Intervals represent unequal chances
Weights create adjacent intervals on a total scale. Standard, Express, and Overnight with weights 6, 3, and 1 total 10, so their probabilities are 6/10, 3/10, and 1/10. The browser chooses a random point from that total interval and returns the label whose interval contains it. A weight is relative support, not a score or a guarantee.
A delivery lottery with numbers
With the default weights, Standard covers the first six tenths of the interval and has 60% chance; Express has 30%; Overnight has 10%. In 100 independent trials the expected counts are 60, 30, and 10, but a run of ten might contain no Overnight result. Probability describes the rule before sampling, not a quota that each short batch must satisfy.
Normalize only when it helps
The entries 6, 3, 1 mean the same odds as 60, 30, 10 or 2, 1, 1/3 because multiplying every weight by a positive constant leaves the ratios unchanged. A zero weight means no selectable interval; a negative weight has no probabilistic interpretation here. Keep the source for every weight so users can tell whether it represents tickets, inventory, or a policy choice.
Weights need a rationale
This picker is unsuitable when each participant has an equal right, because assigning unequal numbers silently changes the allocation. It also does not produce a weighted ranking without repeats: after one selected item is removed, the remaining total changes and a new rule is needed. Use it only where unequal chances are intentional and explainable before the draw.
Test the consequence of the ratio
Before using unequal chances, translate the weights into percentages and ask whether the largest interval is defensible. A 6:3:1 rule gives Standard six times Overnight's chance, which may be appropriate for six tickets versus one but not for a participant whose name was entered mistakenly. Save both labels and weights with the outcome. That record lets a reviewer distinguish an unexpected draw from a disputed probability rule.
Show the odds before the selection
Convert each weight to a percentage and show the reason for the ratio before anyone sees the selected label.
Questions about probability weights
Must weights add to 100?
No. Only their ratios matter, so 6, 3, 1 already represents 60%, 30%, and 10%.
Can a weight be negative?
No. Negative values do not describe a selectable probability interval.
Choose another odds rule
All list tools · Remove-After-Pick Selector · Pair Generator