Understanding the RNG from a mathematical perspective

Some people dislike variance.

For example rather than using a fusing orb which say gives a 1:500 chance (made up number) of getting a 6L, they would prefer to just say give a vendor 500 fusing orbs and get a 6L for sure.

It's pretty much the exact same system, except one has variance and one doesn't. Variance adds excitement and mystery, but doesn't change the math (long term over a large sample that is).

"

Zaqwert wrote:

Some people dislike variance.

For example rather than using a fusing orb which say gives a 1:500 chance (made up number) of getting a 6L, they would prefer to just say give a vendor 500 fusing orbs and get a 6L for sure.

It's pretty much the exact same system, except one has variance and one doesn't. Variance adds excitement and mystery, but doesn't change the math (long term over a large sample that is).

From the long term, large sample view, i.e. the view of a GGG administrator checking the crafting tables of last month, there would be not much difference in those two systems, that´s right. But there is a big difference for the single person. The same difference as there is between faith and knowlegde.

RNG is much like roulette, you may toss some hundred bucks on a single number, just because this number didn´t show up in the last 1000 spins. But still, your chance to win in this special spin is 1:37 (1:38 for american roulette), no matter if there were 1000 spins without that number.

The proposed system is like saving to buy a house. It will take a while, but you know for sure you will have a house in the end.

..but you do get the house in the end..


..if you buy it from the guy selling the house. :D

Informative post, hopefully people give it a read. You might want to touch on RNG in computing as well whereby RNG is not true RNG.

"

Nyruami wrote:

The proposed system is like saving to buy a house. It will take a while, but you know for sure you will have a house in the end.

No. It. Is. Not.

Your example is so far away from the RNG formula, that you went passed the asylum.

"

IExiledOtherExilesToBeTheExile wrote:

No. It. Is. Not.

Your example is so far away from the RNG formula, that you went passed the asylum.

I believe you misread the lines you quoted. "Proposed system" doesn't refer to RNG, but rather to the system mentioned whereby a person pays 500 fusing orbs to a vendor and receives a 6L.

"

So after 10,000 kills your expected number of drops is 1.

I am not following this.

Statistically speaking, if the chance of getting a chaos is 1:10000, then the chance of not getting a chaos orb is .9999. If you raise .9999 ^^ 10000, you get .367. So if you roll 10000 times, you have a .36 (36%) chance, or roughly 1:3, of not getting any chaos orb at all. So I would say the the expected number of drops is .63, myself, not 1.

This can be one of the biggest misconceptions of the RNG, and one of the greatest causes of grief. This is why I suggested what I did in the RNG-itis thread.

While this bad outcome was a full 37% likely, I would recommend intervening against the RNG for the 5-15% (not picky about the intervention point, just listing by example).

"

Courageous wrote:

"

So after 10,000 kills your expected number of drops is 1.

I am not following this.

Statistically speaking, if the chance of getting a chaos is 1:10000, then the chance of not getting a chaos orb is .9999. If you raise .9999 ^^ 10000, you get .367. So if you roll 10000 times, you have a .36 (36%) chance, or roughly 1:3, of not getting any chaos orb at all. So I would say the the expected number of drops is .63, myself, not 1.

What your describing (0.63) is the probability to get at least 1 drop in a session of 10000 kills.

The "expected number of drops" or more precisely "expectation value of the number of drops" is exactly 1 if you have a 1:10000 chance and 10000 kills. If you'd have many sessions of 10000 kills and average out the number of chaos orbs per session, you'd get 1. Sometimes you get 0 (37% of the times on average), most of the time you get 1, sometimes you get more than 1. But it averages to 1, when the number of sessions is sufficiently large.

You have to take into account the algorithm that generates the "random" numbers. It's not really random, per se. Most often, random number generators are seeded with a value ( usually time, but could be something like an analog to digital conversion of signal noise, or something else ). You aren't going to have a perfectly uniform distribution. If I use, for example, C++ and feed the STL rand() function with the same seed, I'll end up with the same sequence of "random" numbers.

EDIT: Read this as "random" numbers are deterministic, not actually random.

"

realatomicpanda wrote:

I believe you misread the lines you quoted. "Proposed system" doesn't refer to RNG, but rather to the system mentioned whereby a person pays 500 fusing orbs to a vendor and receives a 6L.

That is still RNG. Don't try to change the subject. There is a huge difference between working with numbers you can't see, and numbers you can see [money in your hands upon buying the house.]

Philosophical question ( read: derailment ): Is there anything that is truly "random"?