@quinn I want to second this, and to add a note of (relative) optimism about variable pricing. There are at least two well-known models of social media where the users pay. The first are MMO games, where much of what users pay for is to preserve the space where they can interact. (Single-player EVE would be a fraction of it's current size.)
That model requires a degree of indirection -- users are told they are paying for the game, long after work on updated content shrinks to a fraction of the original work by Blizzard or whomever. Ello seems to be planning on doing that with 'Pay for features' model.
The second example is single-source at the moment: Meetup. Many years ago, Meetup faced exactly the problem Ello has today, and which you describe so well. There were only three places Meetup could get income: the meeting places, third-party advertisers, and users.
They wen with #1 first, trying to get money from the businesses where the Meetups were held, so local coffee shops would split some of the extra money in the till when 17 curling afficionados met in their establishment. This failed when those the few businesses that agreed to pay also demanded the right to control the users in one way or another -- excluding competing coffee shops as options for Meetups and whatnot.
Number two was advertisers. Scott and Co thought about that for a bit, but decided they wouldn't be able to live with themselves, for all the reasons you describe.
So #3 it was, even though a decade ago, charging the users was almost universally thought of as crazy. Meetup decided, though, that only if the money came in from the people who valued the service could they keep designing the Meetup they wanted to design. (The principle, of course, is general: Form follows funding. Services assume the values of the people who pay for them.)
The other key thing that Meetup settled on (and I was an advisor back then, and I'm as proud of this decision as of anything I've ever helped a company get to): Charge groups, not individual users. Having a Meetup group cost money, but the people who sponsored the group could decide how to pay it. If it was your group, you could be a macher and pay for all of it, and offer participation free. Or you could charge the regulars but let newbies in free. Or you could split the small cost N ways for N participants. And there were narrower models -- eating groups just threw exta into the tip and so on.
I am skeptical that selling features will work for Ello in the long term, becase really, the only first-order feature that anyone cares about on a social network is "Where my dogs @?" If there are eleven people you like, or would like to like, on a crappy service, that beats there being eleven thousand people you don't know or want to know on God's Own App.
I am optimistic, though, that those of us in your upper quintile can be charged enough to allow for others to join free without the service either collapsing or having to go into hyper-growth to satisfy promises they made during investor storytime.