It's basically an open spec for something similar to the Dropbox Datastore and Files APIs, with open-source implementations that anyone can host. The official client library has some fairly robust offline-first sync'ing capabilities, and also happens to be relatively backend agnostic, being able to act as an interface for either the remoteStorage API itself, or the actual Dropbox Datastore and Files APIs, as well as the Google Drive equivalents (although the latter two are under heavy development and probably not production-ready). This allows your users to choose the storage service they want to use for your app without you having to implement all 3 interfaces (and if they choose remoteStorage, be able to host their own storage instance or use one hosted by someone they trust).
I'm prototyping a distributed app right now that uses remoteStorage to enable multi-client sync and simultaneous logins. I believe this approach strikes a fairly good balance, giving users the privacy and service mobility offered by distributed software, as well as the usability and convenience offered by centralized software.
The only thing I feel they're sorely missing is a transparent encryption layer on top of the Datastore API (and possibly for Files too, though I'm not sure how feasible this is). I ended up implementing this part myself for the app I'm working on, but I feel it's unreasonable to expect every developer to reimplement something as essential and error-prone as cryptography.
While I appreciate the attitude so far I don't see too many applications redecentralization might be useful to. Yes, it's more fault-tolerant. Yes, it would for the most part avoid mass surveillance.
However, decentralized systems are also much more complex to build. So far, decentralized approaches in most use cases provide too little value for most users to really make sense economically. Privacy, fault-tolerance are certainly very important aspects. Unfortunately, in general most people couldn't care less. Despite some public outrage the NSA, GCHQ and the BND are still very much in business and doing as fine as ever.
Are there any applications particularly amenable to a decentralized approach? Quite likely. Is a decentralization a generally applicable model of the way we do our networking? I'm not so sure about that. Having a decentralized social network for instance would be some achievement but what benefit would it provide to the not-so-privacy-minded?
For server operators it makes sense to toss more processing onto the client-side as it lowers cost.
The FreeNet model is just ahead of its time: fully encrypted and fully decentralized using separate apps/plugins on the client-side to make sure it's always decentralized.
One issue with client-side decentralization has been how to pass things around to other clients when your main machine/box/mobile device is off. So you need carriers/exchanges that pass them around; in SaaS the client and data are stored on one server; with decentralization the data is stored on the client and on an exchange. Instead of being a server provider, you'll be a client provider coupled with a data host and you can just outsource that to S3 or some other data host provider.
I hope you're right! I've wanted this for the last 5 years and it seems that more and more good solutions are popping up, sandstorm.io being the most prominent to my eyes.
There are hardly any alternative networks? GSM/3g/4g are illegal to run on an ad-hoc basis most everywhere. Which leaves cabled networks (which are generally not legal to set up as metro/internets -- or even link freely between different buildings if you need to cross community owned land) and wireless mesh-networks.
I think both are interesting, but I don't see them forming a global interconnected network capable of delivering symmetric gigabit to 10 billion people (not that the Internet of today does that either -- yet).
I suppose local 802.11-whatever community networks might be connected in metro-/internets with high-bandwidth point-to-point links -- but it seems likely that they'd end up connected to the Internet -- perhaps on the ipv6 Internet.
Either way, hopefully internet access will continue to evolve like a utility, like electricity and water.
Interference with 'official' networks, besides that spectrum use requires a license in just about every country except for some very specific low power unlicensed segments of the spectrum.
