@ID_AA_Carmack ML definitely seems like a field where many intuitions don't hold and most ideas don't work. There's a lot more trial and error, seeing what sticks, than in software engineering, it seems to me.
