That was me. RNG systems typically use entropy based seeding to produce random results and it pulls from multiple sources.
I explained my theory better in the other thread in regards to DB but this is some basic info on it - https://en.wikipedia.org/wiki/Entropy_(computing)
There is a chance that whatever system they are using to create random results isn’t producing random results due to a failure with certain types of hardware or some other flaw that would instead produce a consistent result.
An example of a flaw would be a system that is setup to pull a certain amount of entropy from a system to seed results but you don’t give the system enough time to generate enough entropy and instead of making the user wait it insteads repeats previous results. Or, the seeding data is somehow getting cached sometimes as opposed to going stale and replaced which would produce a consistent result.
I honestly have no clue what implementation they use so I’m pretty much just spitballing theories.