“Anyone who considers arithmetical methods of producing random digits is, of course, in a state of sin,” [1] is a quote from the famous mathematician John von Neumann. In this work, we compare the most widely used arithmetical methods for randomness generation in terms of generation rate and seeding time. As a novelty of this work, we also consider the energy consumption, both for randomness generation as well as for seeding. We also argue that a “good” randomness generator is overrated for random linear network coding applications. Based on our measurements, we propose a faster method for generating randomness for RLNC with the same or better performance than traditional generators. In some scenarios, we reduce the linear dependencies by 40%.