TRUST

Trust is putting something that matters to you in someone else's hands.

Why This Matters

When It's Missing
It shows up as work. Extra approvals, a status meeting that exists because someone got surprised once, a copy of the document you keep for yourself. Each one is individually defensible and the total is enormous — most of what makes an organization slow is somebody, somewhere, checking.

And the checking is not neutral. In an experiment where people could either restrict what the other party was allowed to do or leave them free to choose, most of those who were controlled performed worse than those who weren't. Asked why, they described the control as a signal of distrust. Monitoring doesn't only cost time. It tends to produce the behaviour it was designed to catch.
When It's Present
Decisions get made where the information is, instead of travelling up to be approved by someone with less of it. That's the return: speed, but also better answers, because the person closest to the problem is usually the one who understands it.
Start Here
Tell one person exactly what you're trusting them with, in words. "This is yours. I'm not going to check it." Then don't. Most trust is never actually communicated — it's assumed by one party and unnoticed by the other, which means nobody gets the benefit of it.

Go Deeper

Trust is not one thing, and that's the most useful thing to know about it. The standard model in the research breaks trustworthiness into three parts: ability — can they do this; benevolence — do they have my interests in mind; and integrity — do they hold to principles I can predict. When someone says "I don't trust him," they almost always mean one of the three, and the three have completely different remedies. A competence gap needs training or a smaller task. A benevolence gap needs a conversation about interests. An integrity gap is the serious one, and it isn't fixed by either.

If you can check, it isn't trust. The definition in the literature turns on vulnerability: trust is the willingness to be exposed to someone else's choices without the ability to monitor or control what they do. That's uncomfortable, and it's the point. What most people call trust is a verification system with good manners. Real trust always costs something, because it always involves the possibility of being wrong.

It builds slowly and breaks fast, and there's a reason. A single failure carries far more weight in your assessment than a single success, and it's not irrational — one betrayal tells you more about a person than one delivery does. What follows from that is practical: you cannot recover trust at the speed you lost it, and the attempt to do so reads as pressure. What actually works is a long sequence of unremarkable, kept commitments, which is slow and boring and the only route there is.

Trust and liking are different. You can like someone enormously and not trust them with a deadline. You can dislike someone and trust them completely. Confusing the two is how organizations end up handing important work to the most personable person in the room, and how families end up unable to say that someone they love is unreliable.

That's what trust is — not a feeling about a person, but a decision to be exposed to them.

The Trust Spectrum

See the Full Spectrum

Too Little	The Middle	Too Much
A new approval step every time something goes wrong	Trusts specific people with specific things	Confidence mistaken for competence
Delegating the task, keeping the decision	Extends it first, in a size they can afford	The pattern excused for the tenth time
Controls that quietly signal suspicion	Says out loud what's being trusted, and why	No verification even where the stakes are high
Information held as leverage	Repairs quickly and without ceremony	Loyalty maintained past the evidence
One failure treated as a permanent verdict	Rebuilds at the speed it actually rebuilds	Trusting to avoid the discomfort of doubting

A note on too little:

Every control here was installed for a reason, usually a real one. The question isn't whether it was justified when it was added — it's whether the thing it was protecting against is still more expensive than the message it sends.

Practical Applications

Business

1. Work out which of the three is missing. Before deciding you don't trust someone, name it: is it that they can't, that they won't look out for you, or that you can't predict them? The first two are usually fixable and often aren't even about them. The third is a different conversation.

2. Remove one control and say that you're removing it. Pick a check that exists because of something that happened once, years ago. Take it out, and tell people why. The removal is worth more than the time it saves, because the message travels further than the process did.

3. Extend it first, in an amount you could afford to lose. Trust is reciprocated more often than people expect, but somebody has to go first, and it will be the person with more power. Size it so that being wrong is survivable, then actually let go of it.

4. Repair within a day. When you break something — miss the thing, say the thing — name it fast and plainly. The gap between the breach and the acknowledgement does more damage than the breach, because it's the part that looks deliberate.

Family

1. Believe them the first time. Where there's no real reason to doubt, don't check. The message a child takes from being verified isn't you were careful, it's you expected me to be lying — and a child who expects not to be believed stops bringing you the difficult version.

2. Give them something that genuinely matters. Not a chore invented to teach responsibility. Something real, where the consequences are real and visible. Trust that's never actually at stake doesn't register as trust.

3. Keep what they tell you. Anything told to you in confidence stays with you, including from the funny anecdote at dinner. Every relationship in a family is being assessed for whether things travel out of it.

4. Tell them now what you'd do if they called you at three in the morning. Say it before it happens: "If you're ever somewhere you shouldn't be, I'll come and get you, and we'll talk about it another day." The promise has to be on record in advance or it can't be used when it's needed.