Asked by: Hasan Cuthbert, Cambridge
That’s a question with practical importance, as randomness is surprisingly useful. Following no laws, random numbers lack any predictability, so when added to text they garble it in a way that no one can unscramble without knowing what numbers were added to encrypt them. Randomness is also useful in simulating the effects of chance on complex systems like stock markets, and for selecting representative samples of patients when testing new drugs.
Researchers typically use random numbers supplied by a computer, but these are generated by mathematical formulas – and so by definition cannot be truly random. In the 1970s, scientists discovered that a widely-used formula produced regularities in its ‘random’ numbers that undermined countless research studies. True randomness can be generated by exploiting the inherent uncertainty of the subatomic world. In 1957, the UK government unveiled ERNIE (Electronic Random Number Indicator Equipment), which used random quantum noise to choose Premium Bond numbers.
While randomness seems ideal for making totally unbiased choices, there’s a problem: the lack of bias only really appears in an infinitely long set of random numbers. In any given collection, there can be astonishingly long patterns. So don’t be surprised if your smartphone track-shuffling feature randomly plays the same song several times on the trot.