The fair-roster problem: why the willing get over-worked

Ask anyone who builds a roster by hand and they'll tell you the same thing: it's not the empty slots that hurt. It's the guilt. Because when Saturday's bar needs covering and you're out of options, you already know who you're going to ask — the same person you asked last time.

This is the fair-roster problem. Hand-built rosters don't distribute work according to availability; they distribute it according to reliability. The people who always say yes accumulate shifts, the people who are hard to pin down quietly drop off the radar, and the coordinator absorbs the difference as guilt and chasing.

The reliability trap

It happens for an entirely human reason: asking is expensive. Every "any chance you could…?" text costs you something, and asking someone who might say no costs more. So under time pressure, you optimise for certainty — and certainty means the willing.

The result is a roster that looks full and a team that's slowly tilting. Three months later your most dependable volunteer is doing every second week, feels like they can't step back, and eventually steps away entirely. The roster didn't just fail them — it selected against them.

Hand-built rosters don't distribute work by availability. They distribute it by reliability — and that's how you lose your best people.

What "fair" actually means

Fair doesn't mean equal. Someone who's free every Saturday should serve more than someone free once a month — that's respecting availability, not exploiting it. A fair roster does three things at once:

• Works from real availability — not from memory, not from who answered the group chat fastest.
• Balances load within the available pool — if two people are free, the one with fewer recent shifts goes on.
• Shows its gaps honestly — an unfilled slot you can see beats a slot filled by guilt.

None of this is hard maths. It's just bookkeeping no human wants to do at 10pm on a Thursday — which is exactly why it should be software's job. When generation is automatic, the ask disappears, and with it the bias toward the reliable.

The part nobody automates: after publishing

Fairness doesn't end when the roster goes out. The drop-outs, covers and swaps that follow are where hand-run rosters revert to type — because a 9pm scramble always lands on the willing. A fair system asks everyone qualified to cover, first to accept wins, and keeps the tally straight.

Your reliable people aren't a resource to spend. They're the reason the whole thing works — and the roster should be protecting them, not leaning on them.