Me too. I'm an impending M8 shooter and would like to think they're less important than we hear. It's hard to believe it's too bad when somebody who shoots as much as Sean Reid didn't even notice it until other people started complaining.
I would like to think a lot of things... but physics says whenever IR energy reaches the sensor of a M8, the color parameter estimates from the Bayer interpolation calculations will be in error. The more IR energy there is, the greater the error. The error can be amplified or attenuated by the lens glass elements and coatings and these errors may not be evenly distributed across the image. The parameter estimate errors can not be corrected in post processing because we can't know enough about the distribution and intensity of the IR energy that happened to reach the sensor. And you are right. It's not "too bad" when there is no, or hardly any, IR light is reflected by your subjects. It is even less bad when the IR light intensity is the same everywhere in your subject and your lens elements transmit IR energy equally at all points.
Sean Reid happened to make a mistake in his early analysis of the M8's color fidelity. This doesn't mean the laws of physics no longer apply. It only means even information one has to pay for on the Internet might not be complete. And we all make mistakes... so it goes.
Common sense dictates that IR filters are profoundly important for M8 color photography. Otherwise when people use infrared film all their negatives would be blank. IR film photography is not limited to subjects that that contain certain types of synthetic fibers that efficiently reflect IR energy. Just search for IR film photos on Flickr and see for yourself the range of subjects where IR light is present.
Technically the requirement for IR lens filters is caused by the fundamental nature of Bayer sensors. One can never estimate the true, but unknown, visible-light photon counts measured by a given sensor site because the amount of IR photons also counted by that site can not be estimated. This means that any time IR energy is present, the M8 sensor visible light photon counts are in error. They are wrong.
Some lenses have IR hot spots. This means the errors induced by IR photon counts may not be the same everywhere in the image.
Early in the M8 release cycle people thought they could design Photoshop actions to estimate the true, but unknown, visible photon counts in any given M8 image. This approach failed for the reasons I explained above.
Color reproduction in photography is difficult and complicated. I do not see how looking at images without filters is of any use at all because except for the extreme cases, you have no idea what the visible color was in the first place. A subtle error is still an error. Nor do you know where the intended focal plane was.
