[roesel]

The magnitue drop is reasonably simple to understand in terms of fields. The oscillating optical field might be less effective at exciting material oscillations in the middle filter due to a mismatch in polarization, but it still does so at the same frequency. You can think of it as multiple photons (incoming field) collectively exciting the same electron on the same frequency but with reduced efficiency. The electron then re-emits fewer photons (outgoing field) of the same wavelength, leading to a lower light intensity detected after the filter.

[Arwill]

I don't think photons are absorbed and re-emitted by electrons. At least that argument does not hold when discussing light slowing down in glass or water. Light is affected by the electromagnetic field of the material it is going trough, is slowed down, or absorbed based on some of its property, but photons that go trough are going trough without collision. Photons that get absorbed and re-emitted are scattered in all directions, and are mostly lost. You would not see a consistent image trough a polarising sunglass, if the photons you were seeing were re-emitted photons.

[roesel]

I think using the term "re-emit" is a mistake on my part. I fully agree that the photons are not absorbed and re-emitted.

However, their speed of propagation in the material, and their properties can be altered by the interaction with dipole moments of the material. Therefore, photons do indeed "go trough without collision" but not without interaction. The dipoles in a dielectric material such as glass need to take the energy to oscillate from the field, and then they give it back to the field by radiating it out again.

I agree that this is not "re-emission" in the absorption/emission diagram sense, and I understand that the field in a material cannot be separated into the material portion and the vacuum portion, since it is both at any given time, but what I have tried to outline in the previous comment is still a useful representation of the role of the material which the original article chose to leave out. I should have used "radiate" and "dipole" instead of "emit" and "electron" to make it a bit clearer.