It just sounds like a great idea all around.
and this is assuming that the whole "lets beg Mr. Zuck to pretty pretty please stop doing that whole business thing and start serving us for free, we'd appreciate it" thing works out.
I mean, G+ couldn't do it, and Google tried damn hard. If they can't do it at their scale, who can?
You're making it appear as through National Security Letters were public, which they very much are not.
> "Not that I'm implying there isn't already secret requests for data"
I appreciate that you consider it worth reiterating the point about national security since I had only glossed over it. But I don't appreciate you manipulating my comment to suggest that I was ignoring that point entirely when I quite clearly did raise it.
The issue here isn't federated protocols (though that is something we could benefit from in other domains) but rather the dependance on a centralised archives in the cloud. Most people simply don't care or even aren't aware of self hosting solutions. And those that are aware often choose cloud solutions because it's cheaper and/or less hassle than self-hosting. Federated protocols doesn't fix this (just look at how hard it is to set up a mail server compared to running Google Mail).
Plus in the case of Facebook you have network effects where a social network is only as good as the size of the member base.
But with IPFS, [1], (which is a federated protocol) you can "host" a HTML file or video or any content without any difficulty yourself.
I'm not sure if it would be suitable for the functionality in Facebook though (sharing stuff with friends only), sending messages, so that's part of the reason why we need more research.
The network effect of Facebook can be mitigated by introducing laws that open up competition, like the telecommunications acts, which opened up the market for telephony.
[1] https://ipfs.io/
However as much as we could argue the differences between IPFS and the plethora of other hosted solutions all day and night, but ultimately it would be a moot argument because as cool as IPFS is - and I do think it's a genuinely interesting platform - it is still only a platform for serving static content. So it's definitely not suitable for Facebook-like functionality. In that regard IPFS is more akin to Amazon S3 than it is Facebook.
> The network effect of Facebook can be mitigated by introducing laws that open up competition, like the telecommunications acts, which opened up the market for telephony.
While I can't see that happening - or at least not inside my professional life time - it is an interesting point. We've seen monopolies split up before and while Facebook is a long way from being a monopoly it would be interesting to see which line they needed to cross before the government stepped in.
The name might be slightly misleading (InterPlanetary FileSystem) but IPFS has more functionality that just hosting files. Two that comes to mind is pubsub which enables you to have a distributed publish/subscribe network to send messages globally and "dialing directly to nodes" which enables you to send/receive messages directly to peers without having to use location-based addressing.
While the primitives in IPFS are simple, they enable you to build powerful architectures, such as CRDTs. So it wouldn't be impossible to re-implement something like a social network on top of IPFS.
Disclaimer: I'm one of the developers working on IPFS