Dividing things up when the numbers don't come out even

Dealing beats debating

The instinct when a pile needs splitting is to talk it through: who wants what, who had the good one last time, who's owed a favour. That conversation works for three items and falls apart for twenty, because every item becomes its own negotiation and the person most willing to argue slowly accumulates the best pile. Nobody's cheating, exactly. The process just quietly rewards persistence.

Dealing removes that. Shuffle the items, then hand them out in rotation like cards — one to you, one to you, one to you, back to the top. Everyone ends up with within one item of everyone else, and nobody chose which. It's the oldest fair-split algorithm there is, and the reason card games use it is that it's fast, obviously even, and impossible to lobby.

The Fair Division Tool does exactly that. Put the people in one box, the items in the other, one per line, and it shuffles the items and deals them round-robin into a card per person. Both lists are saved in your browser as you type, so a half-finished inventory survives a closed tab, and each division is recorded in the history below.

It divides by count, not by worth

Worth being clear about this, because "fair" is doing a lot of work in the name. The tool makes the number of items as equal as it can. It has no idea what any of them are worth. Deal out a garage and one person gets the mitre saw while another gets four rusted screwdrivers, and the tool will consider that a job well done — four items each, perfectly even.

That's fine when the items are genuinely comparable: shifts, chores, tickets, boxes of unsorted paperwork. It's not fine when they aren't, and the fix is to do the valuing yourself before you deal. Split the pile into tiers first and deal each tier separately — all the high-value items in one pass, the everyday stuff in another — so nobody's evenness is measured across categories that aren't alike. Or, if one item is worth roughly two of another, list the small one twice so it carries its real weight in the rotation. The tool can only be as fair as the list you hand it.

The leftover always lands at the top of your list

Here's the detail that actually matters, and it's easy to miss. The items get shuffled. The people don't. The deal walks down your people list in the exact order you typed it, over and over, until the items run out.

Which means: seventeen items between four people gives 5, 4, 4, 4 — and the person with five is always whoever you typed first. Not usually. Always. Run it again and the items in each pile change completely, but the counts don't budge, because the count pattern is determined by your typing order rather than by the shuffle. The randomness decides what you get, never how much.

If the piles come out even, none of this matters. If they don't, and the extra item is worth having, then the fairest-looking tool on the page has just handed the surplus to whoever happened to be typed at the top — which is usually the person doing the typing. The fix takes five seconds: before you deal, put the people list in an order nobody picked. Draw the order with a picker, or just have someone else type the names. Then the remainder lands somewhere no one chose, which is the whole point of using a tool in the first place.

It's worth flagging because a sister tool behaves differently. A chore roster spread over a week reshuffles every day, so having more chores than people evens out on its own — the doubling-up moves around. Fair division is a single deal. There's no next round to correct an imbalance in, so anything uneven in one pass stays uneven.

One small snag: identical names

Each person gets their own bucket keyed by name, so two people listed with exactly the same name share a bucket — you'll get one card with both of their shares piled into it rather than two cards. If you've got two Sams, distinguish them (Sam R. and Sam T.) and the deal comes out as you'd expect.

When you want a different tool

Fair division is for many items into few piles. If you've got exactly one thing per person — four people, four prizes — you don't want a deal, you want a matching, which pairs each person to exactly one item instead of bucketing. And if what you're really splitting is recurring work rather than a fixed pile, a roster that rotates over days will treat people better over time than any single deal can.

Split the pile before it splits the room

The Fair Division Tool runs entirely in your browser — nothing about your list leaves the page. For the neighbouring jobs, the Random Prize Assigner handles one-thing-each, the Two-List Matcher zips any two lists together, and the Random Chore Roster spreads recurring jobs across a week. Shuffle the people, deal the items, and let the pile stop being an argument.