This is a subfile of the primary AI4Communities post, and won't make any sense unless you read its parent first. It explores how the AI4communities idea would look within the ActivityPub-powered Fediverse.
(Notes: This is an early draft. As explained in this newsletter edition, I am publishing these early versions as I develop my thoughts in the hope that constructive comments will help me finish the post. More version control in the footer.)
My manifesto posts of 1/1/23 reflected some knowledge of the Fediverse, which I'd been tracking for a few years (see my Hub's Fediverse overview). In fact, I'm not sure I would have come up with these ideas if ActivityPub hadn't shown that decentralised social media was at least possible. My early versions of the AI4Communities post, also reflected that bias, which is why "cozyweb community" maps so easily to "Fediverse instance".
However, as I researched the more modern decentralised protocols for the AI4Communities post (see AI4Communities on the ATmosphere, AI4Communities on Nostr), I came across an introduction to Nostr from one of my favourite thinkers in this space, Gordon Brander. While I'm not yet convinced that Nostr is the solution, Nature's many attempts to evolve a Nostr provides an excellent explanation of why the ActivityHub-powered Fediverse is flawed: "Federated networks become oligopolies at scale", due to general forces seen everywhere: "airline routes, power grids, trains, banks, Bitcoin mining, protein interactions, ecological food webs, neural networks". It happened to email, and now it's happening with the Fediverse, where calls to defederate Threads (which was 10x the rest of the Fediverse when it arrived) were ineffective.
Nevertheless, the Fediverse has something the more modern ecosystems (probably - tbc) lack: a diversity of apps working and communicating across a decentralised architecture. So despite my own bad experience (see All my toots gone), as well as the mild toxicity Fedizens have towards other protocols, it would be a mistake to not seriously consider its suitability for supporting AI4Communities.
The main AI4Communities post summarises various categories of AI service within these ecosystems, so this section examines how the various categories would look in the Fediverse.
The current draft of these services in the AI4Communities post is very Fediverse-oriented, referring constantly to "the village's inhabitants", where village maps more or less perfectly to Fediverse instance. However, while these services could augment the Fediverse offering enormously, they might not be possible to implement.
When it comes to content discovery, on the Fediverse you only get three reverse chronological feeds:
None of these feeds include content from villages which your village does not federate with, which is widely considered a feature, not a bug (but could limit the effectiveness of AI-supported content discovery, as explored below).
With AI4communities, an instance could employ content discovery AIs to proactively pull in and surface content of collective interest to the village's inhabitants from across the Fediverse. As that AI would be trained on the inhabitants' interests and explicitly guardrailed by the community's values, it should be an interesting, high-value feed.
And if your tastes don’t match the village’s, you can move to another village. Balancing an individual’s feed parameters with the village’s content moderation rulebook should be a fun problem to tackle, but far from insuperable, as each village can find its own balance. Which brings me to...
Fediverse content discovery and moderation services are in a sense two sides of the same coin: while content discovery AI services pull in useful content from across the Fediverse, they would need also to keep unwanted content out.
In addition, however, content moderation AI services would also operate within the instance, as described in AI4Communities post: improving the discussions and collaborations within the instance's local feed. As the local feeds is composed of the community's authored content, this is also where these AI services would find a lot of their training data, and where the community would discuss edge cases.
The main problem is that an instance's content discovery algorithm cannot pull in content from an instance it is not federated with. The main opportunity is that AI4communities could make it safer for instances to federate with more servers, and reduce their need to push the nuclear "defederate" button, opting instead to harden their algorithms against specific posts and users.
to be developed
In AI4Communities on the Fediverse:
The store will provide an array of products and pricing schemes, including lower-cost subscription plans where the village's data, suitably anonymised, is used to affray the costs. Other villages may select higher-cost plans where the village does not share its data. Some may charge some or all members, and/or display ads.
The diversity of both funding schemes and AI plans means that one village may form a data union, while another sets up a for-profit instance and a third lets users choose between paying a subscription fee or seeing ads.
(caption) Different Fediverse communities offering users different experiences, including AI services sourced from multiple AI marketplaces and individual suppliers.
What I don't know yet:
This is one of this wiki's pages managed with the permanent versions pattern described in Two wiki authors and a blogger walk into a bar…