Wind comments on A rational unfalsifyable believe - Less Wrong Discussion
You are viewing a comment permalink. View the original post to see all comments and the full post content.
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (46)
Assume I observe X, Y, Z and form three hypotheses
A obviously has highest probability since it includes B and C as special cases. However, which one of B and C do you think should get complexity penalty over the others?
In you story:
The relevant comparison is: Given that God did X, what is the probability that God also did Y and Z, verses God did not do those things.
P(God did Y, Z | God did X) = P(God did X,Y, Z) / P(God did X)
v.s.
P(God did not do Y, Z | God did X) = P(God did X, and something other than God did Y, Z) / P(God did X)
I am uncertain about how to correctly apply complexity penalty, but I do believe that the multi explanation model "God did X, and something other than God did Y, Z" should get complexity penalty over the sing explanation model "God did X, Y, Z".
The belief "God caused some tiny thing, a couple thousand years ago", should correlated with the belief "God did this big thing right now". This is why I firmly believe that God did not cause some tiny thing, a couple of thousand years ago.
Phrased like this, I see what you're getting at; but in my mind, I was describing extraordinary, but different events. Say, miracle cures and miracle plagues or whatever. A whole bunch of locusts and your aunt being cured of cancer most likely have different causes. In that case, you first have to postulate an entity which can summon a bunch of locusts. The actual summoning need not be magical or spontaneous in nature, only their appearance. So keeping a bunch of locusts hidden away whilst feeding them (somehow), before releasing them like a plague, would do.
This SAME entity then also needs the ability to cure cancer. To me, adding abilities like this incurs complexity penalties on a pretty big scale. Especially when you start adding other stuff and start scaling this influence over time (same entity responsible for actions many thousands of years ago and events now)
This says that if you are, say, an Inca ruler and you hear about Spanish conquistadors, the fact that they can ride weird beasts AND shoot fire out of metal sticks AND do a lot of other supernatural-looking stuff implies that you should disbelieve their existence -- probably not a good idea.
In general terms, the complexity penalties you're are talking about are justified only if these different abilities are unrelated. But if, instead, all of them have a common cause (e.g. massive technological superiority), the penalties no longer apply.
I see.