Re `seed!` ing the RNG at each step of a simulation?

[stevengj]

@nhz2

I’m okay with it only being reproducible in a single Julia version

Then what’s the problem with serializing the RNG state?

@nhz2

but what do you mean by it being less random?

If you re-seed too often it changes the statistics. (Imagine the extreme case where you re-seed after each random number.)