Drawing Lots

Where 제비뽑기 is fine, and where it fails a legal/audit standard

제비뽑기 has been a social-choice mechanism for centuries — ancient Athenians used kleroterion-based sortition to fill around 1,100 government roles per year, and Korea has used it for everything from deciding dinner payers to assigning dormitory rooms. For any situation where (a) everyone involved agrees the lot is being drawn, (b) the stakes are low enough that nobody will litigate the outcome, and (c) participants trust the apparatus, a digital lot like this works well. That covers probably 95% of household, classroom, and small-office uses.

Where it stops being enough: any decision that could be challenged or needs to be reproduced later. Three concrete examples.

• Government procurement tie-breakers. Korea's 국가계약법 시행령 Article 47 permits drawing lots to resolve tied bids — but requires the draw to happen in front of all parties (or their designated witnesses) with a recorded protocol. An in-browser click by one party doesn't satisfy the witness requirement.
• Legal contracts. Deciding which of two co-tenants gets the bigger room via 제비뽑기 is fine as an informal agreement; using it to allocate a 2억원 inheritance is not — courts look for documented, auditable process. The Math.random call here is not auditable.
• U.S. civil procedure. Federal courts use computerized random jury selection from a voter-roll master list (28 U.S.C. § 1863), with the seed and algorithm certified and the source list retained. A one-click browser draw wouldn't pass challenge.

Edge case: setting "winners = 1, losers = 4" on this tool gives each slot a 20% win rate only if exactly one person draws. If two people share one lot set — drawing sequentially — the second draw's probability is conditional on the first (sampling without replacement), not 20%. For binding group draws, do the full assignment upfront and then reveal.