The federation is a network of servers configured to share wiki pages in a standardized json format under a creative commons license. Our focus in this pages is to configure a Server for participation in this network, a subnet of the world wide web.
We consider the risks of building federated wiki or anything else permanent on something as fragile as information.
However, the term means something different than you may imply. Let's compare it to PeerTube. Each instance can follow the indexes of other instances. A visitor can search for any video from any instance (given it's on the allow list), and the search results will be shown in a single unified list. You can click on a search result hosted on any other instance and the video and its comments will be fetched and streamed through the same interface still accessible through the starting instance. A visitor can register a user account and follow user accounts or channels hosted on any other instances, getting notifications about new content from them. You can comment under videos posted on any other instance with your account. Visitors can list videos posted by a given account on various instances. Cross-server compatibility was also a thing a few years ago, but they have since refined on their model regarding this (it's not strictly required to be considered "federated" as long as all instances running the same software federate among each other as equal peers). Bkil (talk) 10:53, 27 March 2024 (UTC) Well, it's not federated (yet ?) as intended in PeerTube or Mastodon. If you have a more correct term in English for meaning many independent servers that can be harvested by a shared meta catalog, you can switch to it --PanierAvide (talk) 11:03, 27 March 2024 (UTC)
~
Does federated wiki deserve to call itself "federated"? A primary goal a dozen years ago was to exhibit "wiki nature" with no centralized control. We do rely on DNS but have demonstrated that we can share numeric ip addresses by our own federation and thus bypass that central authority. We reject TLS for the same reasons. When I read of other schemes they all are struggling to find some robust form of push which is the only fast path to influence. I am suspicious of their motives. c2 
http://found.ward.bay.wiki.org/infected-with-roi.html matrix
a concept mentioned by Ward in passing, but one that speaks to a deeper truth of the creative process.
An inventor invests time for a return. An investor contributes wealth for a rate of return. The same formula, R=rt, except the investor has time t in the denominator and that makes all the difference.
~
The question is, how important is push? It does for IRC where the message turnaround is on the order of seconds (sometimes milliseconds on busy channels back in its golden ages). At the same time, delays of hours are acceptable on forums. On a documentation site, I (bkil) don't see why subscribers would expect updates more than once a day, so polling would be just as good for all intents and purposes. matrix
I (bkil) would add a twist to polling, though, to make it scale. Each peer should sign all of their entries (and indices) and mirror some or all entries on friendly (whihelisted or subscribed to) peers. Then readers and syncing peers could look up data through the swarm and not have to poll the origin as long as said content and indices are properly seeded. Incidentally, PeerTube had implemented automatic reseeding of friendly instances for a similar reason a few years ago. I don't view theirs a perfect architecture and regularly pick at it, but they got at least the basics right. matrix