A lot of the ideas presented on this thread are less applications for federation and more applications for blockchain of some kind. For example, wikipedia or uber eats replacement. Before you blindly downvote me for this suggestion, let me explain why.
In federation, you have servers which talk to each other. Users own their own accounts and there are multiple repositories of information. Lemmy is a repository of links and comments, each lemmy instance has its own repository. Mastodon is a repository of tweets, replies, and DMs. This works great. Everybody makes their own repository of information, and users can subscribe to any repository they like. They can also, via federation, access other repositories and “pull” or “push” data to them. That last sentence is the magic of federation you don’t get on platforms like Facebook. ActivityPub and federated platforms solve this problem of provider lock-in, at least partially.
This fediverse is not great when you need to establish a single repository of information that everybody in the network uses and is in sync for all users. Because it has no mechanism to arrive at consensus as to what should go into that authoritative repository. Even if all participants can be relied to act honorably (something the internet rarely provides), there will be disagreements about what should go into that repository. Edits may come in at different times, how do we resolve which edit goes “first”? Because it may make the second edit irrelevant, etc. Federation can’t solve this problem. ActivityPub can’t solve it and Nostr can’t solve it. But…
This is the exact problem blockchains solve: how can you establish a centralized repository of information (ledger) and administer it in a decentralized, P2P way where you can’t trust all participants to honestly participate? You cannot develop P2P systems which maintain a centralized repository of information without blockchain because no other P2P system has been able to solve this problem. There is no other mechanism of arriving at consensus and prevent sybil attacks.
Wikipedia? Centralized repository of information. Uber eats? Centralized repository of foods available, drivers, customers, and orders. eBay? same. And by the very nature of blockchains, they can also have an economic layer built into them which provides a means of exchange among participants. Useful for an eBay replacement, maybe less useful for a wikipedia replacement. Those means of exchange (“tokens”) can be used not just for transfer of funds, but also for things like building/scoring user reputation and incentivizing specific behaviors, especially if you want to incentivize behavior that is contrary to a user’s normal economic interest, such as providing a subsidy for restaurants on Uber who use more expensive, but more sustainable food packaging.
The non-P2P solution is to trust the administration of this centralized repository to a trusted authority. We trust wikipedia to administer articles and decide what ultimately goes in them. That system works fine for wikipedia, I’m not convinced we need a decentralized version.
There are many blockchains with various technical attributes which may work better or worse for solving these problems. They may use proof-of-work, proof-of-stake, etc. Some are more decentralized than others and have features like censorship resistance, privacy, smart contract, etc. But they solve this exact problem.