Chiming in to say I agree with this viewpoint -- I think conflating centralization as the only way to provide the convenience, indexing, free hosting, etc is definitely what happens when people say that decentralized solutions can't work.
Two big blockers that I rarely see discussed for the decentralized future coming to pass:
- (D)DNS for the 1+ persistent devices per/user (IPV6 support/rollout basically)
- Lowering the barrier to entry for running a server on a users' behalf
If we really want the internet to be a web, we need to make it easier to run servers that act on behalf of users/hold their data, that they control.
If you'll indulge me, I've had in the back of my mind a project for a long time, a - Document management and storage
- Backups (connect your local "pipe" to a backup storage provider)
- Chat/Messaging, SIP + Video conferencing
- Local networked/mesh services spanning from blogging, social networking, ecommerce, whatever.
I'm very much of the I-just-want-to-write-software so I preoccupy myself thinking of what tech I'd use to make it happen, but I think this is the kind of project that people need to really embrace the internet as a sparse group of interconnected nodes.
I also think that humans are ready to accept this kind of structure -- we already have it! People can understand that if you want a storefront to be up all the time, you need to have something run it all the time, the power to run it just isn't in their hands yet.
Of course, startups are hard, hardware startups are harder, and consumer product hardware+software startups are pretty hard, so it's much easier said than done. Decentralized systems absolutely can work, but the amount of work, fore-thought, and careful construction with consideration for generating network effects and making it easier for people to run and use -- basically good product design -- is just missing in the open source world, because it's hard.