Published at LXer:

How to generate a random number from the Linux command line. We will explore how to generate a random number within a range and also a specific length.

Read More...

Thank you.
I was reading the linked article and was about to suggest that you do the same when I got to the end and realised that, nowhere, had they mentioned the true randomness of the numbers. Your method could be used with a true random number generator or, at least, something like the sound-based one to create the least-predictable numbers your system is capable of.

Right on!

If you get into a situation where you need 'more' randomness, you can hit up random.org via http or api. They use atmospheric wash to create random readings.

random.org

For more information.

I'll be honest in that my interest in random, aside from the odd SSH session, is just curiosity. It did occur to me just now, however, that I've a couple of USB SDR devices which ought to be able to find something actually random?
I seem to recall the sound-based one is randomsound?

Sure. But sound isn't as random as might be expected. After a while you will get repetitions in your singular environment. Ie: you sneeze every 4m, heater kicks on 3 times an hour, etc. There are recurring sounds and reasons that your random feed would be psuedo-random or not-random.

I made a random generator years ago that:

Grabbed the news headline of the moment
Converted all the letters to numbers
Grabbed a random subsection of weather data
Converted all the entries to numbers
Did some math on the results and converted to base64

That got me a lot of good random data to mess around with at the time. Not sure I'd do it the same now.

I admit I don't know how randomsound works, but I suspect the developers weren't that stupid? Perhaps?
To be fair though, up until now, only stupid "random" systems seem to have been broken.

We are all only human. Remember when Debian broke OpenSSL randomness which no one noticed for years? https://www.schneier.com/blog/archiv..._number_b.html

Have to admit I dobn't recall that exactly but have read of similar over the years.
Perhaps there's a "Schneier's Law" stating that even the best crypto syystem will fall to implementation eventually?