Here comes the answer you didn't want to read - it depends! ;-]
***
In the early days of recording, the signal chain during recording was far from being linear - frequency-wise as well as regading dynamics; fast transients could'nt be captured by the microphone, levels above a certain range weren't possible on the recording media available ... in a word: compression and limiting occured everywhere, and everybody thought this to be completely normal.
The microphones and pre-amps became better, and when digital audio was commercially available, suddenly fast transients and high dynamic ranges were possible to be recorded - with the result that most people described the listening experience as "small" or "poor", although in fact the chain was dynamically more transparent and "real" then ever before (... I know, I know, there are many caveats, but I don't want to start an "analogue vs. digital"-debate here).
What I'm actually trying to tell you with this long-winding explaination is that it's completely up to you what you want to hear: On many occasions, especially when it comes to AV-music production (film, spots, etc.) I highly appreciate the possibilty to compress the single instruments or -groups without sacrificing the overall sound - something which a very discerning factor from VSL-productions and conventional recordings where this is much more difficult to achieve. - OTOH, in a more "artsy" context, I would be happy to use the 90 (and with the appearence of the 24-bit ProEdition even more) dB of dynamic our library is able to deliver ... without noise, hiss or any alteration of the natural, signal-inherent loudness envelope.
Of course ... other people may have different points of view :-]
/Dietz - Vienna Symphonic Library