1. A model of the client/access provider relationship is that the
client subscribes to content services and the access provider
exercises them. There could be a request-by-request mechanism
for the client to specify which services should or shouldn't
be applied to a request, by embedding proxylet content in
his request. Well, that's cute, I hadn't been able to think of
a reason for request proxylets before. Getting browsers to
support this is a header of a different color, though.
I agree with Hilarie here - I'm not sure that we can reasonably get new
headers on request. If there is a subscribe/unsubscribe concept with
features, I think it will have to end up in some sort of lookup mechanism -
along with identifying the user.
For these features (identifying the user and figuring out what they can and
cannot use) I've always assumed they'd be implemented over top of OPES.
E.g. a proxylet firing off on the request to get the ID and set all
preferences by inserting a header line or cookie that then flag other OPES
rules.
I have actually seen a demo of a feature like this on iCAP from the Zack
Network people.
2. There are business models that induce the content-access-user
chain to operate on behalf of the user, despite the rapacious
practices that we've become accustomed to seeing. I can't
say that these models will dominate the market, but we see
that it's difficult to predict what tomorrow's web pages will
bring.
3. What's going on here is that the "authority" is getting a
finer grained definition of content and the control over
it. This gives more opportunities for doing business.
I think the model brings opportunities for correct control
of content into what is becoming an increasingly complicated
distributed system. A richer model for control is important
for scalability.
Excellent prophesies - I'm looking forward to being able to say "Yeah, I new
about that years ago..."
-Rob