Seems like a pretty misguided point of view. Something tells me a sizable amount of the facebook crowd is comfortable downloading torrents, I'm pretty sure that counts as distributed/decentralized. As far as free software, is your mom unable to use any website that uses apache, ngix, php or ruby? Almost everyone has a broadband router that is almost entirely open source. My Mom and Dad both use an open source browser, what do yours use?
Is your point that only companies worth $50bn can build usable web apps, or that anyone that doesn't want to sell your data can't do UX?
That's from 2008 announcing Pirate Bay had hit 12 million concurrent peers on their tracker. That's not total users, that's total online. Take that number and multiply it by 5 to 10 to estimate total users, double it to account for the rest of the global trackers (low?) and increase it by some percentage to account for 3 years of growth and you're in the hundreds of millions of users range, possibly as high as 300M or more.
I'm not a crazy bittorrent fanboy, in fact I'm quite the opposite, but I think it's clear that the number of users of a decentralized system can far exceed any "rounding errors" of facebook's numbers.
By having 12 million concurrent peers on their tracker, aren't they pretty much de facto centralized even if the protocol is theoretically decentralized?
If The Pirate Bay disappears tomorrow do all of those users just switch over to another tracker? Very likely not. They'd just wait around to hear about the next Napster->Kazaa->LimeWire->Pirate Bay from their tech savvy friends, remaining mostly oblivious to the technology being used under the hood.
The pirate bay hasn't actually run a tracker for a year now or so. BT uses a system called DHT (and maybe something else) that lets the peers handle the tracking themselves. The peers do need to boot strap off of something, but it's very minimal afaik.
Even when they did use a tracker it's probably more reasonable to think of it as a directory server more like LDAP or such. At their base even classic decentralized protocols like SMTP rely on DNS as a directory, IP on routing tables etc.
While DHT is indeed decentralized, a central tracker is still preferred since it can quickly give an authoritative list of peers (rather than a probabilistic one).
Private sites tend to disable DHT anyways, since it's much harder to detect cheating without central trackers.
The important thing is that trackers are easily decentralized, since they're a per-torrent URL.
I think you are massively over-inflating your 300M estimate. The typical usage of bit torrent is to leave it running 24-7 to download the 300 torrents you've got queued. I would bet it is is far lower than 5-10 times and I really doubt it has tripled since 2008.
If so, so many people don't know what a browser is, you can only image how smooth a distributed social network would have to be to work.
Bit torrent works because it's three steps: 1. Download a bit torrent client, 2. Search Pirate Bay, 3. Click on the torrent. Occasionally, you hear the failure mode: why won't this stupid thing download? The only people who say things like "More people need to seed this" are tech literate.
Bittorrent is different. You don't care who the other nodes are, all you care about is you're connected to some nodes, and the client takes care of that for you.
Like others said, the only way a decentralized facebook would work is when everyone uses one (or two) central nodes (site) which will most likely be run by something like Google or Microsoft, and the end result will be the same: big company has access to all your data.
The "Like" button probably wouldn't work though. If you're some external website, which "like" button will you embed? Google's like button? or Microsoft's?
Is your point that only companies worth $50bn can build usable web apps, or that anyone that doesn't want to sell your data can't do UX?