The pseudorandom number generator (PRNG) is a math function. There are
many types. One type is the "Center-Square" method. Start with a seed,
that is, a binary number (typically 64 bits), square it (multiply it by
itself). This results in a 128-bit number. Take the center bits (i.e.,
bits 32 through 95), shift them down, and use that as your random number
and as the seed for the next random number. That's not a very good method.
A better method is "Multiplicative-Congruential." Start with a seed,
multiply it by a constant that will usually cause overflows. Ignore
overflows, divide that by another constant which does not have a common
divisor with the first constant, and use the remainder as the new number
and as the seed for the next iteration. I believe that the method used
in most gambling machines.
That is also the method used in Optimum Video Poker, but the random
numbers are NOT used to select cards. Instead, they are used to shuffle
the deck many times before each hand is dealt. I defy anyone to detect a
pattern in the dealt hands.
So what does it matter whether a number is random or pseudorandom? Many
years ago a new casino opened in Canada (I think it was in Windsor). By
law, it had to close for several hours every night. The keno game used a
PRNG. Everything was fine until one very astute player got a 20 out of
20 win, then a 19 out of 20 win, before they really dug in to find out
how he did it. At first he said he used chaos theory. Then he admitted
that he had noticed that the games were identical every day. It turned
out that the keno computer was shut down when the casino closed, and
every morning the PRNG started with the same seed, no naturally the
results were the same every day. All they would have had to do was use
some seed that wasn't a constant. Even the date would have been OK.
So for all practical purposes, the difference is in how the "random"
numbers are used.
Posted by: Dan <dan@optimumplay.com>
Reply via web post | • | Reply to sender | • | Reply to group | • | Start a New Topic | • | Messages in this topic (11) |