/randoming before lotting

[Hanyoko]

Ok possibly bad example since dynamis is handled different, been a few years but i do recall similar situations. Replace exmaple with 64 people at some sort of poped NM event that has adds, and after death swaping in people to lot/pass multiple times for no reason.

And attaching a lot in a item <-> player connection is very different (and just 1 row in a database) rather then saying every lot is pregenerated.

*data*
Just ran a C# RNG for 100,000,000 iterations

average 100,000
high 101,070 (209)
low 99,095 (147)

[Spekkio]

just consider the extra strain of 5 crystals and 6 items dropping from a single mob that's being xped on in a burn party. calculate the wasted cycles per hour and figure what that would cost SE for the inefficiency over the course of the life of the FFXI servers. why would you needlessly waste a ton of cycles on unnecessary actions? besides, what difference could it possibly make to know when it decides the random unless you know a way to control it.

[The Blackrose]

ITT people with no programming skill speculate on how random numbers must be generated.


here's a hint; does Monopoly come with a sheet of pre-rolled numbers? no?

neither does XI.

As someone who has argued along the same lines of logic before, it's usually best to just let these topics pass by. Or try to enjoy them with some popcorn goodness.

[Necronus]

a hardware RNG hashed into a 0-999 range is hardly complicated and would completely wreck both of your theories.


that said, they could also do a simple rand with unique seed based on number of packet requests since server up, which would also be almost impossible to break in a predictable manner (assuming the rand is any good).

and considering you can get (for free) an algorithm, here, that has a cycle time on the order of magnitude of 10 quattuordecillion (to give you an idea of the magnitude we're talking about here, if we were to generate 30,000,000,000(30b) random numbers a second, it would take 3 trillion centuries before a single seed would cycle to a repeating pattern.

obviously, a software RNG is only as good as its seeding system. but with a software RNG like that, you would only need to seed the RNG during boot up and you'd have good variance.

Like I said, I wasn't defending any theories. Just pointing out that the Gambler's Fallacy doesn't have to apply to computing if you have someone lazy or stupid enough to program something poorly (or someone intentionally going out of their way to do so I guess). Now, while I normally look down on many poor programming decisions SE has made in the past, the fact random number generation can be pretty trivially added to their software, I doubt any of my ramblings in my first post are true, aside from the fact that the Fallacy is not necessarily true, just highly likely.