How the wheel picks a winner
The order of operations is the opposite of what the animation suggests. The winner is chosen first, by drawing a uniform random index across your list of options, and the rotation is then computed so that the wheel comes to rest with the pointer on that slice. The spinning is a presentation of a decision already made, not a physical simulation whose outcome emerges from the motion.
That design has a useful property: the odds are exactly uniform regardless of how the animation renders, how long it runs or where it happens to stop visually. Every option has precisely a one in n chance. A physics-based wheel would introduce subtle bias from its starting position and friction model.
The final angle is built from six full turns for the visual effect, plus the offset needed to bring the winning slice's midpoint to the pointer at the top, plus a small random jitter within the slice so the result does not always land dead centre. The transition uses a decelerating cubic-bezier curve over four seconds, which is what produces the slowing-down feel.
Equal slices and what that means for your options
The wheel divides 360 degrees evenly by the number of options, so every entry gets an identical slice and an identical chance. There is no weighting: an option cannot be made more likely by any setting in the tool.
The workaround is duplication. Entering an option twice in a list of ten gives it two slices and therefore a two in ten chance. That is a legitimate way to build a weighted draw as long as you are comfortable with the duplicate labels being visible on the wheel, and it is how most people express a rough weighting here.
Labels are truncated on the wheel face when they run past about a dozen characters, which is a rendering limit rather than a limit on the option itself. The winner announcement shows the full text, so long entries still read correctly once chosen; short labels simply look better on the wheel.
Drawing with and without replacement
By default every spin is independent, which is drawing with replacement: the same name can win twice in a row, and over ten spins of a four-option wheel you should expect some options never to come up. That is correct behaviour and not a malfunction, though it surprises people who expect a rotation through the list.
The remove winner action converts the process to drawing without replacement. Taking the winner out and spinning again gives each remaining option a larger share, and repeating until the list is empty produces a full random ordering of your entries. That is the right procedure for assigning a presentation order, picking teams, or drawing several distinct prize winners.
Deciding which of the two you want before you start matters, because the difference is not cosmetic. A raffle with three prizes usually means three draws without replacement, while a daily decision about where to eat lunch means with replacement, where yesterday's result is irrelevant.
When a random pick is the right tool
Randomness is most valuable when the options are genuinely close in value and the cost of deliberating exceeds the cost of picking wrong. Choosing a restaurant, picking which of several equally ready tasks to do first, or deciding who speaks next all fit. Spinning also removes the appearance of favouritism, which is often the real reason to use it in a group.
It is a poor tool when the options are not equivalent. If one choice is clearly better, a wheel launders that judgement into chance and hands you a worse outcome with a clean conscience. A useful diagnostic: if you would be disappointed by one of the results, you already have a preference and should act on it rather than spin.
Nothing is stored, so the option list lives only in the page and reloading clears it. For a repeated draw with the same list, keep the text somewhere you can paste back in.