I don't see much in your message with which I would disagree.
> The data owner (the content provider) is often not the owner of the
> distribution network/system.
right, and I see nothing wrong with that so long as the distribution
network is distributing the material with the consent of the content
provider, and according to whatever instructions to which they might
mutually agree.
part of the point (which I'm struggling to articulate even to myself)
is that the system as a whole needs to behave predictably and there
needs to be a clear division of responsibility between the various
parties that handle the data. otherwise, when something goes wrong,
(say, incorrect, altered, or stale data is delivered to the consumer)
there will be no clear direction as to which pieces are at fault
and who is responsible for fixing it. (yes, I realize that such
problems often occur even in well-designed systems, but we should
try to minimize them)
Keith
p.s. note that the content provider is not necessarily the owner of the
data in a legal sense, though in general one hopes that the content
provider has permission to distribute the data. still, there's a subtle
difference between roles that should be specified by the protocol
architecture and those that are specified by laws - the protocol
architecture should be concerned with reliable and predictable
operation (and hence, for instance, that the replicas of data
that are obtained by querying the content provider are actually
those intended by the content provider) but the protocol should not
concern itself with whether a content provider has the legal right
to make that content available.
This archive was generated by hypermail 2b29 : Thu Nov 18 2004 - 11:21:29 MST