This is quirky, but it's funny to think about trusting randomness. It's belief that nobody else has more information about what you are being presented than what you are being presented. I know we have pretty good CRNGs and there's nothing rigorous about this deepity, but having seen how they have been sabotaged in the past, when you start seriously questioning whether you can really trust random numbers, it becomes a very funny existential question. If you can't trust randomness, there's only one thing left.
randomness is just the absence of human ability to understand the pattern. Thats all. It doesn't mean it is truly unpredictable, just unpredictable by what we know right now.
I shared this earlier, in a comment on special relativity, debating determinism vs indeterminism. I thought it deserved it's own post. I am surprised this has not had any love on HN already. A truly (maybe! ;) random RNG, powered by measuring the quantum fluctuations of the vacuum? And free, too!
There is no way to verify that somebody else is delivering truly random data, or that they aren't spooling off a copy of whatever they hand out.
But there is no need ever to get your entropy from somebody else. The low bits of pixels in every camera in your gadgets provide a constant supply of high-quality entropy nobody else has access to, either to face or to copy.
When gathering random bits this way, it is usual to recommend flipping every second bit, to eliminate bias. XORing it to a stream cipher bitstream is also a good idea.
But hardly anybody needs much entropy, except to stay warm.
random.org also offers 'true' randomness, derived from atmospheric noise.
I do wonder if there's any way to determine which of these is more random, and if that even means anything.
Last time I looked at this, you could quantify randomness, but not very well. The best way I found was to compress the random output as compression rate inversely correlates to randomness.
