It's not unthinkable that Facebook will be the top destination 20 years from now if they are (1) successful with internet.org and (2) first-to-market with a viable virtual reality product.
Does Economies of scale outweigh the economic surpluses gained from a more competitive economy? Well, that's something one should ask an ivory tower Economist.
See also: Google's increasingly long list of discontinued products, many of which were acquires from burgeoning young businesses.
Ask an economist and you'll get different normative answers of how we should craft policy, but the positive answer will be the same across the board: it has happened with clothing manufacturing, publishing, radio, TV, food production, etc. and is now happening with the internet. A nascent and initially fragmented medium/industry attracts investment and develops standards; large conglomerations begin to form and M&A smaller players to gain market share, leading to a saturated market of heavy-weight players.
And PCs - it started out as a decentralized market between Apple, Commodore, Tandy, Atari, Osborne, etc, but then centralized under IBM before being decentralized again with IBM clones like Dell, Compaq, etc. That in turn forced the centralization of the OS market under Microsoft (driving out Apple, CP/M, Acorn, Atari) and microprocessors under Intel (driving out Motorola and MOS technologies).
I think it's more accurate to say that the point of centralization moves up and down the value chain based on technological developments, market deals, and corporate strategy. When things get too centralized, a savvy entrepreneur will come in as a small-time complement, then find a way to encourage competition and commoditize his complements, riding the wave to a very prosperous monopoly. When competition becomes too intense in those markets, the bigger players will start eating the smaller players, and eventually the market consolidates.
These networks with small world properties strike a highly pragmatic balance. They are far more robust to insult when compared to centralized networks (though not so much as random) while having much more efficient propogation of information than in more random networks.
However, as technology progresses, we might reach a point where "distributed" can be sufficiently efficient.
no self-respecting third-world intellectual is going to be satisfied by a catered whitelist of websites they can access. to me this seems like an attempt to repeat what we did with confusing "the web" with "the internet" by having third world people confuse "facebook" with "the web"
It doesn't matter what intellectuals are satisfied with, the internet (like TV and Radio) will be driven by what the mass-market consumer wants. If they can get all they need from Facebook for free (running a business & talking to friends and family), then why pay more to develop a network that they don't want?
[1] http://qz.com/333313/milliions-of-facebook-users-have-no-ide...
so what I predicted has basically already happened. sigh.