The more one looks into it, the more one realizes that aspects of mirrors that initially appear to be intuitive are, in fact, extremely hard to explain.
I'll call the other track "mostly classical." As far as I can tell, it is represented solely by Feynman and a Geneva University theoretician who applied the Feynman path approach to detailed calculations of reflection, refraction, diffraction, and other phenomena in 2005. This paper is quite technical but still worth a read. Though all other authors acknowledge Feynman's genius, it seems that few if any actually use his methods in QED calculations, because they are typically more complex to apply than Schrodinger-based approaches.
The Geneva paper seems to be the first time anyone has documented detailed consequences of the Feynman model. (Feynman didn't record his own calculations for this example.) But before I go there, let's look quickly at the other problem: that phase assumption. Path length is certainly a factor in phase at detection, but what about the phase when a photon starts on one of these paths? If the source is a laser, you can assume phases are equal at creation, but this is not the case for a regular light source. If you assume that amplitude summing (interference) at the detector is between different photons travelling on different paths, path differences still affect the result, but lack of correlation between source phases will lead to random and time-varying (noisy) interference at the detector, which is not what we see.
Back to what the Geneva paper has to say:
- Feynman's explanation is more fundamental and more powerful than the Schrodinger approach. Schrodinger can be derived from Feynman, but not vice versa, because Feynman represents correlation between space-time events (between paths), but Schrodinger cannot.
- In detailed calculations using the Feynman method applied to photons, all classical behaviors of light as waves emerge as expected. Reflection, in particular, follows Feynman's example.
- Photons propagate over macroscopic distances in a completely classical manner. (Heisenberg applies to the creation and detection of a photon and to scattering events, but not to simple propagation.) But we must consider all possible paths in the analysis.
- Paths add in the same way that waves add. We add amplitudes with phases to obtain interference at the point of detection.
- This brings us to the creation phase issue and the only conceptually difficult requirement of the Feynman method. Different photons have random relative phases at creation, but any given photon is trivially in phase with itself when created. Therefore, to obtain the results we see, interference cannot be between different photons travelling different paths. Each photon individually must travel along all possible paths and (indelicately) interfere only with itself at detection. This is the only way we can avoid that noisy interference between uncorrelated photons, and it is why experiments testing one photon at a time give the same results as for multiple photons at the same time.
I should caution that the 2005 paper is an interpretation, and that it makes predictions of new behaviors that have not yet been tested. But absent counterexamples, I find this interpretation very appealing. It is in complete agreement with Feynman's explanation, and it conserves all classical and intuitive understanding of light behavior based on photons, with just one exception. That exception is a doozy: A photon must travel simultaneously along all possible paths to the point it is detected and resolve itself through self-interference at detection. If we suspend disbelief on this one point, everything else is completely intuitive.
So what do we make of this one difficult point? Travelling simultaneously along all possible paths is definitely neither classical nor intuitive. Perhaps we see a particle travelling along all paths as a projection from a simpler path in something more fundamental than space-time. There are hints of this in a recent article, "A jewel at the heart of quantum physics," which suggests that the space-time so familiar to us may not be the most basic representation of reality. Whether we will appreciate this as an improvement in intuitive understanding is up for debate.