Thread: random isn't really random?

Originally Posted by Daved

You say seeding srand with MT.out and then calling rand is MAYBE better than just using MT.out. But you haven't proven that it is not worse either.

I actually don't know the internals of pseudo-random number generators myself, but logic tells me that using them the way they were intended to be used is most likely the best choice, especially when others more knowledgeable on the subject than I are saying the same thing.

Your logic is that it might be better because it might make the output more random. I'm still curious as to what exactly you mean by that.

to be fair, i have no solid reasons for believing that a coupling of PRNGs will yield better results. its just a half-hearted assumption that prng * prng = better prng.

i also agree that I haven't proven that multiple seeding of srand() with MT.out is not worse. it just worked well for me, thats all.

what i was intrigued about was why there are seemingly vehement claims that multiple srand() calls would be detrimental. iMalc pointed out a possible reason and Salem gave an example for the same, but that applies for an implementation slightly different than mine (please see my last post).

EDIT: i am aware that you had previously questioned what i meant by the phrase "more random". i had addressed that here.