I think a lot of people forget that when the tonality of an instrument changes, the wave shape changes. Something that has more low end (or, less spikey transients) will generally have a fatter looking (and of course, sounding) waveform.
It's my contention that transients are essential. I think what is being confused by some is "transients = bad" when really, you mean "spikey ice pick transients are bad," which I agree. Tape takes care of the spikey stuff. Digital picks it up.
One of the reasons I feel like digital is pretty acceptable on things like modern Jazz, film scoring, orchestral music, and other (what I would call) "hifi" type genres is because the transients are well preserved. Also keep in mind, however, none of these genres tend to have instruments that create super spikey transients, and even when they do (like drums on a jazz song) they aren't mic'ed or processed in such a way where they need a hard attack or a close mic'ed snare half the time. Plus, lots more room mics are used, mics have distance, and as we (should) know, air is a great compressor. Putting a mic close on something is a sure fire way to have a hard edged transient. The genres themselves soften the blows of the attacks because when they are well recorded, the dynamics of the band take care of that. Digital (in my opinion) sounds great on these genres.
Now when it comes to rock, pop, etc., we seem to like everything "slamming," as if it were played as hard as possible. I feel like the reason a lot of us like the sound of compression is because it has an energy. It makes things more exciting. It helps us try to capture, enhance, or harness that "live energy." Thus, drummers are hitting harder, guitarists are strumming harder, and bassists are plucking harder. One of the best interviews I ever heard was with T Bone Burnett - they asked "how do you get your records so loud, so competitive with the market, but without sounding compressed hardly at all?" He simply said "I had everyone play quietly and evenly." or something like that. Sorta blew my mind. The less hard the attacks are, the louder the sustain/body can seem by comparison. Not always, but in general.
So does that mean we should play soft in rock? No, not necessarily. Let's talk about my memorable experience with transients.
In my experience, there are few things that can solve spikey transients better than tape and good analog front end. I've had some success with saturation plugins, but when I'm recording the project myself, I'd much prefer to do it myself with a good front end, tube mics, nice pres, analog compressors, tube equipment., etc.
One of the biggest issues with compression ITB is not that the compressors suck, or that they don't "sound like the analog versions," I think it's that they're given a much higher transient to look at. For example - my friend Michael and I did a test regarding this phenomenon. We recorded a snare drum with a Radial mic splitter, split each to a 1073, one to tape, and one to the DAW. We then brought in the tape snare into the DAW just to compare the waveforms, and it was pretty obvious. We weren't even slamming the tape or anything. To the ear, they sounded about the same volume, the tape snare was a bit fatter and a little darker. The really interesting thing was when we tried compressing both.
You see, the initial transient hit of the digital snare was much higher in level than the "meat" of the snare. On the tape snare, the transient was the same level, but the "meat" was much higher. (Obviously, it's not that the meat was "really" higher, it was that the tape compressed that very initial part, which made the decay part of the ADSR envelope louder).
Okay, I know this is super rough haha---even a little exaggerated for demonstration purposes, I just did this drawing on the computer in like 10 seconds! I attached a simple drawing of "something" like the waveforms we saw. When we compressed each, the peaky transient of digital registered above the threshold, but the compressor was much less effective (because when it reached full gain reduction, it had very little to actually compress -- The sustain and body were so much lower in comparison to the transient. When we compressed the tape snare, it got squishy, fat, and still had a nice smooth attack. It actually was compressing the body of that snare. We HEARD it easier. It was very obvious, even with just a few dB of reduction. The digital snare compression seemed to not really do anything until we cranked it to -10, when it was really affecting the body of the instrument, but then, the transient sounded super spikey. See the issue?
This is just my rationalization of this argument. Others might have very different experiences, but maybe this makes sense to some of you. I've been writing a book (no masterpiece by any means) about mixing, and I show some of these tests in the chapter on compression because it's so important. A compressor is a machine. It takes a skilled engineer to understand what you're feeding the machine and how to manipulate it to the song's advantage. Constantly evaluate statements like : "X compressor (or Y compressor settings) will always work on snare." Tape snare? Digital snare? Loop snare? Deep snare? Piccolo snare? Sample snare? Wood? Steel? Brass? Fast song? Slow song? You get the point. End rant.
Slut note: having nice monitors is a big help to hearing transient in detail. I feel like a lot of low-end monitoring systems compress them or something - I don't know the details of technical monitor design, but they just don't have the detail on tap to allow you to really hear whether or not a transient is spikey, an "ess" is harsh, or a snare is spikey. Getting "flat" monitors is not the only factor - you need monitors that can actually let you HEAR how you modify the transients. My personal set, Barefoot MM35s...well they're just dandy at that.