From what I read, the wavelength is so long that it is smaller than the smallest 'unit' of conservation of momentum.
So if it takes 10 sillybits to move mass the smallest amount possible (let's call it a moviebit), and this drive generates only 5 sillybits - apparently the Universe is obliged to still move the object one moviebit (even though the EM drive has only paid for 1/2 a moviebit*)
*It's nice to know the Universe doesn't deal in small change :)
One of the questions being asked is if they will still get the same output from *even less* energy :)
Apparently this theory is holding up due to its prediction that if the shape of the drive is different, the thrust would be generated in the opposite direction - I recall reading that this would explain some of the results being observed!
FREE ENERGY! Woot.