For the record, I did mention the relationship between the input impedance of the Intactaudio device and the output Z of my MC2000 (2 ohms into 20 ohms), by way of speculating on how it is working with that particular cartridge. Nearly any very LOMC cartridge with a very low internal R would tend to work with the IA device in a voltage-dependent manner, I guess. Still, the IA device also sounds great with one of my other LOMCs that has a high-ish internal R and where the amplification would be more in the current domain. (Dynavector 17D3 which has internal R = 32 ohms.) Be that as it may, I am driven to try out a very very low input impedance phono some time, with the MC2000, just to satisfy further my curiosity.
Raul, When IA used the term "tests", he meant exactly that. He made measurements using the proper test gear and said nothing about listening tests. Not that there is anything "wrong" with either approach, except I would posit that subjective testing based only on listening is fraught with error, not the least of which is error due to listener bias. Furthermore, Dave (IA) was not per se testing Atmasphere's hypothesis that low resistive loads impair HF tracking. He was testing a finding put forth by Moncrieff, to the effect that unloaded LOMCs (i.e., with a 47K ohm load) exhibit a form of IM distortion (when he used one particular pair of frequencies for his IM test) and that the distortion is reduced when you use progressively lower resistive loads, down to 100 ohms. Nevertheless, Moncrieff's and Dave's measured findings might (or might not) provide a scientific rationale for your position. As for me, I wonder about the particular pair of frequencies that Moncrieff used for the IM test and whether one might get a different result for other pairs of frequencies. (I think the upper one was 4KHz; would that be high enough for the effect described by Atmasphere to kick in?) Or for different cartridges. (I don't know what cartridge was used by either Moncrieff or Dave.) Finally, I personally wouldn't care if using a 47K load produces a bit more measurable distortion, as long as the result sounded closer to live, real music compared to a 100R load.