Code-breaking, at its heart, involves looking at lots of numbers and trying to spot telltale patterns that may provide a key to decryption.
Spies long ago realised that the best way to create truly random encryption keys is to ask a computer to do it. This is because human beings are rather poor at creating lists of random numbers, even (or perhaps especially) when they are consciously trying to do so.
Looking for an insight into why randomness – or apparent randomness – is so challenging, a team of researchers from the Laboratoire de Recherche Scientifique in Paris, France, devised a series of tests and applied them to a pool of 3400 people, aged between four and 91.
Each of the participants was asked to complete a series of tasks, all of which required the construction of lists that would appear random when viewed by an observer. The tasks included making up a set of 12 coin toss results, and inventing the numbers shown by 10 hypothetical die throws.
The results, published in PLOS Computational Biology, reveal that age 25 is the peak for achieving truly random-looking lists.
“This experiment is a kind of reverse Turing test for random behavior, a test of strength between algorithms and humans,” says co-author Hector Zenil.
The findings add weight to previous studies that suggest – perhaps counter-intuitively – that constructing properly random lists requires very high-level cognitive processes. Randomness may be connected to other cognitive activity, particularly creativity.
Just how randomness develops and declines over a person’s lifetime remains poorly understood, but the work of Zenil and colleagues now provides others in the field with an assessment tool to accurately measure its expression.
The team also demonstrated that this measurement can be achieved using only a small data set of 10 or 12 points, rather than the much larger sets employed in previous research.