How could random functions be really random?

Introduction

I know I'm going to lose a lot of reputation for this question and I also know it will be flagged as inappropriate but I'm really curious about that so I'm not giving up if there's any chance I'm getting at least an answer.

Question

Today I woke up thinking:

Hei, how could random functions be really random if they are 开发者_Go百科created by an algorithm?

Think about it. How could you create a function that simulates randomness without the concept of random already built in? I began to think:

Hei, I'd take an array of int, then I'd do [thing], then [thing], than [thing] again, then I'd choose only odd numbers... ecc

But it seems more likely a function that make it more confusing to predict what the choose will be rather than real randomness.

Is it possible to create randomness? How are functions that returns random ints (such as rand() in PHP) created? How can they simulate randomness?

---

Functions that algorithmically produce so-called random numbers are pseudorandom number generators. If you know the seed used to generate the sequence, then the numbers are predictable. The sequence itself is a statistically random distribution but not truly random.

There are true random number generators that typically involve some hardware that samples randomness from the physical world, e.g., radioactivity or acoustic noise. A naive implementation would be to sample hard disk access and mouse movements. See random.org for a real RNG.

Obligatory xkcd strip:

---

There's a reason they're called pseudorandom numbers; they're not truly random. From Wikipedia:

A pseudorandom number generator (PRNG), also known as a deterministic random bit generator (DRBG),[1] is an algorithm for generating a sequence of numbers that approximates the properties of random numbers. The sequence is not truly random in that it is completely determined by a relatively small set of initial values, called the PRNG's state.

---

Read volume 2, chapter 3 of this seminal work if you want the maths behind it. You can buy it to look impressive on your bookshelf. (Just keep in mind that most people who buy it wind up never actually reading it -- for a good reason. It's VERY dense and VERY difficult reading.) The short answer that doesn't involve massive tomes of difficult text is that "random" numbers generated purely algorithmically are pseudorandom, which is to say that they are "random enough".

---

You might want to look into wikipedia's article on PRNGS - what all random number generators we have on PCs (pretty much) are.

About the closest you can get to random, which I think is done somewhere, is to use temperatures in the CPU or some other sensor reading as a seed for one of these. If the seed is random (the temperature is unlikely to ever be exactly the same), the sequence is about as close to random as possible.

---

I usually "get Milliseconds" and divide it to a pseudorandom number. This makes it even more random and unpredictable.