Some ideas become clearer when written down.

There are problems that require your presence and problems that have simply become accustomed to it. Time away has a way of revealing the difference.

The first category survives distance poorly. Decisions stall in ways that are visible, specific and correctable, and the stalling is itself useful information. The second category often improves, because the people closest to the work stop waiting for confirmation and proceed on their own authority.

I have come to think of proximity as a resource with a cost. Every hour a leader spends inside a problem is also an hour the organization spends watching how that leader behaves inside it. At close range that observation is instructive. Held too long, it becomes the point of the exercise.

What changes at distance is not the information. The reports are the same, the numbers are the same, the people are the same. What changes is weighting. Issues that felt urgent because they were adjacent return to their actual size. Issues that were quietly structural become the only things still visible from that far away.

Leadership becomes more useful when an organization can distinguish access from dependence.

I do not think absence is a virtue. I think it is a diagnostic. It tells you which of your problems were ever yours.

A leadership team that only hears agreement has stopped receiving information. The first obligation is therefore straightforward: create conditions in which disagreement is cheap to express and expensive to ignore.

The second obligation is less comfortable. Having heard the disagreement fully, you may still decide against it.

I have watched capable executives collapse these two things into one. They listen carefully, they are genuinely moved by what they hear, and then they treat the strength of the sentiment as though it were the decision itself. In the moment this feels like respect. Over a longer period it produces something worse — an organization that learns to argue by volume, and a leadership team that has quietly relocated its judgment to whoever is most persistent.

The alternative is not to listen less. It is to be precise about what listening is for. Consultation gathers information the decision-maker does not otherwise have: consequences they cannot see, second-order effects, the operational reality behind a number. It does not transfer the decision, and pretending otherwise is its own kind of disrespect, because it sets an expectation that will be broken later and more painfully.

What people are owed is not the outcome they asked for. What they are owed is evidence that their reasoning was actually received — restated accurately, weighed against the constraints they could not see, and answered rather than absorbed.

People deserve to know they were heard even when the decision remains difficult.

The failure mode I watch for in myself is the meeting that produces agreement I did not earn.

Real urgency has a shape. There is a date attached to it, a consequence that can be described in one sentence, and a named party who is materially worse off if the date is missed. Anything lacking those three properties is usually escalation, and escalation is contagious.

Manufactured urgency works, briefly. It moves things. The cost is that it spends a resource most leaders never account for: the organization's willingness to believe you the next time. A team surged four times for reasons that later proved discretionary will surge more slowly on the fifth occasion, and the fifth occasion is invariably the one that mattered.

Executive behavior sets tempo whether or not it intends to. A question asked casually on a Friday evening is not received as a casual question. It is received as a signal, and signals are interpreted upward — an interest becomes a request, a request becomes a priority, a priority becomes a weekend. Nobody in that chain behaves unreasonably. The chain simply resolves ambiguity in the direction that is safest for the person receiving it.

The discipline, then, is not to be calm. It is to be explicit. State what is required, state when it is required, and state plainly what is not required. An organization that cannot tell the difference between your curiosity and your instruction will always resolve the difference in the more expensive direction.

I have not always been careful about this.

The organization needed a shorter decision cycle. It now has one.

There is a version of simplicity that is a lie. It removes detail without removing the underlying difficulty, and the difficulty reappears later, in a worse form, usually in someone else's quarter.

The version worth pursuing is different. It accepts that the complexity is real, places it with the people closest to it, and asks that what arrives at the top be a decision rather than a description.

This is often mistaken for a preference for brevity. It is not. A one-page summary is not simpler than a forty-page appendix; it is more expensive. The page costs more because someone has had to resolve the contradictions the appendix was permitted to leave open. Shortening a document is not compression. It is deferral in a smaller font.

The distinction is testable. Ask what was decided in order to produce the page. If the answer is that nothing was decided and the material was condensed, the complexity is still in the system, unowned, waiting.

What senior leadership should experience is compression, not omission.

