I’m beginning to feel sorry I raised this query. I had no intention of starting (or re-opening) an argument over randomness in the digital world. In my module design, I experienced a set of results from the Symbolic Dice Button facility that seemed a little at variance with what I expected. I merely wanted to check whether other users had reported similar experiences and if my experience was the result of a glitch or simply a demonstration of a perfectly good facility not turning up with a series of results that fitted my expectations.
Joel - please accept my apologies if I’ve unwittingly opened an old wound.
I’m happy to answer questions about it. Being told that we have a problem on no evidence—which you did not do, but happens every few weeks—is a sore point for me.
In fairness, the topic of “Are psuedorandom number generators truly random” is an incredibly deep topic, which IMHO requires PHD level math…
If interested in the topic, it ties closely to the history of cryptography, which is a fascinating topic to read about (e.g., Bletchley Park in WW2).
The question decomposes into a few subtopics:
Is the PRND generator cryptographically secure? (i.e., sequence is not predictable)
Is the distribution of generated numbers correct?
For those people who prefer a more experimental approach, it would be straightforward enough to put together a small module, and generate 1E5 to 1E7 random numbers and plot the results.
Meeting FIPS 140-2, Security Requirements for Cryptographic Modules is a pretty high bar, IMHO…
