The criteria that actually matter
Almost every choice between random pickers comes down to four questions:
- How many outcomes do you need? Two, six, twelve, arbitrary?
- Should the outcomes be equally likely? Or do some need more weight than others?
- Does the process need to be visible to everyone present? A flip you both watched is more convincing than a number a website returned.
- Does the result need to be repeatable or auditable later? For example, in a tournament draw.
Pick the simplest tool that satisfies all four. Adding complexity beyond that doesn't add fairness; it adds opportunities for bugs and disputes.
Quick comparison
| Tool | Outcomes | Best for | Weak when |
|---|---|---|---|
| Coin flip | 2, equal | Binary decisions; kickoffs; "you go first"; A vs B. | You need three or more options, or weighted odds. |
| Six-sided die | 6, equal | Picking from a small set; ordering players; board games. | You need fewer than 6 or more than 6 options without re-mapping. |
| n-sided die / digital dice | n (4, 8, 10, 12, 20, 100…) | RPGs; situations where you've already labelled outcomes 1…n. | People treat the result as more "official" than it is. |
| Spinner / wheel of names | Arbitrary, can be weighted | Drawing one name from many; teacher cold-calls; raffles. | Watchers don't trust the wheel software; doesn't scale to thousands of entries cleanly. |
| Random number generator | Arbitrary range | Picking a row from a list; programmatic uses; large draws. | The outcome feels abstract and "computery" — bad for in-the-room decisions. |
| Cards / shuffled deck | Up to 52, equal, no replacement | Multi-step draws where each pick reduces the pool; tournaments. | Single binary picks — heavy ceremony for a 50/50. |
When a coin flip is the right call
Reach for a coin flip when:
- You have exactly two options.
- Each option should have equal weight.
- The decision is low-stakes or procedural — you'd be fine with either outcome.
- You want a fast, visible process that everyone present can verify (whether it's a real coin or the on-screen flip on the simulator).
Examples: who serves first in a friendly tennis match, who gets the last slice of pizza, picking between two near-identical job offers (after the analysis is done and a tie remains).
When a coin is the wrong tool
Don't use a coin flip when:
- You really have more than two options. Mapping three options onto coin flips ("flip twice, T-T means option C") is fragile and people forget the mapping mid-decision.
- The options are not equally desirable. If one is much better, you don't want a 50/50; you want either to commit, or to think harder, or to use weighted picking.
- You'd be unhappy with one outcome. The reaction "best of three?" after a flip is the giveaway: you weren't really willing to accept a coin flip's verdict.
- You need an audit trail. A live flip witnessed by two people is fine for casual use; for anything formal, use a method that records the inputs and process.
Worked example: choosing between three options
Suppose three friends are deciding which restaurant to go to: A, B, or C. The naive approach is two coin flips ("HH = A, HT = B, TH = C, TT = re-flip"). It works, but every fourth flip is wasted, and people forget the mapping when they're hungry.
Better tools for this case:
- A six-sided die: 1-2 = A, 3-4 = B, 5-6 = C. One roll, no waste, easy to remember.
- A random number 1–3 from any picker.
- A wheel-of-names tool with three slices.
The coin can do the job, but it's the wrong shape for it.
Worked example: weighted decisions
Suppose you want option A 70% of the time and option B 30% of the time. A single coin flip can't do this — you'd have to flip ten times and use a threshold, which loses the appeal of "one flip, done". For weighted choices use a number generator (roll 1–10; 1–7 picks A, 8–10 picks B) or a weighted spinner.
What about "fair-feeling" vs. "actually fair"?
For two people who both watch a real coin land on a table, the flip feels fair — and that perception matters as much as the underlying probability. A digital flip on this site is, mathematically, at least as fair as any plausible physical coin (see the probability of a coin flip). But if either person doesn't trust the screen, the perception side is broken even if the maths isn't. In those cases, use a real coin and let both people see the catch.
Decision checklist
- How many options? 2 → coin. 3–6 → die. More → number generator or wheel.
- Are options equal? If no → use a weighted picker.
- Will both parties accept the result? If no → don't randomise; have the conversation you're avoiding.
- Need an audit trail? If yes → use a process that records inputs.