The probability model is a count of entered sectors
Every non-empty line creates one equal sector. If a list has n distinct lines, each line has probability 1/n on a spin. The browser chooses an integer position from that finite list; visual rotation is not part of the probability calculation. This assumes each entered line is meant to be one separate, acceptable outcome. A duplicate is not harmless decoration: two copies of “Soup” in a four-line list give Soup probability 2/4, while each other label has 1/4.
Worked choice: three lunch options
Enter Soup, Sandwich, and Leftovers. There are three sector indices: 0, 1, and 2. If the local result is index 1, Sandwich is chosen, and its chance before the draw was 1/3 = 33.33%. Now add Soup a second time. There are four positions; Soup owns positions 0 and 3, so it has 2/4 = 50%, while Sandwich and Leftovers each have 25%. The arithmetic exposes whether a preference was intentional before anyone sees the winner.
Construct a decision population, not a wish list
A useful roulette list contains alternatives at the same decision level. “Walk by the river” and “do something outdoors” overlap, so the second label gives one real activity another route to selection. “Museum” and “the museum” may be accidental duplicates. Remove venues that are closed, choices over budget, and options a participant has already rejected. Random selection can break a deadlock among acceptable alternatives; it cannot repair an unfair or incoherent menu.
State the redraw rule before the result exists
A closed restaurant, sudden weather restriction, or a missing ingredient can make an outcome impossible. Decide in advance whether that label is removed and a new spin is made, whether a replacement list is built, or whether the group chooses manually. Selecting a new rule after a disliked result turns an equal-probability draw into an unrecorded preference. For a casual choice, record the complete list, the winner, and any pre-agreed exception rather than saving only the winning word.
Where equal sectors stop being a suitable model
This page does not compare prices, travel times, risk, nutritional needs, or legal consequences. It also cannot verify that someone did not edit the list between spins, because no server-side record exists. Decisions involving money, safety, rights, or medical care need criteria and accountability before chance, not a more colourful wheel. The result is best treated as a tie-breaker for a low-stakes list whose members already accept every possible sector.
A five-second audit before pressing spin
Count the lines aloud. With five lines, the denominator must be five; each intended outcome should appear once and be understandable without a later interpretation. If the group deliberately wants an option twice as likely, enter it twice and announce the resulting 2/6 share. That visible count is enough to test the model without trusting an animation.
Equal sectors implement a categorical draw
A wheel is just a list indexed from zero through n − 1. With seven distinct labels, the first index, the last index, and every index between them each represent 1/7, about 14.29%. The circular picture can make neighbouring labels look related, but no probability attaches to location, colour, or a dramatic near miss. If “coffee” occurs once and “tea” occurs once, they are alternatives. If a label means “choose either coffee or tea later,” it is a compound outcome and should be split before the draw.
A five-option travel example
Suppose a group has already accepted Bus, Train, Bicycle, Walk, and Stay Home. Five entries produce five equally likely outcomes. A spin selecting Bicycle means only that Bicycle occupied one of the five slots; it does not claim it is the cheapest or healthiest option. If rain removes Bicycle before spinning, four options remain and each has 25%. If rain is checked after Bicycle wins, removing it then changes the observed procedure. The relevant distinction is eligibility at the time the list was closed.
Duplicates are a crude weighting language
Repeating one label changes a roulette from equal choice to a weighted choice. A menu of Read, Exercise, Call a Friend, and Read gives reading 2/4 while the other two actions have 1/4 each. That can be useful when a group deliberately grants two entries to a popular option, but it is easy to hide in a long pasted list. Prefer one line per genuinely equal option. When unequal chance is the goal, state the intended ratio in words and check the count: two, one, and one means 50%, 25%, and 25%.
Chance has a narrow role in a group decision
A roulette can settle a tie only after the group has screened for cost, access, allergies, safety, and consent. It cannot aggregate different preferences or compensate someone who always gets the inconvenient result. For recurring choices, rotate a veto, maintain a record of prior outcomes, or use agreed criteria instead of hoping repeated spins will balance themselves. The useful output is a transparent tie-break among already acceptable labels, not a reason to ignore a material difference between them.