The World of Layonara  Forums

Author Topic: Fun with randomness  (Read 701 times)


Fun with randomness
« on: September 08, 2012, 01:02:14 pm »
(subtitle: How good are your dice?)

OneST8 passed this my way and I thought I'd share:

The Epic 20,000 Roll Dice Randomness Test - Forbes

If anyone's curious, I have a compilation of over 3.5 years worth of crafting die rolls and over 4.5 years worth of Save vs. Soul Mother rolls.  See the attached Excel file for the results

I have done this for curiosity purposes, not to brow-beat anyone or start up a debate, and decided to share it with all of you for the same reasons.

The file contains two worksheets, one for crafting results and one for Soul Mother saves.  Each contains the following:
  • The frequency of each die roll result
  • A handy chart for visualization purposes
  • Computed total number of rolls
  • Computed standard deviation
  • Computed average deviation
  • Maximum and minimum roll frequency values
  • Range of frequencies as a percentage of total rolls
  • Sum of values of all dice rolled (from database query)
  • Computed average roll
  • Computed ideal average roll
You may have to scroll down to see the calculated results.

The following users thanked this post: miltonyorkcastle, Hellblazer, Xiaobeibi, RollinsCat

Script Wrecked

Re: Fun with randomness
« Reply #1 on: September 08, 2012, 01:55:17 pm »


I knew it!



Re: Fun with randomness
« Reply #2 on: September 08, 2012, 05:45:40 pm »
What does NWN use for it's dice algorithm? A PRNG? Especially the soul mother saves look rather cyclic in their deviation. No expert on that kind of mathematics, tho.


Re: Fun with randomness
« Reply #3 on: September 08, 2012, 08:56:33 pm »
Layonara uses a lagged Fibonacci RNG algorithm for crafting and loot drops.  It looks like that was never integrated into the Soul Mother saves, though after this much time and the apparent reliability of the algorithm, I could probably do so without much issue.  The Soul Mother checks still use Bioware's default PRNG, and there may be a few other places that do as well.


Re: Fun with randomness
« Reply #4 on: September 11, 2012, 07:25:46 pm »
Are lagged Fibonacci RNG algorithms more random than regular Bioware ones, or do they just cost more and come with a free air freshener?


Re: Fun with randomness
« Reply #5 on: September 11, 2012, 08:08:27 pm »
They are generally more statistically "even" in their distributions. Over a very large sample, this difference becomes increasingly minimal, but in theory it should make small samples (i.e. crafting attempts) a little more flat in the numbers it produces.


Re: Fun with randomness
« Reply #6 on: September 12, 2012, 03:25:34 am »
*Görmungard eats his dice after rolling below 10 for a Troll Ear*
Lousy rotten...