I recently had a thought that was like this:
Man, I go down too many rabbitholes. I learn GPU kernel programming, ThunderKittens, CuTE algebras, before basic distributed GPU primitives. I learn harmonic analysis and character orthogonality relations before basic linear algebra. I have this bad habit where I keep diving deep into one specific spike, losing the forest for the trees, and then the fundamentals come back to bite me.
I think, framed like this, it’s a great insight into my psychology, is it not? I prefer the “deep stuff” over the “basic stuff”. We can further moralize this (your ego can’t take solving the basic stuff; you must calm it).
But would a “rational agent” really think like this? I think not. A psychopathic agent focused solely on the goal, will learn whatever they feel is the most useful skill for them right there.
Even if you strip the emotion out of my preliminary analysis, it’s still projecting some pattern onto the data that may or may not “be there”.
It feels nice to come up with these kinds of explanations, and especially when they’re framed along an axis that both is familiar (you’ve thought about it a lot) and “low” (monotonic argument - a self-deprecating position surely is more accurate and objective than a positive one; you can’t “delude” yourself negatively, only positively!).
Now - instead of taking this, and searching for an even more universal set of foundations that we need to compile down to in order to verify properly, I think it’s helpful to take a hint from philosophy proper - not a huge fan of the American pragmatists, but they’re right here. What is the point of such an action, irregardless of its “truth” value?
You are exposing the frame in this way because you think that there is an issue along this axis - you want to get rid of the supposed recurring bad behavior - so we frame the situation, diagnose the root cause, and we have to fix it. Seems obvious, right? What did this add? Did we not just rephrase the original representational claim?
Not necessarily. Because your representation wanted to represent the situation in such a way that a certain course of action falls out naturally from the framing. And maybe, hey, in that case - I have no objection; picking up a linear algebra textbook is a good thing. But be careful.