I’m sharing this post and it’s thread because I’ve run into this issue of not being able to find communities, users, posts, or comments on other instances, or the communities are missing a substantial number of comments and votes even after interacting with these posts and receiving new replies. Is this functionality standard and desirable behavior? Why or why not? I’m trying to understand how it’s expected to work, because it’s not very intuitive starting out to explore other instances sometimes.
Maybe we could brainstorm a way to make it more intuitive. What are the trade-offs involved? Why does it work the way it currently works? Is there a better way?
The main point that has been brought up is the fact that users from smaller, newer instances will not be able to access the backlog of posts and comments from older more established instances. I’d say a lot of the value of the platform comes from this foundation of posts sorted by votes on a link aggregator site like this. Should new users be encouraged to join the older, more established instances in order to have access to the most content?
cross-posted from: https://sh.itjust.works/post/357141
SOLUTION FOUND - thank you
There is a community called Tarot on lemmy.world, and I want to subscribe to it. The url is https://lemmy.world/c/tarot. I’ve been to lemmy.world when not logged in and I can see it there. But when I try to find it from here, I can’t. I tried putting !tarot@lemmy.world in the search and got “no results found”. I’ve been told that “someone needs to search for it, for it to appear”, but I have already tried to search for it, many times. What else needs to happen for me to be able to see and subscribe to this community?
Or do I just have to make a second account on lemmy.world?
Yes, the process to find a new community on another instance is pretty clunky. For one, part of the issue of the “i searched for it and it didn’t appear” is your home instance has to wait for the workers on the remote instance to actually reply with “yes, this community exists” before you can browse to it. And your home instance must have workers free to actually make such a request. Many communities are heavily overloaded and these federation tasks can run many minutes behind which leads to a frustrating experience trying to find a community.
You will only ever see NEW posts that are made AFTER an instance first federates a community. This is intentional, because loading all back history out of the database once these communities exist for a few years would be a gargantuan, horrifyingly resource intensive task. As such, all back history will only be loaded “on request”, aka, the remote instance will only send pre-federation content if a user on the local instance specifically attempts to browse to it.
Follow up question: After the first contact, will all new posts, comments, votes always be federated between the two instances?
And also, after federation occurs, I guess that means that new posts from the other instance will now show up in the ‘All’ feed, where as they wouldn’t have before.
Yes, all new posts and comments from specific communities that have been accessed will always be federated between the two, as long as the worker threads are working and the databases tracking what’s new and such don’t get fucked up. (So, theoretically yes, but no guarantees, federation can get slow and out of sync sometimes with how much load many instances are under currently)
As I understand it they don’t grab the entire server contents because that would result in a ridiculous amount of traffic. Just communities that individual users have browsed to.
Thanks for the insight.
This is a major hurdle for discovery. My workflow to find new communities is basically to search for them on several different instances, visit their instance domain directly, search for their url through my home instance, and even then it’s occasionally useless because the only posts available to view in a feed from my home instance have no votes, no comments, and are pretty random. All the interesting posts that have had a moment to be voted on and commented on are a couple days old, and thus the only way to find things to interact with is to use the instance through it’s native domain. I suppose I could manually search for each URL of posts and comments as I browse in the native domain back into the search engine of my home domain, but this is insanity. Is this really the way we want to do things?
What if there was a way for one instance to request not the entire backlog of posts at once in these situations, but a series of posts long enough to fill one page of a feed at a time that match some search criterion when they attempt to directly explore a new community/instance, even if they are posts created pre-federation. Then the ‘most commented’ or ‘top of the week’ or especially ‘Top of All Time’ and old pinned posts, basically the ones people would want to see, that define a community, would over time be federated as users browsed, accumulating a select subset of pre-federated posts on-demand. Also, it seems like having all the votes and comments on a posts that you just requested would be important. Right now if I search for a specific pre-federated post, I get a bare bones version of it with a tiny fraction of the votes, and usually 0 comments. Something like that seems like it would be a huge step in usability and discoverability, especially for users on newer, smaller instances. I don’t know much about how the system works though, so I don’t know it this is the right idea.