I know this is an alpha release, but it's invitation only with the invitation being "help us shape the future of our network"
All of which comes across as being much less open and decentralized than the current web.
Even if the technology itself is more distributed than the common standards we're all using now, this sounds more like a walled garden than a truly open network.
Yes is it all closed. Look at bittorrent-sync. Totally cool, but closed and controlled.
I don't see that working here, other than being used for porn and piracy.
The BitTorrent protocol will remain open. This is a private alpha release of a version of it that is more focused on web content. - Project Maelstrom team member.
Ephemeral P2P site hosting has potential, though I disagree with the route Project Maelstrom is taking. Another approach would be to try an in-browser implementation, using:
* WebRTC for the P2P "glue" (i.e. webtorrent[0]/instant.io[1])
You basically just package everything (images, JS, HTML) into a single DHT resource, and share the eph:// URI. For long-term hosting, setup a webtorrent-compatible seedbox somewhere in Europe.
I've only gotten as far as static P2P site hosting, but if I were to speculate:
With a distributed database and a rethink on isomorphic JS, you could build fully-auditable "aether-hosted apps". Like cloud hosting, but more buzzwordy! It'll become even more powerful (but less auditable) if homomorphic encryption ever becomes a thing.
For such a DB, perhaps a P2P document store similar to CouchDB/PouchDB, with aspects of context-aware networking. The user auth table could include public keys for authorized users, and handle modifications through a signed changelog. You could probably get pretty far through a signature chain (users delegating users), though at some point it'd be so collision-prone that a superuser would need to regenerate the database and re-sign the contents. I'm not really a DB guy, though; there are obvious situations where such an approach is unsuitable, plus there might be a simpler DHT-like way to do it.
Short URL's could be handled using a proof-of-work-based DDNS system (Web Crypto API for PoW, WebRTC for P2P). A more simplistic approach to URI->URL is simply to convert into human-readable strings[4].
Thanks for the thoughtful comment. We've thought about a lot of this. This is our first stab at trying to push a lot of these thoughts and ideas forward.
All of which comes across as being much less open and decentralized than the current web.
Even if the technology itself is more distributed than the common standards we're all using now, this sounds more like a walled garden than a truly open network.
AOL with Tor. Yay.