AT&T was split on geographical lines, so maybe FB could be as well, and the regionals would all have to interoperate with open protocols that would allow for regional competitors.
AT&T was split on geographical lines, so maybe FB could be as well, and the regionals would all have to interoperate with open protocols that would allow for regional competitors.
Personally I would want to be able to host my own tiny Facebook instance that I control. Ideally I could even invite friends and family to share my ad-free self-hosted instance with a chronological newsfeed setup.
The right way to think about this is what kind of splitting would produce the most competitive landscape. We have a really successful example in the way that ISP competition has played out in various markets.
The US has regional ISP near monopolies in gigabit fiber, since the infrastructural outlay is so expensive. This is analogous to the moat FB has from building a vast and deeply connected network graph.
The ISP outcome has been that the US mostly pays extraordinarily high prices for substandard service.
Other countries have avoided this trap by regulating the enforced splitting of the network itself and the service provider. Since building out the physical network (the fiber, or the friend graph) is extremely hard, great that is a natural monopoly. Maybe you might get a couple of huge players, but the barrier to entry for new players is nearly insurmountable.
Each network graph / fiber provider is required by law to offer access to their network to third party service providers who you can contract with to provide you internet service over the network.
Japan for instance has this setup, with a handful of players that have built national (or regional) fiber networks, and when you buy service, you have a split bill. Part of it goes to pay for access to the physical network. The other part is to pay for an ISP that is routing your packets over the network.
I think this is the proper solution for FB. Break the company apart between the network graph provider, and the CLIENT layer. Let third parties provide their own friend graph client with features that people want.
Then you would see a flowering of competition as people feel free to try the client experience they prefer. * How about chronological news feed? * no ads, but paying your interconnect cost by making you pay a subscription * fine grained control of how many stories you ever see from specific over-talkative people. * delivering a subset of your friends' stories only on keyword-matched topics / hiding all stories that match specific topics.
I am not hopeful the technologically illiterate people writing the decisions will understand this though. Geographic splitting is nonsensical for a graph that is inherently global.