I was browsing XKCD and came across this comic discussing differentiation and integration:
This resonates true with my college-level understanding of calculus: differentiation is very procedural, while integration sometimes requires a stroke of insight. It appears to me that given a mathematical expression, it's possible to transform it into a computer-science style expression tree and recursively compute the derivative. However, that doesn't seem sensible to do for integration.
Is it an artifact of how we teach calculus? Is it an artifact of how calculus was created/discovered? I could believe that differentiation looks easier because most calculus classes only work with elementary functions and there are simple rules for elementary functions. Or perhaps there is a "master theorem" for integration that is too complicated to cover in engineering calculus (which is my background).
On the other hand, is there a proof (or intuition) as to why integration is more difficult and/or impossible in some cases? If integration is mostly defined or encountered as the reverse of differentiation, then I could believe integration is more difficult in the same way most believe P != NP (with a stretch of metaphor).