Specialization is great, but only when it's necessary and efficient - I find it particularly interesting that you make this complaint here, in a startup forum, where programmers are acting as CEOs, web designers, PR men, testers, etc., often quite successfully. You can trim a lot of crap out of the equation and still do quite well if you're motivated.
Many of the roles you mentioned are all but irrelevant to most musicians, and given the great advances in software, they can replicate them themselves as needed, quite easily compared to the difficulty of mastering an instrument: a good musician that spends a couple weeks learning what to do can produce music that approaches release quality by himself on a laptop with a few $100 microphones. Learning about recording is now almost a prerequisite for taking on the job description "musician." The other things, which musicians rightly have no interest in learning to do for themselves (managing tours, etc.) are what managers should be doing, and I don't think anyone is arguing that managers be taken out of the equation, because they unquestionably add value, much like a secretary adds value to a business person.
Anyhow, as someone that has at one point worked on both sides of the music industry (as a musician and in production), I'd suggest that the vast majority of musicians have always been in the "cottage" music industry mode, but have still been happily and consistently producing fantastic amounts of music. The friends that I have that remained in the industry as musicians almost all make livings by playing shows and teaching lessons; the few that have gotten record deals with real companies have all ended up feeling cheated, and tend to wish they'd never signed the damn papers. Google the horror stories yourself: I assure you, this crap happens all the time, I've seen it from both sides and it's pretty sickening.
Most working musicians have no need for copyright protection, as it doesn't tend to affect their careers one bit. There will always be demand for live music and lessons, and the financial rewards from those things allow them to do what they love and produce new music constantly. Even when they do produce their own CDs (usually to sell at shows), they don't rely on copyright to sell them - to the contrary, most of them freely share almost everything they produce online anyways, as they tend to be more interested in forming new partnerships and gaining respect and attention than locking in sales.
I won't argue with the point that the music industry adds some value to some acts, as I've seen how much work has to happen behind the scenes to get a major label release out the door; but the fact remains, if every big studio was to up and disappear overnight, we would still see tons of music coming out, and the huge demand for music would lead to the popularization of the best stuff in a quite natural fashion (exactly the way it does with websites, videos, and everything else on the Internet).
This is the fundamental problem the music industry is faced with: while at one time it was the only way to produce, popularize, and distribute music, and added massive amounts of value to the market, now these things could all happen without it. Its primary current value is its ability to elevate a few acts high above the din very quickly by throwing lots of money towards promotion, and it's true that this might not happen without it.
But in my personal opinion (one I suspect is shared by many), the ability to create these overnight hits is exactly what is wrong with the music industry, and I would be happy to migrate to a situation where such things could not happen, where quality acts had to fight their way through the rest of the noise to reach stardom. Without the huge acts in the way, the smaller ones would have a greater chance of getting bigger. If this is accelerated by people stealing music online, they you know what? Burn it down and salt the earth. We'll do just fine without that particular industry, and perhaps something better and more interesting will emerge in its place.