There are several articles in the music industry that address what is affectionately known as "The Loudness Wars."
In the early 90's, louder mixes started rearing their heads, and the overall levels of everything in an arrangement began to increase in volume, but not necessarily dynamic range.
What does that mean? Well, imagine listening to Dark Side of the Moon, or Vivaldi's Four Seasons. An invaluable aesthetic which makes these albums what they are and cannot be diminished are the dynamics themselves. Those works wouldn't be what they are without the softness of particular themes, and the climactic explosiveness of their zeniths.
In short, serious musicians work extraordinarily hard to create dynamic range in their works for the purpose of making music that can move a listener beyond simply the notes which are being played. Playing with smoothness can evoke a particular emotional response. Playing with aggression can evoke another, but it is my opinion (and strictly mine) that the nuances that make these distinctions possible are lost in translation when they are not able to flourish and bloom in the way they were intended.
In fact, mastering engineer Bob Katz tells us that the overall volume of recordings has gone up 20 decibels in 20 years. Compare a recording from the 70's to one from today, and you'll see what I mean.
Another thing that will stand out, yet again, is the complete absence of dynamic range. Now, since radio compression is evil and can apply as much as 40 decibels of compression to a work before it hits the airwaves, I would be willing to concede that it plays a role in the overall sonic degradation of a masterwork. However, having played my fair share of MP3's on my iPhone, I can tell you that the builds from verse to chorus to bridge are virtually indistinguishable.
Compression is generally applied to audio tracks with uneven volume discrepancies. In a lush mix, the balance of these textures can be lost in translation, so compression is often used to squash bigger waves of audio to the same volume as the smaller waves, then the entire track is boosted with makeup gain. Voila! A consistent track. The problem with applying these things directly to instruments is that the attack of a drum or the clarity of a vocal is found in those peaks. If you squash them, that no longer exists. Which is why I favor volume riding individual tracks and using parallel compression for character.
Couple aggressive compression with what is affectionately known in the trade as "brickwall limiting" where the average volume of the overall mix is limited to an overall RMS level of 8-10 dB of limiting, you can forget about dynamic range.
Is it wrong? Actually, it, like all things, is a matter of personal preference.
Personally, I'm not going to be overly concerned with producing my music for marketability at the expense of producing the music that moves me. Give me dynamics all day, everyday.