Significantly, high-level web technologies are arguably more privacy-aware than their underlying stack, supporting end-to-end encryption (although to be fair, centralisation is a significant problem at both levels, likely out of purely practical concerns). I don't think malice played any part in this development though -- it's just another accident of history: the network stack got "good enough" to ignite the ecosystem well before privacy implications became clear, such implications weren't so immediately dramatic to force a complete overhaul of the significant amount of deployed infrastructure, and here we are.
The End-to-End principle behind TCP/IP should be considered a must-study in anti-authoritarian thinking.
It's the job of all layers to preserve decentralization to the extent they can, and in this regard HTTP(S) fails miserably.
The decentralized architecture that was the original plan for the "internet" is a worthy goal, not simply for anti-authoritarian or privacy reasons. Or home shopping possibilities (Baran paper predicting this).
How about something more general, like resiliency? A network that is opportunistic and always "up".
Municipalities are building resilient local networks for disaster recovery purposes. We should be building them for all uses, and interconnecting them.
A true "internet" does not depend on nor is subject to centralized control. It only requires cooperation.
http://www.cluetrain.com/book/apocalypso.html
I don't care to debate the topic. Just quietly pointing out an alternative view.
Here's a brief proof. Assume that TCP was never conceived. Would most of what we have today still be around?