I have been using a technique that's almost identical to this for years, based simply on observing that the roots are symmetric around the min/maximum: differentiate, set to zero, then difference of squares.

This cannot be an unrecorded technique can it?