I am once again astounded by how unreasonably effective FEP 1b12 is at federating content completely.
On NodeBB I have a list of “popular” topics, which is mainly populated based on number of posts within a given time period. For most content from Mastodon-based servers, this supplies a decent signal of a given topic’s popularity. The more people you follow, the more effective it is, but overall it’s pretty shit at getting you the whole conversation.
Enter 1b12, Lemmy’s preferred federation method. Follow a community actor and you start receiving everything that happens in that community. Replies, likes, the whole lot.
It also absolutely dominates my popular feed. It’s all Lemmy stuff now because the Mastodon stuff literally can’t compare.
Albeit the SNR is a tad lower, so give and take…
A lot of the effort I’m championing with the ForumWG deals with combating the inability for Mastodon (and other non-1b12 implementors) to federate content thoroughly. A lot of that is due to design decisions that were thoughtfully made, so this isn’t a critique, per se.
It really does highlight that 1b12 is actually quite good at what it does (federating content out), and that 7888, et al., would be a great complement for post-hoc backfill.
@julian i don’t think this is a 1b12 thing per se, it’s more that mastodon is biased toward individuals over communities, and a lot of its activities are private and not forwarded, so you never get the full context unless you follow everyone involved.
on the other hand, 1b12 says entirely too much about the types of everything. imo it misuses Group, Page, and Announce (arguably).
i also would prefer something more browsable
@trwnh@mastodon.social said in The unreasonable effectiveness of 1b12:
Absolutely 💯
That’s in a nutshell what I’d like to achieve in 2025. The building blocks are nearly all there.
@julian well, the building blocks are kinda there but they’re assembled pretty weirdly. i’m wondering how you feel about deconstructing that and maybe publishing semantically correct activitystreams documents as an alternative interface to html? it’s a more “web” approach than a “fedi” approach, but i do think it could be a good fit for nodebb at least philosophically wrt “you shouldn’t have to mirror the entire network”. a linked data browser like browser.pub could make use of that data…