the best question a leader ever asked still finished poorly, so treat your own judgment as something under maintenance rather than something in the bank.
Without scrolling back: name the four weights of the Judgment Load, and state the one thing a leader can delegate and the one thing he cannot.
When the Gut Is Reliable and When It Lies
Philosophy named this competency long before management literature discovered it. In Book VI of the Nicomachean Ethics, Aristotle separates practical wisdom — phronesis — from theoretical knowledge. Theoretical knowledge concerns things that cannot be otherwise: mathematics, the eternal, the necessarily true. Phronesis concerns the opposite — things that can go either way, where you must deliberate about what is genuinely good for human beings and then act. He offers Pericles as his example, on the grounds that such men see what is good for themselves and for people in general.
Two details in Aristotle's account matter enormously for you. The first is that he says a young man can become a mathematician but not practically wise, because practical wisdom requires experience, and experience takes time. Judgment is not downloadable. The second is that Aristotle refuses to separate phronesis from character. On his account you cannot be practically wise while being a bad man, because a corrupted appetite corrupts the perception of what is worth doing. That is the same claim this academy makes in different vocabulary: intellectual honesty is a character operation. Aristotle got there twenty-three centuries before us.
Now to the research, and I want to be careful, because leadership books routinely overclaim here in both directions.
One tradition, associated most publicly with Daniel Kahneman, catalogs the systematic ways human judgment departs from what the situation warrants — anchoring, availability, overconfidence, substituting an easy question for a hard one. Read it alone and you conclude intuition is a menace and every decision should be run through a model. A second tradition, naturalistic decision making, comes from watching experts work in the field rather than undergraduates in a lab. Gary Klein studied fireground commanders expecting them to compare options. They mostly did not. He called what they actually did recognition-primed decision making: the commander recognizes the situation as an instance of a type he has seen, the first workable course of action comes to mind, he mentally simulates it, and if it survives he runs it. Read that alone and you conclude expert intuition is a marvel and analysis mostly slows it down.
The honest treatment is the one the two men wrote together. Kahneman and Klein published a joint paper in American Psychologist in 2009 titled 'Conditions for Intuitive Expertise: A Failure to Disagree.' They set out to find where they disagreed and largely could not. Their conclusion is the most useful paragraph in this lesson, so I will state it plainly: skilled intuition develops when two conditions hold. First, the environment must be sufficiently regular to contain learnable patterns. Second, the person must have had prolonged practice in that environment with feedback clear enough and fast enough to teach those patterns. Where both hold, expert intuition is real and should be trusted. Where either fails, the felt confidence is still produced — the mind manufactures it regardless — but it no longer tracks accuracy.
Apply that to your own domains and it becomes uncomfortably clarifying. A firefighter's environment is regular and the feedback is immediate. Your read on whether a room is with you is probably trustworthy, because you have done it a thousand times and found out within minutes. Your read on whether this executive hire will work out in five years is almost certainly not, because the feedback arrives years late, confounded with everything else that happened, and you have had perhaps a dozen genuine trials in your life. Long-horizon political forecasting is the hardest case; Philip Tetlock's work on expert political judgment found accuracy far more modest than the experts' confidence suggested. And a durable line of research going back to Paul Meehl's 1954 monograph on clinical versus statistical prediction has repeatedly found simple actuarial rules matching or beating expert judgment where feedback is noisy.
Two caveats, so I am not doing to you what I am warning against. Much of the biases literature was built in laboratories, and the size and stability of some individual effects have been debated in the replication era; treat the general finding as solid and any single dramatic study as provisional. Klein's work is field research on real experts, which buys ecological validity at the cost of controlled comparison. Neither camp has a clean win. What they agreed on is what you carry: check the environment before you trust the instinct.
The Desk Where It Lands
Let me put all of this on an executive's desk, because abstraction is comfortable and desks are not.
Consider what a mayor's morning looks like. Information arrives incomplete and pre-shaped by whoever delivers it. Every commissioner has an interest, every number has a methodology somebody chose, and the facts that matter most — what the union will actually accept, whether the developer is bluffing, whether the crisis is over — are unknowable at the moment they must be acted on. The consequences land on eight million people who did not vote on this particular thing. Several of the systems he will be blamed for are not his: the subway is run by a state authority, school governance depends on Albany, and the district attorneys and courts are independent by design. And a reporter needs an answer by four o'clock. That is the Judgment Load with the volume turned up.
