Yes, I also thought that pointing this out wouldn't be helpful to the OP. There are also "better" pseudo random algorithms than the standard rand() function - algorithms that give a more even distribution of numbers and do better on other statistical measures at the expense of more computational effort. I'm sure none of this matters to the OP. Basically given the hints provided, if the OP couldn't come close to my one Perl line, the chance of any significant computation using the results is close to zero.
To get really random numbers, you need a piece of hardware that quantifies some physical random phenomenon. I haven't checked in on the Princeton EGG Project in awhile. I suspect that their hardware is pretty good.