>>12182997Not necessarily "true" random numbers, but rather random numbers which are impossible to accurately predict. In cryptography we distingsuish between cryptographically secure and insecure randomness. Cryptographically insecure randomness usually opreate on seeds, and having a specific seed allows someone to predict the entire sequence of random numbers than an random-number generator will generate.
One of the many Cryptographic use-cases would be initialization vectors (IV) for block ciphers. These are used to add an extra layer of scramble to the cipher, and is useful in preventing patterns from emerging in longer streams of data. A classic comparison would be ECB-mode encryption with CBC-mode encryption. EBC-mode encrypts each block individually without an initialization vector, and thus whenever the same block is encrypted twice, the exact same block of ciphertext would come out. This is not good, as it allows an attacker to run frequency analysis on the cipher and can potentially decipher parts of the encrypted data. Compare this with CBC, where each block is XOR-ed with the last, and the first block is XOR-ed with an IV. The great advantage of this is that no pattern should theoretically emerge, assuming the IV is completely random. and inpredictable.
It's moreover important that the IV is always changed each time a cipher is used. Given the limited lifetime of an IV, it is important that one generates the IV with an algorithm that cannot reasonably be predicted. That is, the randomness is cryptographically secure. We accomplish this by attaining randomness from different sources like mouse movements and memory usage. With enough sources of randomness we get something very hard to predict. Had we not done this, an attacker could potentially predict our IV, and our attempt at obscuring our message would fail. This stuff is the reason you're able to log onto your bank over the internet without thugs takinging all your money by spoofing your internet
