It is not. $e_r$ is a rotating vector, not constant. Also $r$ is not really constant either.

Actually, if you think about it, the only plausible way you could have something in space with zero jerk and non-zero acceleration would be an interstellar ship maintaining constant acceleration, so the suggested rule is almost exactly wrong.

I'm afraid you'll have to explain your perspective to me as if I'm five years old. In your original comment, you say,

>a constant times the $-e_r/r^2$, and the time derivative of that is definitely not zero.

I don't see a term there that is time-dependent.