Well, it is a good subject and my reply probably will be lengthy. However, do not expect to read anything worthy in there as I have no answers.

We all know about the compression, the people with develop listening culture and more or less evolved listening intelligence can recognize compression, or more frequently the signature of the mechanisms that create compression (a loaded statement). We all know about the negative consequences of compression but do we always know the reasons and the mechanism why the compression took place? (Let disregard the situations when a compression applied intentionally)

BTW, Amphissa recently has pointed out an article at  http://spectrum.ieee.org/aug07/5429 that related to the subject.

So, if we do not know where the compression comes from then let at list to identify the source of compression.

1) Recording process. This is a big avenue for compression. The recording equipment, the acoustics, the microphones, the microphoneing techniques and many other things could contribute to compression. I personally do not record and have no experience with subject, perhaps someone who does would be able to say anything rational about it.

2) The format. Different formats have own fundamental influence to compression. I am completely confused about it as I have witnessed in many instances when a format with less dynamic range subjectively sound less compressed then formats with objectively higher dynamic range. Perhaps the key is implementations?

3) The playback electronics. The lowering noises, increasing dynamic range and providing a sufficient extension of to peak-power over operational power do help. Still, there are zillion of other factors… How deep should we go into assessing the contribution to compression? Some people look at the shape of the electron cloud inside of tube and the space between the elements of the tube as a prediction of dynamic characteristics of the tubes. Then we have the operation point, loads, the power supplies, the components, the parasitic capacitance and inductances, the magnetics … etc… etc… etc.. Go figure…

4) The acoustic systems. The sensitivity of acoustic systems is defiantly the key but the highest sensitively number is not automatically brings higher absolute dynamic range. There are many factors in loudspeakers that are responsible for compression but the irony is that by intellectually take care of them would not be an assurance of better subjective dynamics. In this section I have some expertise and I would talk about it a lot but how worthy would it be? Beside there is always a mystery for me that here and there pop up some products that  define all compression rules and being low sensitivity loudspeaker they play subjectively as then have no dynamic limitations (The Wilson Grand Slam and above for instance)  so, go figure again…

5) The acoustic environment of playback. That is very important contributor to but it is not without “issue”. Overdumping increases subjective perception of compression ….

6) Electricity. A big contribution to compression. Recently I was informed that when a very peak of our voltage sinusoid is clipped then it is the key for compression as current is cared at the very last tip of the wave. The current distortions, current patterns and this contribution to sound are a separate subject. It is not well researched; at least I know little about it…

So, did I say anything useful about the nature of compression? I do not think so. I wish I know something useful then I would be able to employ it.