or

3) there's a misunderstanding about ordinary least-squares.