Creating a Merit-Based Music Economy
A paper about the problems with the music market structure, possible corrections, and action items required for the fix -- PDF version

Note: As of 2013, this paper is getting somewhat old, but the underlying issue is in most ways structurally unchanged. Most people refer to the realm of "compulsory" or "blanket" licensing these days as "collective licensing" -- I remain agnostic as to whether compulsory or voluntary collective licensing is preferable, as even with voluntary collective licensing there will likely still need to be government oversight (the way the Department of Justice has a continuing "consent decree" that governs the voluntary licenses of ASCAP and BMI).

There has been a recent emergence of several subscription-model music services, such as Pandora, Spotify, Rhapsody, etc. However each of these services either operates under the DMCA compulsory webcasting license, which prohibits fully interactive programming (play any tune on-demand instantly), or else is required to license tracks individually at-will, leading to an incomplete catalog. This prevents the full potential of the subscription services from being realized, and thus the consumer value is limited. Because of this limited value, the price that users will pay for such services is limited, and the royalty pool available to distribute to composers and performers / producers is limited. There is good reason to expect that extending collective licensing to fully interactive services would increase their consumer value, and thus increase the royalty pool to a level more commensurate with the value of the musical work itself.

Bottom line: This problem still needs to be fixed, and collective licensing still seems to be the way to do it, so far as I can see. There are others with different opinions, but none of them have argued convincingly enough to change my opinion here.