Why the downmods? Everyone can easily verify that there are similar biases if you replace "isPrime(n)" with "random() < 0.1" in the various code snippets floating in the thread. The article even admits that the biases are explained by the prime k-tuples conjecture, which is a model of randomness in primes from 1923. So primes are not less random than we thought -- they are still exactly as random as we thought.