Not Every Decision Is a Discussion

The error rate has been climbing for eleven minutes. The incident channel has forty people in it. The engineering manager, who has seen this exact failure twice before, types: “What does everyone think we should do?”

Somebody suggests a rollback. Somebody else worries about the data written since the deploy. The manager thanks them both. Nine minutes later the on-call engineer rolls back on their own, and the manager writes “great call” in the channel as though the discussion had produced a decision instead of avoiding one.

I have been in that channel, and I have been the one typing the question. Many of the questions leaders ask their teams were never questions. The leader already knew the answer, or was the only person in the room positioned to know it, and asked anyway so that the call would belong to the room instead of to them.

Asking is the right default and the wrong absolute. The leaders who get this wrong are usually the ones who worked hardest to stop being the boss who never asked.

The Pattern: Asking Became the Way to Not Decide

A generation of managers learned, correctly, that command is what bad bosses do, and promised themselves they would ask. They ask about the architecture, which is right. Then they ask whether the team wants to skip the failing security check to make the release date, and the asking has stopped being a virtue.

Aristotle wrote in the Politics that the good citizen has to know two things, how to rule and how to be ruled, and that the mark of a free person is the ability to do both. A leader who can only ask is stuck in the same way as one who can only tell. They have one mode, and the situation does not care which one it is. In facilitation, the leader opens the question and the room’s answer can change the outcome. In command, the leader decides, and the room’s job is to execute and to argue afterward.

Facilitation run as an absolute lets the leader skip the one thing the role exists to do, which is to own a call that could turn out wrong. When you ask a room to decide something you were supposed to decide, you are borrowing the room as insurance, and the people who wait for consensus and call it thinking hold the same policy.

Four Places the Question Was Never Real

1. The standard you opened for debate

A security control, a safety rule, or a line of conduct comes up in a meeting as a question. “Do we think we can ship without the auth check this once?”

The moment a standard is put to the room, the room learns that it is negotiable. Even when the vote goes the right way, the standard has been moved into the category of things a good enough argument can shift, and the next argument will be better.

The team’s voice belongs in setting the standard, loudly, and stops belonging the moment the question becomes whether the standard applies this afternoon.

What you were avoiding: being the person who said no with nobody to share it. A standard survives on someone’s willingness to be unpopular about it, and asking the room spreads the unpopularity so thin that nobody feels it.

2. The crisis you have seen before, run as a workshop

On January 15, 2009, when both engines on US Airways 1549 failed over the Hudson, Captain Chesley Sullenberger said two words to his first officer, Jeffrey Skiles: “My aircraft.” Skiles answered, “Your aircraft.” The transfer of control was spoken in both directions, in the first seconds. Crew resource management, the training that came out of crashes where junior officers saw the problem and stayed quiet, exists to open the cockpit to voice, and the same training insists that when control moves, it moves out loud.

Software borrowed the other half from firefighters. Incident command puts one person in charge for the duration of an incident. That person directs, and nobody holds a vote on whether to roll back. The argument about whether it was the right call happens in the retro, with the logs open, when being wrong costs nothing. A familiar failure has a short window in which one clear voice beats nine informed ones.

What you were avoiding, when you typed the question into the channel: being wrong alone.

3. The expert who pretends the table is level

You have run this migration four times. Nobody else in the room has run it once. You open with “What do you all think?”

The room can feel the answer key in your pocket. People guess at what you want, or stay quiet so as not to guess wrong, and the discussion turns into a slow reveal of a decision you had already made. The condition is the size of the gap. Architecture is a real discussion when the room knows it as well as you do, and a quiz when it does not.

What works instead: say what you think, show the reasoning, then ask for the holes. “Here is what I would do and why. Tell me where it breaks.” You have the rank, and the fake question is how you avoid looking like you used it, which teaches the room that your questions are not questions.

4. The new hire you handed autonomy instead of direction

Someone joins, and on day four you tell them you trust them and step back, because you do not want to micromanage. They flounder for a month, and you call it a learning curve.

autonomy given to someone who cannot yet use it is abandonment. Withholding direction out of respect for their independence protects your self-image at their expense. Hersey and Blanchard’s situational leadership model starts a person new to a task with plain direction, and autonomy arrives as they show they can carry it.

Xenophon marched ten thousand Greek mercenaries out of Persia in 401 BC after their generals were murdered. The army elected its commanders, voted on its route, and put its generals on trial when the march was over. At one of those trials, soldiers accused Xenophon of striking them, and he described the winter crossing of the Armenian mountains and one soldier he had hit because the man was about to leave a wounded comrade behind. The army heard the reason and acquitted him. Every command he gave in the snow was owed an account later, and he gave it.

Telling a grown adult what to do feels awkward, and the awkwardness is what you were avoiding. They would rather have the instruction.

What Actually Works: Say Which Mode You Are In

Each of those four cases calls for command, and none of them calls for silent command. “We do not have time to debate this. We are rolling back. We will take the whole thing apart Thursday with the logs open.” A room hears that as leadership. The same rollback, ordered without a word about why the discussion stopped, lands as an override, and the room files the override as the real policy and the asking as decoration. People can carry a call they were not asked about when they were heard the rest of the time and told why this time was different. You control three things here: whether you named the mode, whether you gave the reason, and whether you set the date. Whether the room agrees is theirs, and the sentence is owed either way.

