On Wed, 29 Dec 1999, Papa Legba wrote:

>
> On a second but maybe related note, I believe the random number genrator
> may be flawed in the game. I see too many long runs of unlikly events
> (both good and bad). This would be far from unheard of, as random
> number generators are eseentially impossible to code. (People just
> write code that approximates random number generartion). I did a study
> of the random number generator built into the 8088 (basically the
> original IBM PC) processor. (I used a few different languages to try to
> eliminate any effect they had on it). I basically graphed a "random" x
> and y cooordinate, drawing a line from point 1 to 2 to 3, etc... The
> plots always followed similar patterns... Weird, eh?

Really good random number generators are notoriously difficult
to write. However, you have to keep in mind that unless
they're saving seeds for every event sequence (e.g. whether
your weapon procs, which mob spawns at a site), your
"sequence" of random number events is not a really a sequence.
It's pieces of the actual sequence, the other random numbers
having gone to determining if the mob hit your tank, whether
your caster friend's spell fizzled, whether the guy meditating
at the other end of the zone "passed" his skill check, etc.

The order of these events are pretty much random, depending on
player input, Internet lag, etc. So even if the random
numbers are only quasi-random, *which* quasi-random number you
get *is* random. So I wouldn't be so quick to blame the RNG
just because you get a few unusual strings of sequences.

--
John H. Kim
kim@...