Has the Random Numbe Generator been tested?

[darkgod]

Because you are assuming uniform distribution

[HousePet]

If the level was not at least 5 then there might only have been 2 possible runes.

[Theyleon]

Most of the things people are bringing up are not problems with the random number generator, they are at worst problems with the way the random numbers are being used.

[Suslik]

Most of the things people are bringing up are not problems with the random number generator, they are at worst problems with the way the random numbers are being used.
Well that's pretty obvious. I just pointed out that distribution is expected to be near-uniform, but according to DG it isnt. Why? If the shop can generate phase door, manasurge and shield, why for the sake of goodness did he use RNG so it usually generates 9 phase doors out of 10?

[HousePet]

Because the chances aren't equal, as DarkGod is saying.

[Waladil]

I think the shop sort method makes it look a lot clumpier than it really is. If you were shopping and saw three Belts of Unlife, you might think you had a really clumpy generation. Three Belts of Unlife in a row? What are the odds? Except the code might well have rolled belts of unlife on the first, fourth, and eleventh roll out of twelve, and then simply sorted those three similar belts together. Looks a lot clumpier than it is. Plus, as has already been said, humans find patterns.

Anyone whose played a Chronomancer or equilibrium-heavy class has thought "Why did my spell fail? There was only a two percent chance!", conveniently forgetting the fact that the last two dozen spells went off without a hitch. Failing three times in a row at under 5% each time is really unlikely, but so is succeeding a thousand times in a row at 95%. You won't notice the constant successes, but you'll notice the failures. (Besides, I've rolled a triple-crit in D&D, with plenty of people watching my dice. Unlikely things happen, yo.)