I wanted to share some thoughts, that’s way I
published an almost finished game in the previous post.
The game I created was inspired by Professor Judea
Pearl’s “The Book Of Why”, about Causal Inference. I’m still in chapter 4, because
of other duties, and because, even though it’s a book for the layman, the
professor, let’s say, respects the intelligence of the lay man a lot (maybe
that’s what really smart people do).
Anyway, I’ll talk about causal inference in terms of
Electricity, basic stuff.
We got these formulas:
In this images, there is no hierarchy, the formulas
all have the same importance, right?
But think: you can’t keep voltage stable, then push
the current to a higher level, causing the resistance to increase, right? No, there are actual causes and effects.
Because the real stuff is this:
So, imagine that a robot was trying to learn how to
project an electrical installation by just collecting data from existing installations, maybe getting to the
formulas as first presented here, with no causal model. How much data he would
need to collect if he didn’t allow himself to say “There are individual
mathematical one-way relations here”.?
I mean, how would he avoid making something that just
wasn’t as precise and economical as it could be, if he doesn’t separate things
in cause and effects?
So, I was inspired by Worlde (I even got stuff from a “How
to code Wordle” article at code camp) and also by the Professor Judea’s book, and
made this game.
Like, how long a “just relations, no cause” data
collector would take to master this game, and how long a robot that knows “there
are individual mathematical one-way relations here” would?
Nenhum comentário:
Postar um comentário