I'd argue that a 3rd party having the personal details and communication logs of 2 billion people is a massive problem. Privacy, governmental data requests, accidental information leakages, profiling, job applicant scrutiny, fake news, spam, data misuse by employees etc. Or even worse, in times of war.
If the store is decentralized, then any party could be equivalent to a third party, and therefore anyone has access to your data.
How is that going to work then? Every single company holding your data in a central place is mining it today and selling the information to advertisers.
On a distributed network you replicate data only with people you choose. Since you're not going to add a government (or an ad company) to your friend list, they won't have your data. Private communications are encrypted and flowing directly to the recipient, so there's no port that agencies can tap into to listen. There won't be any big data to analyze, since data is spread across the network. Backup is easy, it's just a folder. Fake news never gets promoted, since there are no paid promotions.
Spam is a non-issue since there are no advertisers. There's no way to get into your feed than through your friends. There's no server or central storage, so employee misuse is not an issue. And you can have as many identities as you choose, there won't be a mandatory real name policy.
I could go on.
That doesn't mean it has to be that way in the future though. There are centralized social networks that are charging money for their use, and they are not selling to advertisers nor are there any.
With decentralized, you still have the same issues (I don't really know how you solve those unless data transfer is truly peer-to-peer which I think causes discoverability and similar issues to usability), but at least the scope of them is inherently limited and you can opt in/out of the source based on your trust of it.
It doesn't matter how it works, the fact of the matter is that while logistically third parties are typical in today's environment, they aren't implied by centralization.
Does Amazon's s3 sell your data to third parties? No. It's still centralized though!
> Since you're not going to add a government (or an ad company) to your friend list, they won't have your data.
Except when people use things like hashbase.io because they can't afford to run the server program 24/7 themselves. Then the hashbase.io analogous service sells your data to a third party, and congratulations you've got a system that has all of the lose of distributed systems, with all of the lose of centralized systems.
If Alice wants to communicate with Bob, she shouldn't be obliged to trust Facebook too.
It's odd that we got used to the idea that "(1) registration, (2) strong password creation, (3) username selection (in case of conflicts), (4) email verification link, (5) login" is somehow a good user experience.
Cycle.js and xstream are amazing, as an aside. Thank you.
Eventually, the client will probably have a “that's also me” button, which you'll press on both devices to confirm.
Facebook isn't an alternative to email if my grandma can't use it.
Computers aren't an alternative to telephones if my grandma can't use it.
Telephones aren't an alternative to mail if my grandma can't use it.
etc.
---
It'll get easier and simpler, I promise.
Just because things have gotten more complicated and people have used them doesn't mean that people will start to use any complicated thing.
For example:
Google+ / Google Wave isn't an alternative to email if my grandma can't use it.
GPG isn't an alternative to talking in-person communication if my grandma can't use it.
... etc
Yes, you have valid examples where things that are relatively complex did become popular, but there are many many more examples of things that were more complicated and failed.
Just because scuttlebutt is complicated is not a reason it will succeed, but is indeed a valid reason it might fail.
Arguably it isn't FB if regular college kids can't use it. If they can subsequently onboard less tech savvy people, so be it, but that could be a future goal.
The idea that centralized storage is the problem masks the actual concern. There is nothing inherently wrong with centrally stored data. There is a problem is when it is locked down by a 3rd party, and/or you don't control how it is used.