The thing here is that not everyone listens to music on systems that are actually capable of reproducing the dynamic range and subtle nuance of the voice while still delivering the peaks of a drumkit. Most are listening to music on cheap computer speakers, stock car systems, iPod headphones (GOD AWFUL PIECES OF CRAP), and generally far-less than ideal systems in far-less than ideal environments, so to make up for this they try to cram as much energy into the signal as they can... which unfortunately for those that actually listen on notable systems, means the signal is harsh and soul-less.
If major productions nowadays were released in a fairly uncompressed format, with the wide dynamic range and sweet transients we like hearing, people would bitch and moan because they can't hear the quiet passages... and that the loud passages would subsequently be too loud, since their listening environment can't cope with that sort of thing.
This is why IMO we can't just mix on a high-end monitor rig in a tuned room and be done with it, we gotta take our mixes to the type of speakers that the average listener will be hearing it on. I'm pretty sure it's the job of mastering to make the end audio sound it's best on whatever system it's played on, and not just high-end audiophile rigs.
As far as vocal stacking and effects, my mixes are fairly dry in comparison to most others, so I suppose I'm with you on that one. We need singers that can sing!
But crappy systems generally compress; cheap boom-boxes and mini systems, along with most car stereos, compress and limit a lot anyway, so why not just regulate this somehow, allowing the compression to be done by/at the end user. Those of us with audiophile systems can sit in armchair in between big speakers and revel in dynamic range, and then get in the car or go to the kitchen and listen to the same music, but squashed to get over the noise of the engine/frying pan. Surely this is the sensible way to go about things?