New Math Keeps Chore Splitting Fair When People Show Up One by One
Splitting the office dishes could get fairer — no spreadsheet arguments required.
Imagine you're the unlucky person assigning chores. Not the fun kind of dividing, where everyone wants a slice of cake. This is the opposite: nobody wants the dishes, the late shifts, or the weekend on-call. Add a twist — people wander in one at a time, and once you hand someone their tasks, you can't take them back. A new paper from a team of computer scientists tackles exactly that problem.
Their goal is what researchers call "MMS" — Maximin Share. In plain English: everyone should get no more than the amount of unpleasant work they'd accept if they were splitting things fairly themselves, before knowing which pile they'd end up with. The team designed a method that keeps everyone within a modest multiple of that fair share, even with zero advance warning about who's coming or what they hate doing. Notably, their guarantees match the best results you can get when you know everyone in advance — which is a bit like matching your best-case plan while flying blind.
They also drew a hard line in the sand. For the simplest version of the problem, they showed that no method can ever beat a two-to-one guarantee — and theirs achieves three-to-one, remarkably close. That kind of "here's the best anyone could possibly do" proof is what makes the result interesting to other researchers.
So what's it for? Think hospital shift scheduling, cloud computers divvying up tedious background jobs, delivery fleets spreading the awkward routes, or a co-op housing group splitting cleaning duty when new members join mid-year. The catch is real: this is a theoretical paper, not a tool you can download. Real people have complicated preferences, and the guarantee is worst-case, not everyday. Still, it nudges the world toward fairer automation — and one day, software might do the chore-splitting argument for you.
- It solves 'chores' rather than 'prizes' — dividing unpopular work like dishes, night shifts, or tedious computer jobs instead of rewards.
- The method works even when people arrive one by one with no warning, matching the best results possible if you'd known everyone ahead of time.
- In the simplest case, the team proved a three-to-one fairness guarantee and showed nobody can ever beat two-to-one — so they're within striking distance of the theoretical limit.
Why It Matters
Fairer automated scheduling could mean less resentment over shared chores, shifts, and workplace drudgery.