Match the sample space to the decision
Chance is only fair after the possible outcomes have been defined. A coin has two equally likely faces, a three-move game has three states, a numbered lot has one state per ticket, and a card deck changes after each card is removed. Choosing the wrong sample space is a bigger error than using the wrong random button.
Small examples, different mathematics
A coin gives each of two people probability 1/2. Four names drawing one short straw gives each person probability 1/4. Five cards from a standard deck are a without-replacement sample, so the chance of a second heart changes after the first card appears. A best-of-7 series is different again: it ends when one side reaches four wins rather than after seven compulsory flips.
Make the rule visible first
Write who is eligible, what each outcome means, whether repeats are allowed, and what happens on a tie. A transparent rule prevents a convenient redraw or a late change of labels from looking like chance. These tools supply a local random outcome; they do not decide whether the underlying allocation is appropriate, safe, or permitted.
Where quick choices stop helping
Randomness is useful for equal claims and low-stakes tie-breaks. It is not a substitute for a queue, a contract, informed consent, skill requirements, or a safety procedure. When the stakes depend on facts that a random draw cannot represent, settle those facts before using any of the tools in this collection.
Check the mechanism, then the agreement
A reproducible random result can show that the stated selection rule was followed, but it cannot prove that the rule was wise. Keep the names, ticket range, deck convention, or series length with the outcome. That distinction matters when people later need to distinguish a fair draw from a fair allocation.
Choose an outcome structure, not a familiar ritual
A raffle needs a numbered pool; a card hand needs a deck convention; a two-person tie needs two labels; a game needs a payoff rule. Those choices cannot be swapped merely because each uses an unpredictable result. Before a group begins, name the object that is being allocated and whether an earlier selection is removed. That one sentence distinguishes a one-off privilege, a sequence of winners, a deal, and a contest, and gives participants a way to audit the result later. Clear rules make a surprising result explainable rather than merely memorable.
Record the evidence that defines the draw
Different selection methods leave different evidence. A coin tie-break needs the two labels and the face assigned to each person. A lot selection needs the complete eligible register, including any withdrawn number. A card draw needs the deck convention and whether the card stays out after it appears. A series needs the winning threshold and the results actually played, because a contest can end before its maximum length. Saving that small input record makes a later check possible without claiming that the random output itself proved the allocation was sensible.
For example, a committee choosing one of eight equally qualified presenters can publish eight names, state that one numbered position wins, and retain the selected number with the roster. If a ninth name was never eligible because the person was unavailable, the record explains why the chance was 1/8 rather than 1/9. By contrast, a dispute about who has the strongest claim should be resolved with criteria before any number is selected. Randomness can distribute an equal remaining chance; it cannot convert unequal evidence into an equal entitlement.