A Super Mega Bowl guide
When this site flips a coin or spins a wheel, no physical object is involved. A computer produces a number that behaves like chance. But computers are built to be perfectly predictable, so how do they generate randomness at all? The answer is cleverer, and more limited, than most people think.
A computer follows exact instructions and always does the same thing given the same input. That is the whole point of a computer, and it is also why a computer cannot invent true randomness on its own. Everything it produces is the result of a definite calculation. So instead of creating real randomness, most software fakes it convincingly with a pseudo-random number generator, or PRNG.
A PRNG starts with a number called the seed and runs it through a formula to produce the next number, then feeds that back in to produce the next, and so on. The stream that comes out is spread evenly and passes statistical tests for randomness, so for any practical purpose it looks random. But it is fully determined by the seed. Start two PRNGs from the same seed and they produce the exact same sequence forever. That is why it is called pseudo-random: random-looking, but not truly unpredictable.
This is what a browser's built-in generator does, and it is what powers the coin, dice, and spinner here. For fair play the results are uniform and unpredictable in practice, which is all a fair randomizer needs.
The fact that a seed reproduces the same sequence is often exactly what you want. In testing, a fixed seed lets you replay the same run and debug it. In simulations, it lets other researchers reproduce your results precisely. Reproducible randomness sounds like a contradiction, but it is one of the most useful tools in computing.
When random-looking is not enough, computers reach outside themselves for real unpredictability. A true random number generator draws on physical noise: electronic thermal noise, the timing of radioactive decay, atmospheric static, or tiny variations in hardware. Because these sources are believed to be genuinely unpredictable, the numbers they produce are truly random, not just statistically random. They are slower to gather, so they are usually used to seed other generators rather than to produce every value.
Security is the one place where an ordinary PRNG is dangerous. If an attacker can figure out the seed or the formula, they can predict every future value, which would be catastrophic for passwords, encryption keys, or shuffles in online gambling. For these, systems use a cryptographically secure generator, or CSPRNG, designed so that even seeing many past outputs gives no useful clue about the next one, and seeded from true physical randomness. This is the difference behind the common question of whether a randomizer is cryptographically secure. A casual coin flipper is uniform and fair but not secure, and it is not meant to be.
Computers cannot make true randomness alone, so they use pseudo-random generators: a seed plus a formula that produces random-looking but fully determined sequences. That is perfect for games, fairness, and simulations, and the reproducible seed is a bonus. True randomness comes from physical noise, and security demands a special cryptographic generator. Match the tool to the job and pseudo-randomness covers almost everything.