These days I found a
website providing service to generate random numbers. What I actually need was a randomized sequence generator, which I was about to use for my user study purpose. However, I was shocked by a statement of the website saying: "In reality, most random numbers used in computer programs are
pseudo-random, which means they are a generated in a predictable fashion using a mathematical formula. This is fine for many purposes, but it may not be random in the way you expect if you're used to dice rolls and lottery drawings."
It's interesting to think about what we are
expecting about our random numbers. Although we all know the computer programs generated random numbers are
pseudo-random, what kind of numbers are
true-random? How can we generate them and distinguish them from those pseudo ones?
What they offered is a
TRUE random number generator which shown below. They do that by recording and interpreting atmospheric noise with several radios.
However, how can we prove the generated numbers are truly random? It's a rather difficult question then you might have imagined, in fact, it's impossible to prove randomness definitively. A visual analysis approach can be found from that
website, which visualize the generated numbers into image and decide their randomness with human eyes. Besides, you can find some other methods from the same page.
Meanwhile, though most random number generator program are actually
pseudo-random, it's not always possible to predict the generated numbers. Most of these programs rely on 2 points: the generating formulae, and an initial number (
seed). In most of cases, the system time is used as seed. Thus, in order to predict the resulting numbers of a specific program whose algorithm is already known, we'll have to know the time when it generated them as well, which could be difficult to determine. In this way, pseudo-random number generator can obtain some true randomness by involving the randomness from the physical world, which is also exactly what the
true random number generators do.
But, since we can't prove a truly randomness, it's also impossible to say the numbers genererated in that way is truly random or not. After all, how can we prove the randomness of the world, while the world itself could be
deterministic? How can we say the atmospheric noise is definitely random?
We don't have to. What's the problem of those pseudo-random number generators? Simply because patterns can be spotted in the results, which made them predictable to some extent. Nevertheless, what we really need is
unpredictability, rather then true randomness. Regarding to the theory about
deterministic universe, it's even hard to say the true randomness ever exist at all.
Posts
