The big picture: we're approximating functions (some nasty) with nice ones (polynomials):
Let's talk error now. Remember error? As we discussed for the case
of Simpson's rule, it's nice to give an estimate -- but even nicer when
you can provide an error estimate, also....