Random pseudo generator

How it works

A pseudo-random number generator, also written pseudo random number generator or pseudo-random number generator, is a function. You give it a seed — a starting value — and it produces a long sequence of numbers that are hard to guess by eye. Same seed, same sequence. Different seed, different sequence. That is the contract. Pseudo random number generators of this kind power shuffled playlists, game loot tables, simulations, and most “random” buttons on ordinary web pages.

The sequence is deterministic. If you know the algorithm and the seed, you can replay it. That is why a bug report can recreate a shuffle, and why a classroom can reuse tomorrow’s warm-up. It is also why this class of tool is the wrong place to hide a secret. A password, a token, or a cryptographic key needs a generator designed so that seeing some outputs does not let you predict the rest. A toy linear formula is not that generator, even if the numbers “look random” on a poster.

Period matters. A generator with a short period will eventually loop. For a handful of dice rolls you will not notice. For a long simulation you might. Modern general-purpose generators used in common languages have very large periods compared with a classroom draw. That still does not make them cryptographic. Large period and unpredictability to an attacker are different properties. Ask for the property you needed.

Seeds, ranges, and leftovers

If you do not set a seed, the program usually sets one from the clock or from the operating system. Two people clicking at different times get different streams. Two tests that forgot to seed can disagree for that reason alone, not because the function is “broken.” If you need agreement, set the seed and write it down. If you need disagreement, do not reuse yesterday’s seed and then call the collision fate.

Turning a raw stream into an integer from 1 to N has to respect N. A remainder that maps a range whose size is not a multiple of N can lean on some faces of the die. Libraries often ship a uniform integer helper for this reason. Use it when the library has one. Inclusive bounds are still English: 1 through 6 includes 6. A half-open programmer interval that drops the top value is a different die.

Seeded stream: replayable, good for tests and shared puzzles. Unseeded stream: different each run, good for a party picker. Uniform integer helper: prefer it over a homemade remainder when you care about equal faces.

Cryptographic generator: a different tool, for secrets, not for this page’s party die. If you can replay the shuffle from a seed, you did not hide a secret. You hid a recipe, and recipes leak.

What a pseudo stream is good for

It is good for games, Monte Carlo work, procedural maps, and any UI that needs a mixed order without a physical noise source. It is good when two machines must share a sequence: send the seed, not a million values. It is good when you want to freeze a lesson. It is a weak raffle record if the seed was the current millisecond and nobody wrote it down; then you have a result without a recipe, which is the opposite of what this tool is for.

Independent calls can repeat. That is the stream behaving. Sampling without replacement is a shuffle of a finite list, then a take. Do not accuse the generator of bias because 7 appeared twice in a 1-to-10 stream of length five. Do accuse the setup if 11 appeared, or if 10 never can. Write the algorithm name, the seed if you used one, the range, and the count. “We used a random pseudo generator” is not enough for the next person to replay or to audit.

Do not scrape a public noise service and then call the result pseudo, or the reverse. Names describe the source. An algorithm is an algorithm. A radio is a radio. If the page cannot tell you which one it is, you do not have a generator description. You have a button.