While I do believe, cautiously, that the thrust Shawyer found in the emdrive is real (emdrive: a microwave oven shaped as a truncated cone) and that Shawyer and then Juan et al & NASA should be respected for finding and testing it respectively, I'd like here to criticise Shawyer's theory of it, which I believe is confused. I'll try to do this in an accessible way, and briefly discuss my alternative MiHsC explanation.
In a New Scientist article on the emdrive Shawyer said: "Key is the fact that the diameter of a tubular cavity alters the path - and hence the effective velocity - of the microwaves travelling through it. Microwaves moving along a relatively wide tube follow a more or less uninterrupted path from end to end, while microwaves in a narrow tube move along it by reflecting off the walls. The narrower the tube gets, the more the microwaves get reflected and the slower their effective velocity along the tube becomes. Shawyer calculates the microwaves striking the end wall at the narrow end of his cavity will transfer less momentum to the cavity than those striking the wider end (see Diagram). The result is a net force that pushes the cavity in one direction. And that's it, Shawyer says." (from New Scientist: the end of wings and wheels).
I think there is a lot wrong with Shawyer's explanation, and for once I'm not alone on this! For example he uses special relativity for an accelerating system and his explanation wrongly indicates a thrust towards the wide end of the cavity, but I'll focus here on the asymmetry in force he claims to exist between the large end plate and the small. I haven't seen an accessible refutation of this, so here goes. Referring to the quote above, he does not seem to consider that the photons 'reflecting off the side walls' will exert a pressure force on those walls, which will have a component towards the narrow end. The schematic below shows a simplified setup for ease of explanation: first of all an equilateral triangle which can be generalised to a 3-d cone by an integration using a rotation around the central axis of symmetry (the horizontal line) without changing the conclusions, so this is a nice accessible way to do it.
In the diagram on the left the microwave photons bouncing around inside the cavity make a force (F, per unit area) that pushes outwards on all three faces equally (the blue arrows). Assuming that the length of the sides are equal (l) it's easy to show that the left-right components of force balance:
Force right = Fl
In the diagram on the right I've truncated the triangle/cone halfway down its sides to represent the truncated cone of the emdrive, and again the result can be generalised to a 3-d emdrive by integrating things in rotation around the axis of symmetry (dashed line) without changing the conclusion. So now we have:
Force right = Fl
Force left = 2*left force on side walls + left force on flat end
Again the forces balance in contradiction to what Shawyer said in his interview (which may have been an over-simplification on his part that he needs to address). This, and his use of special relativity for an accelerating system, and the wrong direction of his predicted thrust indicates that his theory is confused.
I have suggested an alternative, which is perhaps less confused, but requires new physics. The schematic below represents the MiHsC explanation of the emdrive. The brighter colours on the right are meant to indicate that more modes of the zero point field can exist at the right hand, wider, end of the cavity. That means that the mechanism MiHsC uses to produce inertial mass from an asymmetry in Unruh radiation caused by a Rindler horizon (an asymmetric Casimir effect, see McCulloch, 2013) is more effective on the right hand side. So, as photons go right (lower arrow) they gain inertial mass (arrow thickness) and as they go left they lose it (upper arrow), so to conserve overall momentum the cavity is predicted to move left (left hand arrow). The predictions from MiHsC fit the data, reasonably well (see McCulloch, 2015), but not yet such good agreement that I'm writing to Nature about it..
McCulloch, M.E., 2015. Can the emdrive be explained by quantised inertia? Progress in Physics, 11, 1, 78-80. Link
McCulloch, M.E, 2013. Inertia from an asymmetric Casimir effect. EPL, 101, 59001. Link