magicforfun wrote:I have to disagree. The randomness will follow a normal distribution, hence we can predict the outcome with a % deviation. That offsets 100% randomness. It's important to add that every throw off your buttered toast is independent from the result of the former throw. But if you do 1 gazillion throws, you can predict the outcome of these, e.g 52% butter down, 48% butter up or whatever.
Trust me, I did my maths degrees ages ago, but I did'em.
I would disagree with that! The fact that a result over a period of time is predictable within bounds doesn't make it non-random; the radiation example shows this well.
If you took a single atom of an unstable isotope and waited for it to decay you could wait the entire age of the universe and it would not happen. Then again, it might happen a nanosecond from now; entirely unpredictable and random.
However, if you took a larger quantity of that isotope, Uranium-238 for example, it is well known that half of the atoms would decay sometime in the next 4.5 billion years, or Iodine-125; half the atoms would decay in the next 8.1 days. Fairly predictable bulk outcomes of an entirely random process.
So, a random number generator could be made simply by measuring, to arbitrary accuracy, the time between consecutive decays in a radioactive sample. It would give you a fairly predictable number of responses in a given time and each one would be totally random.
