Randomness Guide

How random selection works

A fair random picker needs two things: a sound source of randomness and a method that maps that randomness to list positions without favoring some entries. WordFinderHub handles both parts in the browser.

The list becomes numbered positions

If your list contains five entries, the picker treats them as five positions. A fair one-choice draw should give each position the same probability when every item appears once.

The winner is chosen before the animation

WordFinderHub uses the browser's crypto.getRandomValues() source to obtain random values. The winning list index is determined before the moving-block animation starts. The animation is a visual reveal only.

Why ordinary modulo can introduce bias

A common shortcut is to take a random integer and use the remainder after division by the number of entries. That can be slightly biased when the full range of random integers is not evenly divisible by the list size.

Rejection sampling

WordFinderHub avoids that problem with rejection sampling. Random values outside an evenly divisible range are discarded and another value is generated. Only values from the clean range are mapped to a list position.

Duplicates are separate entries

Randomness can be technically fair while the list itself is weighted. If one name appears twice, it occupies two positions and has twice the chance of a name that appears once. That may be intentional or accidental, so check the list before drawing.

Repeated draws

When Remove selected item is off, each draw uses the current list without automatically removing a winner, so the same entry can be selected again. When it is on, the chosen line is removed before the next draw, which is useful when you want unique winners or a full random order.

Device speed does not change the result

Because selection happens first, a fast phone, slow laptop, different frame rate or reduced-motion setting cannot change the winner. Those factors affect only how the reveal looks.

What this does not guarantee

WordFinderHub is suitable for ordinary choices such as classroom order, team rotation and low-stakes group decisions. It is not an audited lottery system or a certified source for regulated draws, cryptographic key generation or other high-stakes uses.

Try Random Picker

A small model of modulo bias

Imagine a source that returns each integer from 0 through 7 equally often, and a list with three positions. Taking the remainder after division by 3 maps 0, 3 and 6 to position 0; 1, 4 and 7 to position 1; and only 2 and 5 to position 2. The first two positions have three chances each, but the third has two.

Discarding 6 and 7 fixes this model: the remaining values 0 through 5 divide evenly among three positions. If a discarded value occurs, draw again. WordFinderHub uses the same rejection principle with the much larger range of an unsigned 32-bit value.

Separate list rules from random generation

Removing a winner changes the eligible positions for the next draw. Repeating a name creates more than one eligible position for that name. Neither operation changes the need to map random values evenly; they change what each position represents.

The visible grid can show repeated labels during the reveal. Those appearances are decoration, not additional chances or a record of how many times an item was considered. Record the final selected text. See how to run and record a sequence of draws.

For the browser API used by this implementation, see MDN’s getRandomValues documentation.