So here's an example of a particle analogue of the effect I think they're seeing in their EM drive model, and how "equal and opposite" might be universally true, but still only in a limited sense.
Say you're in a spacesuit inside an enclosed cube with no engine in a Zero-g vacuum. You need to get moving somewhere, and all you have are 15 billiard balls which you either don't want to or can't just throw into space outside the cube; they must come with you one way or another. How do you get moving in one direction more than another? Well, you arrange 14 of the billiard balls in a certain way in the middle of the cube, put your back up against one wall near that wall's middle, and throw the 15th billiard ball at the other billiard balls in such a way that the ball pushes you against the wall, propelling the cube forward, while the ball travels to 2 of the other balls, hits them so it stops relative to where the other 2 balls were before it hit them, and sends each of them in 45 degree angles to the thrown ball's trajectory before it hit them. Then those two hit balls each hit two more balls each, each of which in turn also travels at 45 degrees to the former trajectory of the ball which hit it, and so on, until some of the balls are even travelling in the direction the cube is travelling. Assuming you threw the ball hard enough to give the cube enough momentum, the cube will slow down but not stop when it collects all of the balls on the far wall, because not all of the original counterthrust of the thrown ball is going in the opposite direction of the cube when the cube collects the balls bearing that counterthrust.
The general idea is to throw a finite number of particles into a numerically and voluminously larger but still finite group of particles, so the momentum of the thrown particles can scatter via the larger group into directions which are not exactly opposite in direction to the counter-momentum given to the vessel by the thrown particles before the particles bearing that momentum are recaptured by the vessel.
So a more advanced example might be an enclosed cylinder containing air. A fan with a diameter significantly smaller than the cylinder's diameter is fixed to the cylinder, close to and in the middle of one end of the cylinder, facing and blowing the air in the cylinder towards the other end of the cylinder, and we let some sort of convective system do it's work. I don't know what the optimal power/periodicity of fan use would be, nor the optimal temperature/pressure/chemistry/dimensions of the various components would be, but that's the basic idea. A heat source exciting particles instead of a fan blowing them might also suffice.
It might even be that an astronaut stranded in space with nothing but a spacesuit could propel himself forward by simply tapping his finger to the front of his suit near his center of mass repeatedly.
What are your thoughts? Did I miss something?