Heya everyone, it's been a while since we've posted anything on here. Mostly didn'tve ideas or'f we did wanna post sth't generally felt so short n non-important or personal we'd just put't on Fedi instead. But here we go, we got a topic for a longer-form blog post again. Disclaimer: This is a blog post bout philosophy, apart from what we learned in school, most'a our knowledge on philosophy comes from Existential Comics, so credits'a them for inspiring the thought process that lead'a this (n also being our main source for stuff lol).
So, Scottish philosopher David Hume was known for, among other things, questioning the validity'a inductive reasoning n causal inference. We've thought about this a lot, like, y exactly do we do that? What leads us'a believe ? Intuitively we thought'a the argument that "well it's worked out so far", but that's just begging the question really. N then thinking bout't more, we found't surprisingly hard finding any non-circular argument for't.
One'a the most famous refutations'a this's Immanuel Kant's a priori ideas. Basically, he argued that for making sense'a any empirical observation, certain preconceptions were needed, like unity vs plurality or existence vs nonexistence, but also including causality, so empiricism without causal inference would be a nonsensical idea. Maybe that's right but't feels like there's maybe still a missing piece here, also we only found that while doing research on this post, having conceived a slightly different approach beforehand. Another thing that would come'a mind's Judea Pearl's do-calculus for verifying causal models from observational data, which does help for establishing causality based on certain axioms n otherwise only correlational data, but then't still relies on induction being valid in the first place.
So, when would induction be valid? When would't be invalid? Induction is somewhat reliable if the world we live in follows certain rules that stay mostly constant over time, assuming we wanna learn more about those rules. It's invalid if things are entirely arbitrary and there are no rules or the rules constantly change. So let's say the rules could just change like that. What would that mean about the world? We wouldn't just risk that emeralds suddenly turn blue. What could happen is planets loosing their stable orbits, atoms falling apart, elemental particles just vanishing, etc. I.e. all life could be wiped out in one instant.
So, let's put ourselves in an original position where we don't know which world we live in, the mostly-consistent-rules one (world A) or the rules-can-change-at-any-time one (world B). Which world's more likely? We wouldn't know unless we use induction, which for the purposesa this argument, we can't, for now. Ergo, we gotta assume they're both equally likely. So, how would we wanna act? If we live in world A, we'd prolly wanna act as'f we lived in world A, since then figuring out the rules behind how things work might be very useful n acting as'f that was impossible'd be pretty detrimental. Let's say we lived in world B, would there be any benefit'a acting differently? Maybe, I mean'f we could all just vanish in an instant we might as well live every moment as'f't were our last. But then again, would acting as'f we were in world A be so bad? I mean we might "waste" some time tryna figure out rules that're entirely invalid, n then ofc we'd try behaving as'f they were, but then again'f there are no persistent rules anyway there's also no "wrong" rules'a live by, so, whatever, not that bad. So, I'd argue from a game theoretical perspective here, a self-interested actor should prolly embrace using induction up'a the point where't might be useful in this world A.
Edit: I just realized another thing. In world B, we wouldn't even know whether pleasure would always be a good thing, or whether a thing that was nice a day ago would still be nice today. So a self-interested actor wouldn'tve any "correct" way'a behave in world B n their only hope'd be that they actually live in world A. So the world A strategy's in fact the dominant strategy.
That's all folks, remember correlation doesn't equal causation but persistent significant correlations across large samples do imply some sorta causal relationship, whether direct or indirect! Stay safe n have a good pride month <3