My own experiments in microphone loading, done in a class I teach at the university, have had the following results:
SM57s and their relatives are greatly improved by a load impedance of about 500 ohms, achieved either by an alternate tap on the preamp's input transformer or a load resistor added in parallel with the standard tap.
On the other dynamic mics we tried (Electro-Voice PL20, Sennheiser MD421 mk I, Electro-Voice RE15) a 500 ohm load had audible effects that were barely detectable, and the listeners were split on whether they were an improvement or not.
The one condenser mic we tried with a 500 ohm load, a Shure SM81, sounded distinctly worse.
All these results make sense, if we assume that earlier mics like the SM57 family were designed with minimal mechanical damping, relying instead on electrical damping from a low load impedance, while recent dynamic mics were designed with significant mechanical damping, making electrical damping superfluous. The SM81's response to a lower-impedance load also makes sense, in that it's being required to source more current than it was designed for.
So far we haven't tried any transformerless-output condensers; I would expect them to also sound poorer with lower-impedance loading, but no data as yet.
As for higher-impedance loading: my speculation is that it would marginally benefit transformerless condenser mics (less distortion), but it might be a problem for transformer-coupled condensers, which could develop a peaky high end and ringing because of the inadequate electrical damping of the transformer.
Incidentally, while reviewing the ADL 600, I tried switching the load on a Beyer M260 ribbon microphone. The "600 ohm" position (actually lower due to the manufacturing error described above) sounded significantly better on mandolin, which is a real torture test on the high end. That surprised me.
Peace,
Paul