The gap
The Unit Is the Night, Not the Hour
Teams with no overlapping hour keep measuring in hours out of habit, and almost every decision gets clearer the moment the unit changes.
three things · four nights
Three things, and the third is the one people skip.
The same six things, measured twice: once in working hours, once in nights
Four minutes of somebody's attention, spread across sixteen and a half hours of calendar
Eleven minutes of attention in total, and three days before the asker could act
Under an hour of reading and writing, spread across the better part of a working week
The only row where the elapsed time and the effort are anywhere near each other
Two hours of work at one end, two at the other, and a night in between that nobody could use
No quantity of nights produces it, so the hours column is not the interesting one
six things · ten nights · one that no number of nights settles
Six things, ten nights, and roughly four hours of actual attention between them. The two measurements disagree by a factor of thirty, and they disagree in the same direction every time. Nothing here divides one column by the other: the resulting fraction reads as idleness, when what it records is that there was nobody awake. One team's own week, logged as it went.
A team whose halves are never awake together still talks about work in hours, because everybody has always talked about work in hours. The habit survives the arrangement that made it sensible.
The operational risk in The Unit Is the Night, Not the Hour is losing the difference between waiting and working. visit the website can preserve that distinction through timesheets, activity records and project-level reporting.
For broader context, Clockify's time tracker offers an official example of the tools teams use around this kind of process.
It produces estimates that are correct and useless. A change takes two hours. It also takes three days, and the person asking when it will be done wants the second number.
Why hours mislead
An hour is a measure of attention. It answers the question "how much work is this?" and it answers it well. In a team sharing a room, or even a team sharing a morning, the answer to "how much work is this?" is close enough to the answer to "when will it be done?" that nobody has to distinguish them.
Remove the shared hour and the two numbers come apart. The attention a thing needs stays what it was. The elapsed time becomes a function of how many times the thing has to cross the gap, and crossing the gap costs the same fifteen or sixteen hours no matter what is crossing.
A one-line answer and a four-paragraph answer cost the same night. That is the fact the hour cannot express.
What a night actually is
A night, in this sense, is not darkness and is not eight hours. It is the interval between a thing being ready to move and the next person being awake to move it.
For a team whose halves sit twelve and a half hours apart, a night is usually somewhere between fourteen and eighteen hours, and its length depends on what time of day the thing became ready. Something finished at the start of one half's day waits longer than something finished at the end of it. Nobody can do much about that, and the variation is small enough to ignore for planning.
What matters is that the interval is roughly constant and roughly indivisible. You cannot have half a night.
The arithmetic that follows
Once the unit is the night, planning becomes a counting problem rather than an estimating problem. The question is no longer "how long will this take?" but "how many times does this have to cross?"
One crossing: a question with a single unambiguous answer, a piece of work handed over complete, a review with nothing contentious in it.
Two crossings: anything where the first response is likely to be a clarification rather than an answer. Most questions about a decision somebody else made land here.
Three to five crossings: anything that requires two people to converge rather than one person to inform another. Agreeing an approach, not just approving one. Settling a disagreement about priority.
Beyond five: usually a sign that the thing is in the wrong format. Five crossings is a working week for something that would have taken twenty minutes in a conversation, and the useful move at that point is to change the format rather than to keep counting.
| Crossings | What it usually is | Elapsed |
|---|---|---|
| 1 | A question with one answer, or finished work | One night |
| 2 | Anything whose first reply will be a clarification | Two nights |
| 3–5 | Two people converging rather than one informing another | Most of a week |
| 5+ | A conversation being conducted by correspondence | Change the format |
Where the unit breaks down
The night is a poor unit for two kinds of work and it is worth saying so rather than pretending otherwise.
Work that is genuinely parallel — two people building separate things that do not touch — does not cross anything, and counting nights for it produces nothing. Measure that in hours, as usual.
Work that is continuous rather than discrete, such as an incident that somebody is actively holding, does not come in crossings either. It comes in handovers, which are a different and more expensive thing, and which have their own section here.
The unit is for the middle case, which is most work: discrete pieces that need somebody else's input before they can continue.
What changes when you adopt it
Three things, in practice.
Estimates stop being wrong in a way that looks like dishonesty. "Two hours of work, four days of elapsed time" is a sentence people can plan around, and it stops the slow erosion of trust that happens when everything takes much longer than anybody said it would.
Batching becomes obviously correct rather than vaguely virtuous. If each crossing costs a night, sending four questions in one message costs one night and sending them in four messages costs four. That is not a style preference, it is arithmetic, and people act on arithmetic.
And the cost of a badly formed question becomes visible. A question missing a noun does not cost the extra thirty seconds it would have taken to include the noun. It costs a day. People who would never describe themselves as careful writers become careful writers once somebody puts that number in front of them.
The honest limit
Changing the unit does not make anything faster. It makes the real shape of the week legible, which is a prerequisite for deciding what to do about it, and it is not a substitute for doing anything.
A team that counts nights accurately and changes nothing has the same week it had before, described better.