Many services rely on individual downloads and, worse,
on registrations.
These services have thousands of files which must be downloaded
individually with a few button clicks each. If I wish to have
a complete collection (and why wouldn't I), I certainly would break
my mouse hand.
Well if this is the only issue, then why not just maintain a central
*list* of urls? The user downloads the list, as a textfile, feeds it
into a wget or curl script, and it downloads all the tracks from lots
of different places.
This avoids the requirement for a single large server with large
mirrors (which *will* be a genuine obstacle, in the early days of
getting this project off the ground).