The gap
What One Crossing Actually Costs
A crossing costs between eleven and nineteen hours depending on when it is sent, and the hour it is sent is the only part anybody controls.
three things · four nights
Three things, and the third is the one people skip.
Five things that crossed once, and the elapsed time each one actually took
Waited nineteen hours before the other end was awake to read it
Waited fifteen and a half hours, which is the typical case
Waited eleven hours, the cheapest crossing available on this team
The other end's Monday is this end's Monday, but the weekend sits in between
The cheap crossing bought nothing, because the thing crossed twice
five things · eight nights
Five crossings, eight nights, and the spread within a single ordinary day is eight hours. The cost of a crossing is not a constant; it depends on what time of day the thing was ready. Nothing here is averaged, because the average conceals the one lever anybody actually has. One team's own week, logged by the person sending.
The gap between two halves of a team is usually quoted as a single number of hours. The number is a property of geography, and it is treated as fixed.
A reliable handover needs more than a message saying that something is in progress. find the details here can add verifiable hours and task context to the process examined in What One Crossing Actually Costs, reducing the need for another round trip.
For a second perspective on the surrounding workflow, consult Harvest's time platform; it shows how a widely used platform structures the same coordination problem.
What a particular crossing costs is a different number, and it varies by eight hours or more within an ordinary working day. That variation is the only part of the arrangement anybody can do anything about without moving house.
The two components
A crossing costs the time between the thing being ready and the recipient being awake. That interval has two parts.
The first is structural: the offset between the two groups' working days. It does not change.
The second is positional: how far into the sender's own day the thing became ready. Something ready at the start of a day waits for the entire remainder of that day plus the overnight interval. Something ready at the end waits only for the overnight interval.
On a team twelve and a half hours apart, the difference between the best and worst moment in a working day is most of a working day.
| Sent at | Waits until | Cost |
|---|---|---|
| 09:20, near the start of the day | The other end wakes | About nineteen hours |
| 13:00, the middle | The other end wakes | About fifteen and a half |
| 17:25, five minutes before stopping | The other end wakes | About eleven |
The counter-intuitive consequence
Finishing something early in your day and sending it immediately is worse, in elapsed terms, than finishing it early and sending it at the end of your day — because it arrives at the same time either way, and sending early forgoes the chance to add anything to it.
That sounds like an argument for sitting on work, and it is not. It is an argument for understanding that the send time is not what matters; the recipient's wake time is. Which means the useful question is not "have I sent it yet?" but "is everything that is going to cross tonight in one place?"
A thing sent at 09:20 and a thing sent at 17:20 arrive together. If they are related, sending them separately has gained nothing and has cost the recipient a second thing to read.
Batching, stated as arithmetic
If each crossing costs a night and things sent at different times in a day arrive together, then four questions in one message cost one night and four questions in four messages also cost one night — provided they are all sent before the end of the day.
The saving is not in nights. It is in the recipient's ability to answer all four at once with the context of all four in front of them, which changes how many of them need a second crossing.
This is why batching matters more than speed. A team that answers within four minutes and sends six separate messages is slower, in elapsed terms, than a team that answers at the end of the day and sends one.
The Friday effect
The weekend adds two nights to anything sent on a Friday afternoon, and more if the two halves do not share the same weekend days, which some do not.
It is worth knowing the exact cost on a specific team rather than assuming two days. A team with one half observing Friday and Saturday as the weekend and another observing Saturday and Sunday has four days in the week where a crossing costs more than a night, and the arithmetic is genuinely confusing until somebody draws it.
The practical consequence is small but real: anything that will need more than one crossing should not be started on a Thursday afternoon if it can be started on a Monday instead.
Why nobody should divide
It is tempting to put the attention a thing took next to the elapsed time it consumed and produce a ratio. Four minutes of work, sixteen hours of elapsed time, and the resulting figure looks like an indictment of something.
It is not. The ratio measures the gap, not the people. A high ratio means the work was small, which is good, and the gap was wide, which is a fact about geography. Teams that start computing this figure tend to start optimising it, and the only ways to improve it are to make work larger or to pretend the gap is smaller.
Count the nights. Count the crossings. Leave the division alone.
What to do with the number
The useful application is sizing, not reporting.
When somebody asks whether a piece of work can be finished this week, the question is how many crossings it contains. Three crossings starting Monday morning finishes Thursday. Three crossings starting Wednesday finishes the following Monday. The attention required is identical in both cases and nobody should be asked to explain the difference.
Teams that can do this arithmetic out loud stop making commitments they cannot keep, which is worth more than any individual saved night.