How to Make a Fair Random Choice Between Several People
A fair random choice between several eligible people gives each listed person one equal chance, fixes the roster before the draw, and treats a rerun as a new draw rather than a correction. With 8 eligible names, each name has a 1/8 chance, or 12.5%.
A worked example with numbers
A volunteer group has eight names: Amina, Ben, Chen, Dara, Eli, Fran, Gio, and Hana. One person must open a meeting. Put the eight names in a list, state that each appears once, and use a one-name draw. The probability for Amina is 1 ÷ 8 = 0.125, which is 12.5%. If Chen says before the draw that they are unavailable, remove Chen first; the remaining seven people then each have 1/7, about 14.29%. Drawing from eight and clicking again whenever Chen appears is not the same rule, because observers cannot tell whether an unwanted result was silently replaced.
How to set the rule before the result
Start by writing four facts where everyone can see them: the purpose of the choice, the eligible names, the deadline for corrections, and the replacement rule if the chosen person declines. Remove blank rows and decide whether repeated names are intentional extra entries or a paste error. Then make one visible result with the random name picker or draw a complete order with the no-repeat draw order. Save the input list and the first output if the result affects a shared duty. A local random result can be transparent without collecting an account or a private roster.
Common mistakes that change the odds or the process
The common failure is confusing equality of people with equality of rows. Two entries for Ben and one for every other person give Ben 2 of 9 outcomes, or 22.22%, while each one-row person has 11.11%. Another failure is changing eligibility after the result appears: “Dara did it last time” is a reason to set a rotation before the draw, not a reason to spin again. A third failure is letting a late arrival enter only after seeing the winner. State a closing time and use the same rule for every late name.
Where this method stops being appropriate
Use chance for low-stakes allocation among people who accept the rule: a speaking turn, a casual game captain, or a household task after exclusions are known. Do not use it to decide employment, school eligibility, medical care, legal rights, safety work, discipline, access to scarce help, or a regulated prize draw. Those decisions need authority, criteria, records, and usually an appeal path; equal probabilities do not supply any of them.
How the random source fits into the rule
The browser method matters too. DiceDecide’s selectors use a uniformly sampled index and use browser cryptographic randomness when available; the implementation rejects out-of-range 32-bit values before reducing them so a range of 8 names is not favoured by remainder bias. The external description of cryptographically strong browser values is provided by MDN’s Crypto.getRandomValues reference. The social rule still matters more than the animation: a secure random number cannot repair a missing name or a hidden duplicate.
How to audit a people draw after it happens
Read the saved roster from top to bottom and count people, not just visible lines. In the eight-person example, the audit question is whether there were eight eligible people when the draw occurred, not whether the selected name was popular. Confirm that each person had one entry, that the announced deadline had passed, and that the output was retained. If a selected volunteer cannot serve, use the replacement rule that was written before the draw: for example, take the next name in a previously generated order. Do not rebuild the roster around the result. That distinction keeps a replacement from becoming a quiet second contest.
A fair outcome can still be inconvenient. The appropriate response is to change next week’s rota policy—perhaps exclude last week’s volunteer before the next draw—not to retrospectively alter this week’s probability. When names are sensitive, display initials or ask a neutral observer to verify the count. The useful record is minimal: date, purpose, eligible count, rule, and first result. It need not turn an informal household decision into surveillance.