this post was submitted on 28 Jan 2024
145 points (98.0% liked)
Fediverse
28490 readers
353 users here now
A community to talk about the Fediverse and all it's related services using ActivityPub (Mastodon, Lemmy, KBin, etc).
If you wanted to get help with moderating your own community then head over to !moderators@lemmy.world!
Rules
- Posts must be on topic.
- Be respectful of others.
- Cite the sources used for graphs and other statistics.
- Follow the general Lemmy.world rules.
Learn more at these websites: Join The Fediverse Wiki, Fediverse.info, Wikipedia Page, The Federation Info (Stats), FediDB (Stats), Sub Rehab (Reddit Migration), Search Lemmy
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Yea it's one of those awkward things I suppose, where the instance (.world) is big enough that its operational concerns are kinda at odds with where lemmy as a project is up to.
With that many users, who also kinda expect a more reddit-style experience AFAICT, a certain amount of professionalism, stability and, in effect, slowness, is expected. And that's great.
But meanwhile lemmy is a small essentially underfunded project doing its best with a small group (2 main and a few voluntary on the side), which means bugs and then bug fixes and tweaking until things work ... all of which works well over a distributed array of smaller instances so that no single node is a major let alone fatal point of failure.
And so we've got this situation now where you could be critical of how lemmy.world relates to the bug fixing and testing load on the lemmy-verse. lemmy.world is likely the best funded instance (last I checked their donations exceed their infra needs) and yet the job of testing and working through bugs is offloaded onto all of the other smaller instances while they wait until it's all been ironed out. I don't know if it's a fair critique in the end, but it certainly seems to be there and worth considering.
Interesting point.
On the other side, as LW is more cautious about updates, that might have suggested some users to switch to instances that were more up-to-date.
But indeed I agree that people should be more spread, having 25% of Lemmy on one instance is less than ideal: https://fedidb.org/software/lemmy
Maybe? I'd fear instead that centralisation is pretty sticky without some massive failure and wouldn't expect much movement in the proportion of users on lemmy.world.
Which means, if I'm onto something with my critique, there may very well be a bit of problematic dynamic there.
It happened in August during the long DDoS attacks on LW.
I guess here it's less prevalent because the site is still accessible, and most of the users don't really follow closely the Lemmy versions.
Yea exactly. A DDoS is the sort of failure I'd imagine you need to shift people off. Not that big/central instances are completely bad. I think they help get people into the ecosystem that wouldn't otherwise ... and they turn out to be a pretty natural and constant phenomenon in the fediverse (mastodon.social is also about ~25% of masto) ... to the point that I personally start to ask questions about alternative structures which we spoke about in another thread.
Mastodon was actively pushing to be the default instance.
https://fediversereport.com/defederation/
Lemmy's community seems to be aware of the issue, and hopefully it will resolve over time.
Mastodons community is also very much aware of the issue. There are some who even call for defederstion from large instances. The main impetus being mastodons people’s concern for safety and moderation, where big instances necessarily allow much more to pass through. So while there’s more of a push from the top to make mastodon.social the flagship, from the grassroots there’s very much a push against it (I’m loosely a voice in that might self).
Lemmy.world’s push to be a big vanilla instance is not much different from what’s happening over at mastodon I’d say not least because it’s run by people running one of the big mastodon instances.
In the end I’m not sure the two spaces differ that much in their dynamics around this, notwithstanding the differences in the attitudes of the lead devs, especially given that the same result has occurred. Which is why I say it’s likely a natural phenomenon. You’ll found the distribution of instance sizes likely follows a power law which is common in nature.
That's good. I still think that having the developers pushing or not for an instance makes a different, wasn't Mastodon 40% Mastodon.social at some point?
Hmmm ... it might have been, I'm not sure. If so, not since or during the twitter migration, as my understanding there is that mastodon.social couldn't scale well and so it was up to all the other and often new instances to take on the load (similar to lemmy's story).
Since then, AFAIU, they've put their server on kubernetes so it can horizontally scale, and of course, put themselves as the default instance in the mobile app. But as user growth hasn't really continued on mastodon it's hard to tell what the dynamics are now. Having glanced at the numbers a bit over the past year my impression is that the proportion of users on mastodon.social has been slowly growing. I haven't seen any analysis though. I did some analysis early last year (https://hachyderm.io/@maegul/110331433071884694) and maybe I'll repeat that soon.