The obligation this creates runs downward rather than upward. If the executive layer is going to consume compressed information, it has to accept the judgments made during compression, defend them publicly when they turn out to be wrong, and resist the temptation to reopen the appendix the moment an answer becomes inconvenient. A leadership team that demands the page and then relitigates the working is not simplifying anything. It is charging the organization twice.

Compensation matters precisely because it is concrete. It is the one part of the relationship that can be verified, compared and carried to another employer. Any leader who treats it as a secondary consideration is not being high-minded. They are being imprecise.

It is also incomplete — not because money is unimportant, but because it is the only term in the equation that a competitor can match exactly. If compensation is the sole reason someone stays, the relationship has a fixed price, and eventually somebody pays it.

The strongest organizations understand that people calculate value in more than one unit. Belonging: whether the work is done alongside people whose judgment they respect. Growth: whether the next two years will leave them more capable than the last two. Purpose: whether the outcome would be missed if it did not happen. Trust: whether what they were told last year turned out to be true.

None of these substitute for pay. They compound with it. A well-paid person in a low-trust organization is a well-paid person who is interviewing. A trusted person who is underpaid is doing you a favor that will expire, usually without notice.

The non-financial terms are harder to counterfeit than compensation, and they are the ones under a leadership team's direct control. Pay bands are set by a market. Whether people believe what you tell them is set by you.

So the right question is not whether compensation matters. It is what remains true about the work once compensation has been matched — and whether the answer is something anyone would say out loud.

The decisions I can trace most clearly did not happen in the meetings convened to make them. They happened in the ten minutes afterward, when two or three people stayed behind, or in a corridor, or standing near a window while someone worked out what they actually thought.

Scheduled time is good at confirmation and poor at formation. By the time an item has a slot, a title and an owner, its shape is largely fixed; what remains is agreement. Formation happens earlier, at lower stakes, and it is almost always incidental.

This is an argument for proximity, but it is not an argument for attendance. Being inside a building is not the same as being available in it. I have watched leaders occupy an office five days a week and remain perfectly unreachable — back-to-back, door closed, present in the way furniture is present.

What matters is unstructured adjacency: enough shared physical context that a question can be asked without first being justified. That is difficult to manufacture and easy to destroy. It does not survive a calendar with no gaps in it, and it does not survive a leadership team that treats informal contact as a lower-value use of an hour.

The interaction that changed the outcome was the one no one scheduled.

The organization has not moved yet. Evidence is incomplete. Consensus still belongs to the previous model.

This is the uncomfortable period.

Both conditions — being early and being wrong — produce the same external signals. Capable people disagree with you. The data can be read either way. There is a growing sense that continuing to argue is costing more than the argument is worth. Nothing available to you at the time distinguishes conviction from stubbornness. That distinction is made later, by outcomes, and then applied backwards as though it had been obvious throughout.

What can be examined is the reasoning. A position that is honestly held will name, in advance, the evidence that would overturn it. A position that is merely stubborn will keep moving that threshold as the evidence arrives. I have found this to be the only test available from the inside, and I do not think it is a strong one.

Leadership often means staying with an idea long enough to discover which kind of discomfort you are experiencing.

It also means accepting that the organization pays for the discovery period either way, and that the cost is not distributed evenly. The person who is early carries the intellectual risk. The people implementing carry the operational one. Those are not the same risk, and they are rarely borne by the same person.

A program that ends early has produced something real: it has removed a hypothesis from the set of things the organization still has to wonder about. That has value, and it is almost never recorded as a return.

The difficulty is that this is also the most convenient thing an executive can say about a decision that did not work. So it needs a test, and the test is timing. If the case for stopping was made when the evidence appeared, it is judgment. If it was made after the outcome had already become public, it is narration.

Sunk cost is usually described as a failure of arithmetic. Inside an organization it is rarely arithmetic. It is that stopping requires someone to stand in front of the people who did the work and tell them the work will not continue — and the more capable those people were, the harder that conversation becomes. The instinct to protect them is decent. Extending a program in order to postpone the conversation is not protection. It is deferral, with interest, paid by them.

The signal I have learned to watch for is a change in a program's justification. A thesis that was originally about decision velocity, and is later about optionality, and later still about the capability it built, has stopped being a thesis. It is being defended rather than tested.

