This is why I see the BlueSky lifetime as a grace period for fedi to get its act together.
-
@matthew Those are very complex issues related to how distributed collaborative systems work. There are solutions, but they:
- need to be carefuly considered on the performance standpoint, multiple things we looked at had the potential of multiplying the instance-to-instance trafic by a lot
- same for anything related to privacy and trust & safety
- require a lot of careful and expert work to be done correctlyA contributor is currently attempting to solve this here: https://github.com/mastodon/mastodon/pull/32615
-
Isn't this an AP problem, not a Mastodon problem? I'm replying to you now from a non-Mastodon AP instance, and I suffer from this problem greatly (because I run a solo instance and there's not as much chatter).
What's the AP-level solution to this problem? I have a hard time thinking there can be one, for the reasons you just mentioned.
-
@matthew The above PR attempts to refresh the replies from the upstream instance when the status context is requested from the API, if it is less fresh than X seconds, asynchronously. Then the frontend can check if the refresh has finished, and get the updated list of replies
-
I'm not following. How does a PR to Mastodon fix a protocol-level bug? Shouldn't there be a change to the protocol to address this problem?
-
@matthew We think the protocol does not need a change, but most implementations do need a change. And it might not work in the real world due to performance issues. I can not say what other AP implementations plan to do about it, but so far nothing in AP has proved blocking on our side from what I understand (I have not looked at this topic very deeply yet)
-
. Also I don't think most (white) people on fedi understand the impact of Blacksky, it's a great example of how technology focused on community can lead to people being able to find their people and created a self-reinforcing cycle.
@[email protected] @[email protected] @[email protected] @[email protected]
