I’m not a mathematical person, but I am a crazy guy, so I’ve always wanted to use Bayes Theorem to determine Mythic/Oracle odds for me.
I quite like the idea of assigning the odds through an unbiased calculation that allows me to somewhat eliminate my own biases from the equation. It would make the illusion of an autonomous oracle stronger for me because I wouldn’t be directly deciding the odds of the roll (yet they would be consistent with my assumptions about how the setting behaves).
——-
Today I may have gotten a little closer to executing this idea, though a a free iOS app that performs Bayesian network calculations. (App name is mBayes:
https://itunes.apple.com/us/app/mbayes/id1086016884?mt=8 ).
The screenshots here area based on a question of whether a fictional set of Hyperborean-ish tribe will raid based on four criteria: enmity, fear, whether they want something your tribe has, and whether they can only get it from your tribe.
The first picture shows the prior probabilities of being raided given a set of conditions and their assumed probabilities (starting at a base %50):
# Whether they’re your enemies (+5%)
# whether they fear you (-5%)
# whether you have something they want (+5%)
# whether they can get the same thing elsewhere (-5% )
The next picture shows the probability of being raided when the all the conditions are unknown. The next one shows the probability given true or false values for the variables. The last two show the calculations in reverse showing the probability of the variables depending on whether you’ve been raided or not.
This felt like a mini game in of itself.
Grey Knight thanks for the tip. I was using this pocket mod from Pocket Dungeons that attempts to minimize the bias. Good to keep in mind, though.
ReplyDeleteboardgamegeek.com - PocketModDice | Pocket Dungeon
I think we could improve that a bit by having a table where you choose a random row, with each "standard" die type having its own column. That would save on modulo calculations and make most things just a table lookup. You could still use a halving operation for d3 and d10 etc if you're low on column space.
ReplyDeleteYou can do the standard die types with just 120 rows (=2*2*2*3*5), which is about 15 rows per page if you have an 8-page pocketmod. 20 rows per page if you want to keep two pages for a cover.
I did a straight pick-one version inspired by the Pocket Dice Roller, if anyone's curious.
ReplyDeletehttp://www.drivethrurpg.com/product/195037/Pocket-Dice-Roller
https://drive.google.com/open?id=1hiR4atig-u5YmR0n1XEInD8cqcP6l0yb
I looked into the Bayesian thing and I'm still noodling with it -- I have some thoughts on it to share a bit!