Bob_for_short

Should a good physical theory predict phenomena that happen always? Yes, of course. What is a probability of a phenomena that never happens? Zero, of course. Consider then a Rutherford scattering of an electron from a proton in QED. The first Born approximation gives indeed a Rutherford (or Rutherford-like) cross section and the textbooks represent it as a success. At the same time any scattering is experimentally accompanied with photon radiation. The probability of any photon radiation is equal to unity. So QED predicts a phenomenon that never happens - scattering without radiation. Only much later, when treating the infra-red catastrophe, QED books correct this QED failure but not before. I advanced a theory where the radiation is unavoidable: the elastic cross section (i.e., without radiation) is equal to zero, as it should be. Only inclusive cross section is different from zero. In my theory the electron charge and photon degrees of freedom are coupled intrinsically and permanently. They cannot be decoupled unlike QED construction. But my pet theory is in an embryonic state, it cannot be compared to the fourth-order QED calculations of (g-2) yet due to lack of funding.

[math] \frac{g_{\rm th}-2}{2} = 1159652140(28) \times 10^{-12} [/math] In some simple cases of the perturbation theory, like in http://arxiv.org/abs/0906.3504, page 6, the relative accuracy can be better even in the third order.

Let me mention one oversight (many studied, none noticed): the positive charge atomic form-factors. I was really surprised to discover this simple thing myself (1985). It is still unknown to the majority of physicists. You can find details in my article "Atom as a 'Dressed' Nucleus".

Such a precision is explained with a very small value of the expansion parameter (about 0.001). Currently the calculation is made to the forth order so the precision is very high. It is not correct to demand from new theories to overcome this precision - it would take too much effort from one person.