Good judgment is not being right forever. It is recognizing when reality has earned the right to change your mind.

I began thinking seriously about time during a month when I was moving between Zürich, New York and Singapore and found myself increasingly protective of the hours without scheduled conversation.

It was not fatigue. It was that the quality of the decisions made in the remaining hours was visibly different from the quality of the decisions made in the others. A calendar with no gaps does not produce a leader who is more available. It produces a leader who is reactive, and whose largest judgments are made in the ninety seconds between two meetings that were each considered more important than the judgment.

Protection is the wrong word if it implies withdrawal. What is being protected is not the person. It is the decision. An hour held open ahead of a consequential choice is not rest; it is part of the work, and it should be defended in the same terms as any other input, with the same willingness to say out loud what it is for.

The uncomfortable part is that this is a privilege of position. The ability to decline a meeting is distributed unequally, and it is distributed in inverse proportion to how many meetings a person is asked to attend. A leadership team that protects its own hours while removing everyone else's has not adopted a principle. It has adopted a perk, and the organization can tell the difference long before anyone says so.

Good leaders protect the boundaries of their teams.

Delegation is usually described as an act of generosity. It is more accurate to describe it as a test — of the structure, and of whether the leader believes their own account of it.

The test is simple. Remove yourself, and observe what the organization does with the decisions that were previously routed through you. If they stop, the structure was never real; authority had been described but not transferred. If they proceed, and proceed differently from how you would have decided, the more interesting question begins: whether you can leave the difference alone.

I have found that harder than the delegation itself. There is a particular temptation to enter a resolved situation and improve it — to add, after the fact and at no cost to yourself, the refinement you would have made. It feels like contribution. What it teaches the organization is that resolution is provisional until reviewed, which is precisely the dependency you were trying to remove.

Sometimes trust is simply the decision not to enter the room.

Ambiguity is tiring, but it is not hard. Where an answer is genuinely unknown, more work usually helps: another week of analysis, another perspective, a smaller reversible step. The difficulty has a bottom to it.

The decisions that are actually hard are the ones where further analysis has stopped changing anything. The direction is established. What remains is a set of consequences that will fall on people who did nothing wrong, and a leadership team that would prefer, understandably, to wait.

Waiting has a cost that is easy to overlook because it is paid quietly. An organization carrying a decision that everyone senior knows is coming does not operate normally. Hiring continues into areas that will not survive. People commit to work that will be stopped. Delay does not reduce the harm. It distributes it over a longer period and adds uncertainty to it.

I have come to believe the obligations in these moments are narrow and non-negotiable. Decide once, so the organization is not asked to absorb the decision twice. Say it directly, in the first sentence, without preamble that requires interpretation. Be the person who says it — not a communications function, not a document. Answer the questions that follow, including the ones with no satisfactory answer, and do not mistake the absence of anger in the room for the absence of anger.

And afterwards, remain available. The instinct is to give people space. What is often experienced instead is a leader who made a decision and then became difficult to find.

None of this makes a difficult decision less difficult. It makes it honest, which is a lower standard than most leadership writing implies and considerably harder to meet.

I am wary of both available positions. The claim that these systems change nothing is not serious. The claim that they remove the need for judgment is a category error, and an expensive one.

What they demonstrably change is the cost of exploration. Options that were previously too expensive to investigate — a second architecture, a fourth market model, the version of the strategy nobody had time to write — can now be examined before anything is committed. That is a real expansion of range, and range is where most strategic advantage has always come from.

They also compress execution, which is the part that attracts attention and the part I find least interesting. Producing an artifact faster does not improve the decision the artifact was meant to serve. Frequently it just means the organization now has more artifacts.

The thing that does not move is responsibility. A system can generate the options. It cannot hold the consequence. It has no position to lose, no relationships to damage and no obligation to the people affected by what it produced. Those remain exactly where they were.

So the question I ask is narrow: does this let us consider more, decide better, and remain accountable for the result?

Where the answer is yes, I have no reservations. Where the answer is that it lets us produce more, I would rather we produced less.