How to run a Secret Santa, and how to check the tool you run it with.

How these guides are written

Each guide is anchored to something you can check without taking our word for it: an arithmetic result, a test you can run in your own browser, or a sentence from a published product page. The counts quoted throughout come from the same code that runs the draw and are asserted on every build — nine valid draws for four people, 1,334,961 for ten, 4,752 for eight people arranged as four couples, and zero for a household of four inside a group of six. Where a guide makes a claim about behaviour rather than arithmetic, it tells you how to watch that behaviour yourself instead of describing it and asking you to agree.

Where the honest answer is a trade rather than a rule, the cost is printed next to the recommendation. Repairing a drop-out after you have closed the draw tab costs one dead edge and no live ones. Adding a late joiner costs one disclosed pairing, given by consent, in the open. Printing fold-over slips puts the entire draw on paper in front of whoever collects it from the printer. Twenty-five people means twenty-five separate messages, because there is no server here to send them from. The first guide runs a four-test audit on this site as harshly as it runs it on anyone else's, and a Quick draw fails one of the four tests: we would rather write that down here than have a reader discover it.

Two limits apply to the set as a whole. The repair and announcement guides describe cooperative protocols for a group that already knows one another — they are social agreements, not mathematics, and anyone in the group can post a false announcement. And any fact quoted from someone else's product or platform carries the date we checked it, because that kind of fact goes stale quietly. If something here no longer matches what you find, the date is the first thing to look at.