Can we artificially force people to read...

Okay i can stop there, thats never happening.

Was gonna write about forced understanding of context and how the system sees the wide context gap between the thread youre replying and your own

I feel like this is the wrong solution. Feels too heavy handed and also is too all-or-nothing to work. Maybe algorithmic silo-ing is just a relatively easy way for big platforms to prevent infighting that doesnt solve the root problem? Idk, that implies theres a root problem to solve

How can people trust that the system is helping them, and is not censoring them? Idk

Would preventing the out-of-network interaction to begin with be better? Probably, though it needs widespread adoption

I think it can still work (kinda) by implementing it on both sides, so even if a user from outside of your algorithm-defined bubble interacts, the notif will be suppressed somehow

Its also not very decentralized of you to agree on one single network splitting system. But the alternative is endless conflicts from "dumb algorithms" that mix oil and water.

it doesnt really work on new networks because its not large enough where it becomes a problem. existing networks also doesnt work until every raw firehose consumers shuts down and switches over

also the point of using a restrictive moderation configuration for the main RD instance is to promote self hosting but also to avoid liability on my part because i dont want to deal with moderation im sorry

i got a week worth of increased free time but i cant use any of it because im still too busy with stuff and also still too sick to think straight and do shit not to mention the terrible burnout after midterms last week

also idk how to fit this in but i really want to do the RD compositor at some point to make RD way more performant and efficient too but itll certainly make RD hosting less easy but it does allow for using a bsky.app fork instead of a bespoke client

To legacy-unaware users, or legacy-aware but disputing users, the legacy link is just a dead meaningless link. Its only meaningful locally, and maybe globally assuming consensus is reachee

Honestly i think its fine to always have a claim in the did doc forever.

Maybe an alternative to this could be like directly using the legacy<->did mapping verification system as a resolver so no need to have legacy declaration in the user side at all

What about uhh progressive enhancement. So like every time it finds a new acc declaring a legacy link it edits the records to change the legacy link to a did link. Nah its too fragile, what if you need to undo it? What if the mapping used was disputed?

It would probably work right now as is without infra changes. And you probably dont even need the legacy ids to resolve once you index everyone.

Though it probably does need a verification system cuz anyone can just claim to be any legacy id

Cool! I think i get it now.

So is the only purpose of legacy identifiers in place of dids or handles is for the alsoKnownAs linking? That seems way less complex now.

It works well for that sole purpose and its backwards compatible (did links to legacy) without needing to be forwards compatible

This morning i had ideas on how to have forkable user accs for collaborative content and like i thought about a bot that syncs to a git repo but this might be a better solution (?)