PHY Research Is Sick but Curable: An Empirical Analysis