|Posted: September 15, 2005, 5:59 am - IP Logged|
With the latest generation of RNGs, seed size isn't the issue it once was. I doubt lotteries employing computerised draws are using the older algorithms, such as simple linear congruence generators. One example of an LGC is:
xn+1 = (a xn + c) mod m
where a = 16807, c = 0, m = 231 - 1, and the seed x0 = 667790
This is a 32-bit generator; it is quite sensitive to the value x0; and its period is small enough to be exhausted in a few seconds on modern computers (including fast PCs). Obviously you wouldn't want to use it for serious applications - and a state or national lottery can be considered as such, given the amount of money involved.
One of the best ways to choose the seed the value is to tap into the computer's internal clock, preferably at the nanosecond level: there's no earthly way anyone can know what the seed is.
If you're planning on predicting a lottery based on what RNG is used, don't bother. Whilst it's true that any given seed will always generate the same sequence of numbers, you can never know beforehand what that sequence will be the first time you try said seed.
As if all that isn't bad enough, current work on hardware-based RNGs considers such sources as thermal noise and quantum-level behaviour to generate numbers!
"God is subtle but He is not malicious." - Albert Einstein
"In my experience there's no such thing as luck." - Alec Guinness, Star Wars Episode IV