True random numbers in R

Random numbers are important for lots of things including statistical analysis (e.g., using bootstrapping) and cryptography (e.g., producing a one-time pad). But computers can’t produce random numbers. They produce pseudorandom numbers, which are deterministic and hence predictable given the random seed.

Pseudorandom numbers more than suffice for most purposes (so long as everyone isn’t using the same seeds like 1234 or 42); however, if you want or need the real thing there are various ways to pipe true randomness into computers. Two are available via R packages.

The best known is {random}, which exports pure randomness from Ireland. The basic idea is that a radio receiver is tuned to a frequency where no station broadcasts. This is continuously recorded and sampled to generate random numbers.

Another is {qrandom}, provided by the Australian National University. This works by measuring the quantum fluctuations of the vacuum.

01001110 01101111 00101100 00100000 01110100 01101000 01101001 01110011 00100000 01101001 01110011 01101110 11100010 10000000 10011001 01110100 00100000 01110010 01100001 01101110 01100100 01101111 01101101