![[sh.it.post] We're #5! We're #5!](https://sh.itjust.works/pictrs/image/58e76358-5e82-4f03-b7ed-a6754b7875b9.jpeg){kind=link}
Just saw this on AskLemmy at .ml, thought this and chuckled, and now here we are.
Will take the opportunity to thank our admins for what they do, and all you humans for being here and generally being cool.
Just saw this on AskLemmy at .ml, thought this and chuckled, and now here we are.
Will take the opportunity to thank our admins for what they do, and all you humans for being here and generally being cool.
Wouldn’t total posts bias towards older instances though, counting posts over time rather than activity today? So then good point that sh.itjust.works is so high up by both metrics:-).
While lemmy.ml continues to fall - by active users I think I recall it was #3 at some point, then #4, while now it’s #5, where based on the gap below it, it seems likely to remain since users are now more distributed than previously (which is a good thing!:-).
Oh yeah, monthly active users is definitely the better metric, I just wanted to point out that in this case, it falls in the same place either way.
I wonder what the best metric would be for measuring how distributed Lemmy is. Maybe the ratio between total active monthly users vs the top 5 or 10 instances?
People used to say how Lemmy.World had ~80% of all users on the Fediverse. I’m not sure if that was older defunct accounts or what. But it does illustrate one thing: does it even matter where user accounts are located, when the federation model means that someone can access the entire thing, minus only whatever someone’s instance has chosen to defederate from?
On Mastodon that matters greatly, due to the discoverability aspect, but here on the Threadiverse (or whatever we want to call ourselves to distinguish the forum vs. microblogging nature of our spaces, accessible via Lemmy, some app, Mbin, Friendica, PieFed, Tesseract, perhaps Sublinks one day, etc.)? On that note, my instances (Kbin.social, then StarTrek.Website, Discuss.Online, and now a mix between that and PieFed.social) have mostly been extremely tiny, but I never felt like I was excluded, being able to browse by All.
In fact quite the opposite! Having wandered into [email protected] and lemmygrad.ml and thereby exposing myself to their echo chambers, right inside the very ones hosted on their own instances but due to federation, hosted likewise on my instance as well, I strongly wished that the Fediverse would have been a little less connected - or at least if it has offered me some warning! (The sidebar text is only shown on a “community” page, not an individual post when arrived at via browsing All.)
And then there’s communities to consider - so many are on Lemmy.World, but how much should that matter, vs. the users? Moderation though is primarily something related to communities. So like sh.itjust.works doesn’t have all that many, there’s e.g. [email protected], and lemmy.ca also, like there’s [email protected] and [email protected], yet these general-purpose instances have so very many users, even if the communities themselves are mostly on lemmy.world.
If lemmy.world were to go down though, we’d lose a LOT, at least in the short term. Archived copies of older posts would remain cached on remote instances, but a new community would have to be created somewhere in order to allow continued posting.
So I don’t think that the Threadiverse is all that distributed - but I also don’t think that it matters for us?
It would be nice to see a way of unifying communities on multiple instances for redundancy and improving the situation with redundant communities across instances.
I’d imagine it would probably need to be an opt-in option on a per-community basis where one community can request to unify with another one, then from the other community the moderators can accept or reject the request, then the posts, scores and comments would be mirrored and maintained simultaneously across instances. Differences in block lists between instances would probably be a challenge but not an insurmountable one. Bigger challenge might be latency problems with the mirroring and federation but there’s enough existing redundancy protocols that allow for servers with rediculous latency so that’s probably also not insurmountable
It’s really on the mods to accept to consolidate in one community.
All the issues you mention in your other paragraph are why this is probably never going to happen.
I’m mostly talking out a technical solution for community redundancy, similar to setting up redundant VM hosts or more accurately like redundant network hardware, where the data and configurations all exist on both servers, and might be load balanced to some degree between both servers, but ultimately should one go down there’s no loss of uptime as the other server takes over until the time that it’s mate comes back online or a new one is setup and connected
But that’s infrastructure redundancy, it should rather happen there than at the Lemmy level, no?
Mods are generally not going to surrender their power, especially as differences in moderation techniques is what most often leads to new community formation in the first place:-).
Other than PieFed’s Categories of Communities, the only other ways I know of to make Topic aggregations like that is to either find and switch to using an app that provides such (I have heard of at least one, but I don’t recall which:-), or else have many Lemmy accounts and constantly switch between them - like one for News, another for Memes, etc. Or you can browse by All, and see mostly only memes and news from the USA all the time:-P.
But PieFed provides numerous methods handle this: not only Categories but also Topics to aid community discovery, and you can trigger Notifications for anything - a person, a comment, post, or even an entire community.
One day very soon I strongly believe that PieFed will surpass Lemmy in terms of usability. It already has in so many ways, though not quite all.
I’m thinking more unifying communities that either have the same mods or for annexing communities with inactive mods, and I keep referring to redundancy because that’s the specific purpose in my mind, with the side effect of cleaning up the multiple dead communities with the same name on various instances.
There’s a real risk in the Fediverse of the one server hosting a community going offline, and we’ve already seen at least one notable Lemmy host shamble on as a zombie server with absent instance administrators. Instead of forcing communities to tell eachother to migrate or to recreate themselves on a new instance should one disappear suddenly, by having the community effectively load balanced and replicated across 2 or more instances is a lot more resilient
I fully respect when moderator teams have different opinions running similar communities with different rules and expectations and in not saying that should be taken away. I’m just thinking about technical solutions to improve overall Fediverse health
There’s a reason the current Lemmy sourcecode is named 0.19.8 - this is beta version software, awaiting very many changes to reach completion. Heck, reports don’t even reach moderators on remote instances yet, as the whole drama with 196 is showing (they left the mod reports there for days, then got mad when the instance admin did what she said she would and cleaned up in the absence of the mods being willing to do anything about the situation).
But people either don’t know Rust (it’s reputedly difficult), don’t want to deal with those developers, or simply don’t want to help with the writing of code. Hence the creation of K/Mbin, PieFed, and Sublinks to deal with the former pair of issues, though not catching up to feature parity with Lemmy yet (except PieFed is already quite ahead, in some ways even though not ready for the masses in some fewer but more foundational and crucial ways).