View Single Post
Old December 5, 2020, 07:47   #7
bughunter
Adept
 
Join Date: Nov 2019
Location: Pasadena CA
Posts: 129
bughunter is on a distinguished road
Percieved disagreement between posted random chance event rates and subjective player experiences have come up in many of the games I've played and moved devs to examine the RNG to verify that they are indeed "random."

And in every case they find that the "randomness" is within expected parameters. I put this in "scare quotes" because I'm using shorthand for "probability distribution across a large, statistically significant sample."

The distribution of random results across very large sample sets does not address "streakiness" however. Meaning, unevenness in the probability distribution within small samples. This is a topic that I am not qualified to discuss, but spawns long threads populated by people not as humble or as unqualified as I... Bottom line is, some random number generators seem to be more "streaky" than others, and can produce results that seem non-random over the course of a single play session for a single player.

If you haven't done so already, read these pages for introductions to the statistics far better than anything i could offer:

- https://www.random.org/analysis/
- https://en.wikipedia.org/wiki/Randomness_tests

There's a Dilbert cartoon in the first link that illustrates the problem quite nicely.

As a counterexample to the OP, I offer this:

I started a halfling rogue last night. Since he got the spell at cl3, he has been casting Detect Monsters as needed (after descending stairs, etc.) for a total of maybe a dozen times. It has a failure rate of 42% at his current cl7 (down now from 50% originally).

It has not once failed.

(Edit: and of course, now on his next cast, he failed. The RNG gods are fickle.)

Last edited by bughunter; December 5, 2020 at 07:54.
bughunter is offline   Reply With Quote