Distributing the unsocial hours fairly
Somebody has to take the call at eleven at night. In most distributed teams it is the same person every time, and nobody decided that.
Every distributed team generates work that lands outside somebody's normal hours: the meeting that suits three of four regions, the incident at four in the morning, the launch that has to be coordinated live. For broader context on planning work across time zones, Timeanddate Meeting Planner is a useful reference.
How that burden is distributed is decided in every team, whether or not anyone decides it. Where nobody decides, it accumulates on whoever is furthest from headquarters, most junior, or least likely to object.
The drift is structural, not personal
Meetings get scheduled to suit the majority, which usually means the headquarters time zone. Each individual decision is reasonable. The cumulative effect is that a specific group is permanently inconvenienced and everyone else experiences the arrangement as fine. For a practical software-side reference on overtime structures and unsocial hours, this explainer adds useful detail.
The people absorbing it rarely complain, because each instance is small and complaining about a single meeting looks unreasonable. It surfaces as resignation instead.
Meetings outside normal hours, out-of-hours incidents, and launches worked unsocially. The tally is almost always more skewed than management expects, because it builds one reasonable decision at a time.
Counting is the intervention
Simply publishing the tally alongside the schedule changes behaviour, because it moves the conversation from an argument about fairness in the abstract to a discussion about specific numbers.
The figures are easy to obtain and rarely gathered. A calendar export answers most of it, and where a team already records hours against work — with a spreadsheet, or with a tracker such as Monitask, which totals hours by person over a period — the out-of-hours share can be read off directly. The counting matters far more than the mechanism.
Rotate what can be rotated
A recurring meeting with no time that suits everyone should move. Rotating it is disruptive and it is the fair option, and the disruption is smaller than the retention cost of not doing it.
The same applies to on-call. A published rotation with a defined cycle prevents the slow accumulation of duty on the most agreeable person, which is a reliable predictor of that person eventually leaving.
Pay for it, or give the time back
Where unsocial hours cannot be avoided, they should be compensated — a premium, or time back in lieu, applied consistently and stated in writing.
This is partly fairness and partly a forcing function: work that has a cost attached gets questioned, and a meeting that generates three lieu-hours a week tends to get re-examined in a way that a free one does not. Note also that on-call and out-of-hours arrangements are regulated in many jurisdictions, and the rules on rest periods and compensation differ.
Protect the rest interval
Someone on a call until midnight should not be in a meeting at eight. This is the constraint most easily violated, because each meeting is scheduled by a different person who cannot see the other.
Making working hours and the previous night's calls visible in the calendar is what prevents it. Where that is not possible, a stated minimum rest interval, treated as a hard constraint rather than a preference, is the fallback.