Yeaj,Yeah, this concerns me, as well.
Emissivity is a complicated beast. IR sensors do not measure the surface temperature. They measure the thermal energy that is visible from the surface of an object. Case in point is trying to measure a chrome group head with an IR device. Chrome reflects IR wavelength light, so you read the temperature of of the reflection off the group, not the temperature of the group. In my experience working with Omega IR sensors, you have to calibrate the sensors any time the surface finish or color changes. I use these regularly to monitor the heat build un in polyurethane nip rollers and they are sensitive enough that a blue roll cover will read a significantly lower temperature than a black cover when the core of the rolls are at the same temp.
What this means for coffee remains to be seen, but I think the end result will be a reliably consistent and high-precision reading, even if it isn't particularly accurate.
We could get grumpy about the fact that a green bean at 375°F would look different to an IR sensor than a bean entering FC at the same temp would look, but the fact is that a green bean can't exist at 375°, so these issues may be more academic than actual issues with real-world scenarios... Also, I think that as the bean roasts and changes color, it will emit more thermal energy (emissivity approaches 1) than when in the earlier phases, so the accuracy should get better as the roast progresses, which is way better than the alternative.