quartz/content/notes/randomness.md

title	aliases	tags
randomness		comp210

• a lack of predicability
• no patterns
• stochastic (can be analysed but not predcted) vs deterministic

used for

• one time pad
• generating key-pairs
• generating salts for password hashing
• seeding pesudo random number generators

humans are not good at randomness

• e.g., see patterns when there are none
• e.g., random dot pattern
• shuffle algorithm
• gambers fallacy

how to prove randomness

• cannot prove
• but can check for uniformity, bias, distribution etc.
• difficult with small samples
• can identify data that is unlikely to be random
  • (rngtext, diehard, dieharder etc)

In computers

computers have the same problem: they are deterministic

• cannot really produce true random numbers
• CSPRNGS crypto secure pseudo random number generators
• hardware entropy generator

entropy pool

• modern OSs implement an entropy pool that processes can draw on when the need random data
  • some systems provide blocking and non-blocking random source devices
    • blocking: will stop when entropy is exhausted