Random numbers in a range: batches, repeats, sorting

How the generator works

Set the bounds — for example, from 1 to 100 — and the number of values. The range is inclusive: both the minimum and the maximum can come up. For a draw, enable "no repeats": every number then appears once, as on a lottery ticket. Sorting does not affect randomness, it only orders the finished set.

The sum and the mean are shown below the result — a convenient way to check that the batch looks uniform. For five numbers from 1 to 100 the mean is usually close to 50, but spread in a short run is a normal thing.

Where it comes in handy

Lotteries and draws, picking a speaker, test data for tables, a "roll of two numbers from 1 to 6" instead of a die. If you need access keys or passwords, use a dedicated password generator instead: it has a different character set and entropy estimate.

Why not Math.random

Math.random() produces a pseudorandom sequence: given a few values, the sequence can be reconstructed and the next one predicted. For an everyday task that is a minor detail, for a prize draw it is not. That is why the generator takes bytes from crypto.getRandomValues, and rejection of values removes the bias toward the start of the range.

"No repeats" is limited by the size of the range: you cannot pick twelve unique numbers out of ten — the generator will say so plainly.