The four conditions are a test, and the leader is the one being tested. A crisis you have never seen is not a familiar crisis, and a preference is not a standard. Claiming a condition you do not meet is the same avoidance turned around: you want the room’s obedience without the room’s argument.

Ancient Greek split time into two words. chronos was the clock. kairos was the opening, the moment an archer’s arrow has to pass through the gap in the armor, the moment in weaving when the shuttle has to cross. Reading the kairos of a decision is the skill here. Aristotle’s name for the faculty that reads it was phronesis, practical judgment, the kind of wisdom that knows the case rather than the rule. The skill is a fit, the right mode for this room on this day, and splitting the difference, asking a little and telling a little, satisfies neither condition. You read the room, and then you say what you read.

That last step separates calibration from inconsistency, and the whole difference is whether you said why out loud. Four steps make it a habit:

1. Name the mode before the content. “This one is a decision.” “This one I want to argue about.” Say it in the first sentence of the agenda item.

2. Give the condition in one sentence. “This is a security standard and it is not moving.” “I have seen this failure before and we have four minutes.” “I have run this migration and you have not, so I am going first.” “You are new to this, and I am going to be specific for a few weeks.”

3. Date the return of the discussion. Command is a loan against the team’s voice, and a loan has a due date. “Retro Thursday.” “In thirty days I stop being this specific.” Without the date, the room will assume the shift is permanent.

4. Take the challenge afterward and lose in public when they are right. The retro where the incident commander says “the rollback was wrong and here is what I missed” is what buys the next command. A leader who never loses the after-argument has stopped holding it.

I run agents at my desk all day, and the instructions file is where this got concrete for me. The file has two kinds of lines. Some are open, like how to structure a module or what to name things, and the model is told it can argue with me about those. Some are closed: never touch the billing schema, never force-push to main. When I first wrote the file as one flat list, the model either treated every line as negotiable or froze on every line as sacred, because I had not said which was which. The non-negotiables live in the harness, where a smart model cannot talk itself past them, and everything else is the model’s to push on. That is the announced mode, with the reason next to every rule. It turned out to be the same file a team needs.

Diagnostic Questions

• When did you last make a call without asking the room, and did you say out loud that you were doing it?
• When did you last ask a question you had already answered in your head?
• Who on your team has autonomy they cannot use yet, and how long have you been calling that trust?
• After your last override, did the discussion ever come back, on a date you named?

Reading which of the four conditions is in the room is the same skill as reading what your team needs from you this week.

The room owes something here as well. When the leader has not said which mode this is, ask. “Is this a decision or a discussion?” is a fair question in any meeting, and a leader who cannot answer it has told you what you needed to know.

The challenge: For one week, write one of three words next to every item on every agenda you own: Decide, Discuss, or Inform. Say the word out loud before you start the item, and count at the end of the week. If every item said Discuss, you have been hiding in the room, and if every item said Decide, you are running a team that has stopped being asked.

Final Thoughts

The manager in the incident channel typed that question to protect their own record, and the team paid for it in nine minutes of climbing error rate.

The reason is owed because the people in the room can understand it. A command that arrives with its reason lets them obey from understanding, which is the only obedience that survives your absence, and a command without one trains them to watch your face.

The people who get trusted with a command are the ones who ask for real the rest of the time, and who say, out loud and with a reason, when they have stopped asking. A room that has watched you do both will hand you the call when the moment needs one.

If you want to work with people who know which meetings are decisions and say so before they start, MasteryLab is built for that.

FAQ

When should a leader stop asking for input and make the decision?

Four conditions justify the shift from facilitation to command: a non-negotiable standard such as security, safety, or ethics; a familiar crisis the leader has handled before with a short window for action; a sharp expertise gap where the leader knows far more than the room; and inexperience, where someone new needs direction before autonomy can help them. Outside those four, asking remains the right default. Inside them, the leader should decide, say out loud that they are deciding, and give the reason.

What is the difference between a directive leader and an inconsistent one?

Whether the shift in mode was announced. A leader who says this one is a decision, here is the reason, and we will revisit it Thursday is calibrating to the situation. A leader who switches from asking to ordering without saying so is inconsistent, and the team learns to treat the ordering as the real policy and the asking as decoration. Same decisions, different trust, and the only difference is the sentence that named the mode.

Should safety or ethics decisions ever be put to a team vote?

No. Opening a safety, security, or ethics standard for debate teaches the team that the standard is negotiable, even when the vote goes the right way. Standards should be stated, explained, and enforced by someone willing to be unpopular about them. The team’s voice belongs in designing the standard and in the review afterward, and it does not belong in deciding whether the standard applies this time.

How do you switch from consensus to command without losing your team’s trust?

Name the mode before the content, give the condition that triggered it in one sentence, put a date on when the discussion returns, and hold that discussion for real, including changing your mind in public when the team was right. Command handled that way is a loan against the team’s voice with a due date, and teams extend that loan to leaders who repay it.