I always considered magnetic and electrical "waves" to be distortions in the space-time continuum, much like gravity, not as something physical that moves through space. Clearly it is a different sort of distortion, but the ways in which electrical and magnetic fields interact with solid objects are not entirely different from the way gravitation interacts with mass. Somewhere in there lurks the Unified Field theory.
What if spacetime was quantized? Imagine spacetime not as a continuum but as a multidimensional map of successive discrete "tiny squares" each described by an unknown "quantum entity" (don't know how to call such thing surely equal or below the Planck length). Then an electromagnetic wave would be the physical interpretation of a step by step propagation through one "case" to the next one, of the "activation" and "deactivation" of such a quantum entity through spacetime. As for the associated particle (a photon, but why not also any particle composing matter) would be the physical local interpretation of an "activated case" (or the average of a group of activated cases, if one wants to include Heisenberg's uncertainty principle in that very rough idea).
Following that idea, nothing really "propagates" physically, only the information, which triggers the apparent movement of the wave and of its associated particle. A moving particle would then be like those light arrays on top of a wall in the dark, sequentially switched on and off rapidly with a little delay with respect to the previous one, giving the impression of a luminous object quickly propagating through space, whereas in fact there is only an apparent propagation and no movement at all… 
Another way to express such a view: considering the atomic orbital of an electron in an atom. With the appropriate amount of energy, an electron can "jump" from an orbital to the other. This is a quantum leap, a discrete atomic electron transition. The wave function changes. But fundamentally, is it really the same electron which jumped from one orbital to the other, or is the higher energy electron a different one than the previous one described on a lower energy orbital…
There are fundamental differences between gravitation and electromagnetism, even considering the field theory without any quantization.
One very interesting thing about gravitation is that in 4 spacetime (3 D space + time) (*) one can have a zero stress-energy tensor, and still have non-zero energy and momentum in the gravitational field.

This follows from the fact that zero right hand side of the gravitational field equation (zero stress-energy tensor), means zero left hand side (zero Einstein's tensor). But zero Einstein tensor in 4 spacetime does not necessarily mean a flat spacetime. The equality is between the Einstein tensor and the stress-energy tensor. Zero Einstein tensor does not equal a flat spacetime geometry. The Einstein tensor is equal to the difference between the Ricci tensor and the scalar curvature (times the metric tensor).

Both can add up to zero, and yet have non-zero components. ADDED IN EDIT: In 4 dimensions the Ricci tensor can be zero and yet the space be curved: non-flat. Since Ricci tensor equal zero does not necessarily mean flat spacetime, therefore one can have zero stress-energy tensor in 4 spacetime and still have non-zero energy and momentum in the gravitational field ! One can have gravitational wave disturbances with zero source: zero stress-energy density tensor.
This is very different from electromagnetism where the electromagnetic fields (photons) do not carry any charge. In a gravitational field one can have a zero energy density, and still have gravitational waves. Thus we have self-interaction in gravitation due to the nonlinearity of the gravitational equations. A gravitational wave with a small energy relative to the curvature will travel along a null geodesic in the curved spacetime geometry. This is a different path than it would travel in the absence of the spacetime curvature. Thus one can have self-interaction: the gravitational field interacting with itself.
This issue involves energy conservation and self-interaction in 4D spacetime, something that many posters discussing "overunity" really struggle with. In General Relativity you can have energy and momentum on the left hand side of the equation, unlike charges in electromagnetism (electromagnetic waves in vacuum or in space without charges do not carry any charge: photons have no charge).
(*) This is only possible in 4 spacetime (3 D space + time) and higher. In 3 spacetime (2 D space + time) a zero stress-energy tensor necessarily implies a zero curvature of spacetime (because in 3 dimensions or less zero Ricci tensor means flatness) and hence in 3 spacetime (2 D space + time) the gravitational field would not be able to carry energy and momentum. In 4 spacetime electromagnetism, the electromagnetic fields (photons) do not carry any charge.
ADDED IN EDIT:
The issue has to do with the number of components of the tensor that specifies curvature of space: the number of independent components of the Riemann curvature tensor. The Riemann curvature tensor has 4 indices:

But the curvature tensor that appears in Einstein's equation is not the Riemann curvature tensor, but is instead the Ricci tensor which has only two indices:

In 3 D the Ricci tensor has 6 independent components, exactly the same number of independent components as the Riemann curvature tensor has in 3 D: also 6.
Therefore, in 3 D, vanishing of the Ricci tensor implies also vanishing of the Riemann curvature. In 3 D, vanishing of the stress-energy tensor implies vanishing of the Ricci tensor, and vanishing of the Ricci tensor implies vanishing of the Riemann curvature. Hence in 3 D vanishing of the stress-energy tensor implies a flat geometry.
However in 4 D, the Ricci tensor has 10 independent components and the Riemann curvature tensor has 20 independent components. For 4 dimensions or greater, there will be fewer components of the Ricci tensor than components of the Riemann tensor.
Hence for 4 dimensions or greater, the Ricci tensor can vanish, and yet the Riemann curvature tensor may not vanish. Therefore for 4 dimensions or greater vanishing of the stress-energy tensor does not imply flatness of spacetime.