Michael Bloomberg's administration is the most instructive modern attempt to solve this by measurement, in both directions. He brought 311 to the city in 2003, giving residents one number and the city a continuous stream of ground-level data. He pushed performance measurement out of policing and into agency management generally, and ran City Hall from an open bullpen so information would not have to climb a hierarchy to reach him. PlaNYC in 2007 set targets that outlasted the news cycle. The public-health measures — indoor smoking rules, the restriction of artificial trans fats in restaurants, calorie posting — were built on evidence and imposed real costs on organized interests. That is competence, and this academy is not embarrassed to say so. Running a city badly is a moral failure, not merely a technical one.
And now the limits, which are the reason the case is worth teaching. Measurement tells you about what you pointed the instrument at. The December 2010 blizzard buried outer-borough streets for days in an administration that prided itself on operational data. Stop-and-frisk expanded to roughly 685,000 recorded stops in 2011 — an activity metric going up, and to hundreds of thousands of New Yorkers something else entirely; a federal court held in 2013 that the practice as conducted violated the Constitution. The 2012 cap on large sugary drinks was struck down as exceeding the Board of Health's authority — a lesson about legal bedrock, not nutrition science. And the 2008 extension of term limits to permit a third term was not a data question at all. It was a character question wearing a governance costume. Competence at measuring is not wisdom about what to measure, and neither settles what you are entitled to do.
So here is the practice. Chester Barnard, writing in 1938 after nearly three decades inside the Bell telephone system and eleven years as president of New Jersey Bell, put it better than anyone since: the fine art of executive decision consists in not deciding questions that are not now pertinent, in not deciding prematurely, in not making decisions that cannot be made effective, and in not making decisions that others should make. Notice that three of the four are restraint. Judgment includes knowing which calls are not yours and which are not yet.
But restraint is not drift, and here is where I want you honest with yourself. Take the decision you have been carrying for three weeks and ask which of the four weights you are stuck under. If it is incomplete information, name the specific fact that would change your answer and price what it would cost to obtain — if that exceeds the value of the decision, you already have enough. If it is irreversibility, ask whether you can convert it into a smaller reversible version, the way Solomon converted an unanswerable dispute into a test. If it is entrusted stakes, say out loud who is paying for your delay right now. And if it is the clock, put a date on the calendar and decide on that date with whatever you have, because that is the job.
One more thing before Lesson 3.2. Everything here assumed your premises were sound and only your reasoning was at risk. That is generous. Most leadership errors I have watched were not errors of logic — the reasoning was fine and the starting point was inherited from a predecessor, a competitor, or a crowd, and never once examined. So next we go underneath the reasoning to the premises themselves, and ask what happens when you refuse to accept any of them without checking. That is first principles, and the Bereans were doing it before it had a name.
In one sentence: under what two conditions did Kahneman and Klein jointly agree that expert intuition can be trusted?
Through the Six Lenses
Evidence levels labeled per the Truth & Intellectual Integrity standard.
Biblical
Solomon asks not for wealth or victory but for "an understanding heart to judge Your people" (1 Kings 3:9, NASB) — the Hebrew is literally a hearing heart. God grants "a wise and discerning heart" (3:12). The case that follows has no witnesses; he designs a test that forces the truth out. Proverbs 14:15 sets the posture: the naive believes everything, the sensible considers his steps. 1 Kings 11 is the footnote — wisdom given is not wisdom kept.
Philosophical
Nicomachean Ethics VI distinguishes phronesis, practical wisdom about things that can be otherwise, from theoretical knowledge of what cannot. Aristotle names Pericles as an example, and insists a young man can become a mathematician but not practically wise, because phronesis requires experience. He also refuses to separate it from virtue: corrupted appetite corrupts perception of what is worth doing. Judgment is therefore a moral capacity, not a computational one.
Scientific
Two traditions long disagreed: heuristics-and-biases cataloged systematic error; naturalistic decision making, notably Gary Klein's recognition-primed model drawn from fireground commanders, found expert intuition working well. Kahneman and Klein's joint paper in American Psychologist (2009) settled the boundary — skilled intuition requires a sufficiently regular environment plus prolonged practice with adequate feedback. Absent either, confidence persists while accuracy does not. Meehl's actuarial literature and Tetlock's forecasting work show where that bites hardest.
Historical
Bloomberg's administration is the fullest modern test of managing a city by measurement: 311 launched in 2003, performance data pushed across agencies, an open bullpen at City Hall, PlaNYC in 2007, and evidence-driven public-health rules. Its limits are equally instructive — the December 2010 blizzard response, stop-and-frisk expanding to roughly 685,000 recorded stops in 2011 before a 2013 federal ruling against the practice as conducted, and a portion-size rule struck down as beyond the Board of Health's authority.
