Is 1010101010101010 the result of tossing a coin and writing 1 if it comes a head and 0 if it comes a tail? And what about this one: 1001101111001101? One puts in doubt the first one but trusts the second. Although both strings are equally probable (because both have the same length), one feels that the second is more random than the first one. The theory of algorithmic randomness gives several precise mathematical definitions of what is a random sequence, using tools from computability theory. But what about our own perception of randomness?