From George Barnard
May I add a historical footnote to Dana Mackenzie’s brilliant article on
understanding randomness
(6 November, p 44)?
In the early 1960s, Andrei Kolmogorov gave a lecture in Belgrade in which he
said that he had thought it impossible to define randomness. But he had recently
learnt from a young Swedish mathematician named Per Martin-Löf that one
could say that the binary digit sequence specifying a real number x
could be defined as random if the shortest binary sequence specifying the
program for computing x was longer than x itself.
That it was Kolmogorov, not Martin-Löf, who went on to found the theory
of computational complexity may perhaps reflect the latter’s enthusiasm for
birdwatching.
Colchester, Essex
