The story of internet centralization is one of network effects. Even though the internet connects every computer on the planet directly, finding people to interact in the ways we want is difficult. We post things to Facebook and Twitter because that's where our friends and like-minded strangers are actually looking for things.
To fight this, we tried federation. We tried building social systems that allowed people to choose their own provider but participate in the same network, but federation hasn't worked since email. If you take a long look at email, you'll see why: federated systems can't keep up with the pace of evolution that we see in successful social systems. The design-test-iterate loop that design thinkers use to build products people want doesn't work when the rest of the federated providers don't keep up, and they never do.
Where federation failed, I believe decentralization will succeed. There is an honest-to-god renaissance happening right now in decentralized systems, and the tools to continue the internet dream are reaching maturity. Core to this renaissance is our new ability to use blockchains to establish universal cryptographic identities with no intermediary—no federated server to hold back the tide of progress. When you want new features, you don't switch servers, you switch clients. They all speak the same peer-to-peer protocol using the same portable, blockchain-based identities. This has a very useful side-effect: systems that don't rely on servers never shut down as long as people keep using them. Decentralized systems are forces of nature.
This combination makes network effects a public good. Second place companies in network effect industries today have their own private networks that just aren't as useful. It is in their best interest to make their networks as large as possible, and treating the decentralized network as part or all of their network achieves that. It creates a network used by all the competitors who aren't in first place, even as those competitors die off. I think the immortal network used by competitors will over time become as large as the front-runner's. At equal size, the open network is more valuable than the closed one—the open one can be built wholesale into any other product that benefits from it. If you want, you can use a ride-hailing app that uses your social network connections to share rides with friends of friends instead of complete strangers, and anyone can build that app when the ride-hailing and social networks are decentralized.
When our industry gets great at creating decentralized application protocols, it will no longer be feasible to build a business on privately controlled network effects. It's not a defensible business anymore. Users will be able to choose which clients they use instead of being forced to sacrifice their data to a company to benefit from its network.
The internet dream is back, y'all. If you want to help rebuild it, come join the folks using Ethereum, IPFS, and related protocols to build networks we can't control.