Entropy

In information theory, entropy is a measure of the amount of information a message carries, or equivalently the unpredictability of its content. A message that is easy to guess in advance carries little entropy, while one that is genuinely unpredictable carries a lot.

In cryptography, entropy is what makes a key or a random value hard to guess. The entropy of a cryptographic system is bounded by the size of its keyspace. A larger keyspace gives more possible keys and, provided the algorithm itself is sound, more entropy and a harder system to break by brute force. It is not enough for cryptographic input to merely look random. It must also carry high entropy, since a predictable process for generating "random" values undermines the security the randomness was meant to provide.

Gathering genuinely high-entropy randomness from a computer system is non-trivial, and can be expensive enough to affect performance. This is why pseudo-randomness, generating values deterministically from a smaller seed, is preferred wherever true, high-entropy randomness isn’t strictly required.