7.1 Random Number Generators

A random number generator is a subroutine you can call. For most practical purposes, you needn't worry about what's inside it. The values you get for random numbers on the computer differ somewhat from the values of real-world random events as measured, for example, by detecting nuclear decay events. Some computers actually have devices such as geiger counters attached so as to have a source of truly random events. But I'd be willing to bet your computer doesn't. What you have in place of a geiger counter, is an algorithm called a random number generator.

The numbers that are output by random number generators are not really random; they are thus called pseudo-random numbers. A random number generator, being an algorithm, is predictable. A random number generator needs a seed, an input you can change to get a different series of (pseudo-)random numbers.

The numbers from a random number generator give an even distribution of values. This is one of the most important characteristics of randomness and largely justifies the use of these algorithms where some amount of random behavior is desired.

The other "take-home message" about random number generators is that the seed you start them up with should itself be selected randomly. If you seed with the same number every time, you'll get the same sequence of "random numbers" every time as well. (Not very random!) Try to pick a seed that has some randomness in it, such as a number calculated from some computer event that changes haphazardly over time.^[1]

^[1] Even here, for critical applications, you're not out of the woods. Unless you pick your seeds carefully, hackers will figure out how you're picking them and crack your random numbers and therefore your passwords. The method used to generate seeds in this chapter, time|$$, is crackable by dedicated hackers. A better choice is time() ^ ($$+<<15)). If program security is important, you should consult the Perl documentation, and the Math::Random and Math::TrulyRandom modules from CPAN

In the examples that follow, I use a simple method for seed picking that's okay for most purposes. If you use random numbers for data encryption with critical privacy issues (such as patient records), you should read further into the Perl documentation about the several advanced options Perl provides for random number generation. In this book, I use a Perl method that is good enough for most purposes.