As this article alludes to: the heart of email's longevity, the thing that prevents it being closed by a single entity, is being a classical IETF protocol: federated, decentralized, open and interoperable.
The first two of those properties arise from being based on the DNS for SMTP endpoint discovery.
This is why every protocol needs to specify that it uses the DNS, and how.
And that is why I get so worried that the draft HTTP/2 editors so steadfastly refuses to do so.
It would be useful to draw a dependency graph of IETF and W3C protocols that characterizes each one on their degree of divergence from classical IETF values. Identify the point(s) in time where history took a wrong turn, revert to that point, fork and start over with new protocols. It can't be any less disruptive than new protocols which have a non-classical values.
Are there modern use cases for NNTP? Could UUCP be used for sneakernet or bluetooth intermittent mesh applications? What's the verdict on WebDAV, CalDAV and CardDAV as neutral protocols for sharing data, contacts & calendar?
Yes - I have used NNTP as an eventually-reliable pub/sub message bus in an unstable mesh environment. The specific use being transport of per-device usage/billing & monitoring data.
The implied Gossip algorithm of the server-to-server subprotocol was well suited to the intermittent availability of links in that particular network. Messages (articles) were PGP verified. NNTP is channelized by design, and servers like Diablo have built-in expiry management. It's also easy to monitor with a lightweight news reader.
This was only a few years ago. I was pleased with it, at the time - simple to build from readily available parts.
We need a neutral, replicable client-side data store with standardized interfaces for automated analysis/filtering/reading. This could be an offline archive of docs, web browser history or online/cloud bookmarks, along with _private_ annotations about content and authors (e.g. killfiles, quality ratings).
If the datastore is open (standard & source), there could be healthy competition among proprietary apps to use the datastore, but not control it. NNTP could predictively replicate (via home wifi) a subset of this cache to mobile based on personal calendar, reducing public network usage.
What's good about this approach is that it would surpass the usability of mobile apps + public cloud, thanks to the private cache reducing the impact of public network latency. With neverending hijinks on net neutrality and web standards, it may be worth exploring NNTP pub/sub between home server/laptop and mobile devices.
We are ways off, but my personal goal for http://hood.ie is to be able to allow people to build gmail/gdocs competitors without having to do any heavy lifting. We could use your help, though :)
After the death of Google Reader, I've switched to news.gwene.org, which operates a RSS to Usenet bridge. It solves a lot of the problems that other RSS readers are still dealing with, since the Usenet people figured this stuff out back in the 80s.
The first two of those properties arise from being based on the DNS for SMTP endpoint discovery.
This is why every protocol needs to specify that it uses the DNS, and how.
And that is why I get so worried that the draft HTTP/2 editors so steadfastly refuses to do so.