The tests I have aren't random because then its difficult to meaningfully communicate the outcome. Instead it each NIP had a test suite, and then each test in the suite attempts to obtain samples from relays and then uses the uses those samples to construct filters than satisfy the respective test.
nostr:nprofile1qqswwud0pvzu362lehm0av6sq4zd97cue5uy0z8f7jgtk0hz368dvmcprdmhxue69uhhw6r9v96zu6rpwpc8jarpwejhym3wvdhj7qgawaehxw309ak82mnrdp3x77pwwdskuerhd93kstnxv9ex6tcfwvfpm has his thing that does something conceptually similar
Discussion
I picked the word “conceptually” very carefully 😉
They are conceptually much different 😂
Both useful on different ways. Randomness is great in fuzz testing and finding edge cases that are difficult to find without randomness, the auditor I wrote is better for unit testing known vectors to validate expected behaviors.
And to be clear, they are complimentary and so is relevant. Just wanted to clarify the functionality.