A draw made in front of an audience has two jobs: produce an impartial result, and be seen to produce one. The randomness handles the first and the presentation handles the second — visible entries, a slow enough spin to follow, a winner in large type, and a record of what has already come up. A perfectly fair draw that nobody could watch settles nothing.
This is the whole reason a wheel exists alongside a list picker. The list is faster and the wheel is witnessed.
What makes a draw look fair?
Four things, in rough order of how much complaint they prevent.
- The entries are visible before the draw. If the room can read the list, nobody can claim a name was missing or added.
- The result takes time to arrive. An instant answer looks decided; a spin that decelerates looks discovered, even though both were drawn the same way.
- The winner is legible from the back. Large type, high contrast, and no reliance on colour alone to identify the segment.
- The history is on screen. Previous winners in order removes every argument about whether someone has already had a turn.
None of those change the odds. All of them change whether the result is accepted.
Is the animation the draw?
No, and it is worth being straightforward about that. The result is drawn from the browser cryptographic random source before the wheel starts moving, and the wheel is then rotated to land on it. The spin is presentation.
That order is deliberate. Deriving the winner from where an animation happens to stop makes the outcome depend on frame timing, on how fast the device is, and on whether the tab was in the background — all of which are worse sources of randomness than the one already available. Drawing first and animating second keeps the fairness in the part that is designed for it.
It also means nothing about the motion can be nudged. There is no physics to lean on and no point at which a slower or faster spin changes anything.
Should the winner be removed?
For working through a group, yes. Turning on removal makes each spin draw from whoever is left, so a class of thirty is covered in thirty spins with nobody appearing twice.
Leave it off when a repeat is legitimate — picking a discussion topic, a warm-up, a category. And be explicit about which mode you are in before the first spin, because a room that expects removal and sees a repeat will assume the tool is broken rather than that the setting is off.
The difference is not cosmetic. With removal off, thirty draws from thirty names leave about eleven people never picked, which reads as favouritism whatever the arithmetic says.
How many entries can a wheel carry?
More than it should. A wheel divides itself into equal segments, so a class of thirty produces segments of twelve degrees each, and the labels stop being readable well before the tool stops working.
Past roughly twenty entries a wheel is doing the audience no favours, and a plain list picker with a large result is the better instrument. The choice is about legibility rather than capacity: use the wheel when the room needs to see the field, and the list when the field is too big to see.
For a projector specifically, contrast matters more than colour. Segments that alternate between two tints of one accent read cleanly at the back of a hall; a rainbow of similar-luminance colours does not, and disappears entirely for anyone with a colour vision deficiency.
Segment count also affects how long the spin needs to be. A wheel with six entries reads clearly at speed and a wheel with twenty is a blur until it has almost stopped, so a longer deceleration is doing real work on a crowded wheel and is merely padding on a small one.
What about accessibility and motion?
A spinning wheel is a large moving element, and a proportion of any audience finds sustained motion uncomfortable. The system-level reduced-motion preference is the right signal to honour, and the fallback is not a broken tool — it is the same draw, announced without the rotation.
The result also needs to survive without the animation entirely. If the winner is only communicated by where a pointer ends up, then anyone who looked away, anyone on a screen reader, and anyone whose browser skipped the animation has no result at all. The winner in text, large, is the actual output.
Questions people ask
Can I weight an entry? Not directly. Adding a name twice gives it two segments and double the chance, which is visible on the wheel and therefore easier to defend than a hidden weighting would be.
Will it remember my list? No. Nothing is stored between visits and nothing is sent anywhere, so a class list has to be pasted each session and leaves nothing behind afterwards.
Does it work with the animation turned off? Yes. The draw is independent of the motion, so the result is the same either way.
Is it really equal probability? Every remaining entry has an equal chance on every spin. Equal segment sizes are a consequence of that rather than the cause of it.
Fairness comes from the draw and acceptance comes from the presentation, and a public draw needs both. Spin the wheel is built for the room, the random person picker is the faster version for a daily rota, and the yes or no wheel is the two-option case.