BTW: I was always confused by the discussion of emissivity vs. temperature vs. radiation. To me, emissivity is a fudge factor that makes radiation and temperature approach an ideal black body. If you just make emissivity 1, then you will always underestimate temperature. You can see that if you plot the Jones data and match it to a pure t^4 curve. The emissivity will vary to make the curves match.
That was Levi's argument - sort of. "Well even if you set e = 1 you still get COP=2". But of course if you make emissivity = 1 you overestimate power output. In the Lugano case the Al2O3 gives emissivity in IR band close to 1, but total emissivity (determining output power) of only 0.5 or so. The conditions and material used just happen to maximise the errors from any assumption that emissivity seen by the camera is the same as emissivity used for power output calculation.