Influence
Cialdini's authority principle explains why a confident expert moves rooms — and confidence is persuasive largely independent of accuracy, which is precisely the danger. A leader who shows the reasoning rather than only the verdict trades a little immediate certainty for durable credibility, because followers can audit the path and follow it again without him. Borrowed certainty transfers well and survives contact with reality poorly.
Executive
Chester Barnard's Functions of the Executive (1938) frames the fine art of executive decision as largely restraint: not deciding what is not pertinent, not deciding prematurely, not making decisions that cannot be made effective, and not making decisions others should make. Pair that with reversibility as the pacing variable — decide reversible questions fast, irreversible ones as late as the clock honestly allows — and remember that non-decision is a decision that ships the default.
Case Study
New York City, December 2010: The Blizzard a Data-Driven City Did Not See
SITUATION. On December 26, 2010, a blizzard dropped roughly twenty inches of snow on New York City. The Bloomberg administration had built its identity on measurement — 311, agency performance data, and a City Hall bullpen designed so information would not have to climb a hierarchy. CONSTRAINTS. The storm landed the day after Christmas into thin holiday staffing. Forecast confidence had shifted late. The mayor is responsible for Sanitation and emergency medical response but does not run the subway, which is a state authority — a distinction the public does not make. DECISION. The city proceeded without declaring a snow emergency, and City Hall's early public posture was reassuring. ANALYSIS. Outer-borough streets were impassable for days. Buses and ambulances were stranded and emergency medical calls backed up badly. A subway train sat stranded overnight in Queens on a system the mayor answers for politically and does not control. Bloomberg later apologized; hearings and an investigation followed. The failure was not a shortage of data. It was a judgment made on a clock by a culture confident its instruments would raise an alarm if one were warranted. Instruments report what they are pointed at. DISCUSSION. What in your own operation would fail badly without ever appearing in a number you currently review?
Reflection Questions
- Name a decision you are currently studying rather than making. Which of the four weights are you actually stuck under, and who is paying for the delay?
- In which of your domains does feedback arrive fast and clear, and in which does it arrive years late and confounded? Rate your trust in your own gut separately for each.
- Recall the last time evidence started pointing away from a position you had already stated publicly. What did you do in the next ten minutes?
- If your judgment were graded only on the decisions nobody ever found out about, what would the grade be?
Practical Exercise — The Judgment Load Worksheet
Take one live decision. On a single page, answer in order: (1) Is this a puzzle or a judgment call? If a puzzle, stop and go solve it. (2) What is the specific missing fact that would change my answer, and what would it cost in time and money to obtain? (3) Is this reversible, and can I convert it into a smaller reversible test the way Solomon converted an unanswerable dispute into one? (4) Who bears the cost of the outcome, and who bears the cost of my delay — name people, not categories. (5) Does this domain meet both conditions for trustworthy intuition, or am I about to trust a feeling my environment never trained? (6) The date I will decide by, regardless. Put that date in your calendar now, then bring the page to Lesson 3.2.
Assessment
This Week’s Commitment
Name one decision currently sitting on your desk. Write down, before you decide: what would have to be true for me to be wrong, who bears the cost if I am, whether this is reversible, and what date I will decide by whether or not the information improves. Put the date on your calendar this week.
Identity statement to carry this week: “I am a steward of other people's outcomes, so my thinking is not mine to indulge. I will follow the evidence to conclusions I did not want, on the clock, and own the call I made.”
Discussion Questions
- Steelman the pure analyst's position: that expert intuition should never be trusted over an explicit model in organizational decisions. Where is that position strongest, and what does it fail to account for?
- Bloomberg's administration was demonstrably competent and still produced outcomes many New Yorkers experienced as unjust. What does that tell you about the relationship between competence and wisdom in public office?
- Where in your organization does the culture reward decisiveness in a domain that has never given anyone real feedback? What would it cost to say so out loud?
Reading List
- 1 Kings 3:5-28; Proverbs 14:15 (NASB) — a hearing heart, and the man who considers his steps
- Aristotle, Nicomachean Ethics, Book VI — phronesis distinguished from theoretical knowledge
- Gary Klein, Sources of Power: How People Make Decisions (1998) — recognition-primed decision making
- Daniel Kahneman & Gary Klein, 'Conditions for Intuitive Expertise: A Failure to Disagree,' American Psychologist (2009)
- Daniel Kahneman, Thinking, Fast and Slow (2011)
- Philip Tetlock, Expert Political Judgment (2005) — how good is long-horizon expert forecasting, really
- Chester Barnard, The Functions of the Executive (1938) — the chapter on the executive decision