129
How to avoid the Reddit downfall
(lemmy.world)
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 01 Jul 2023
129 points (97.8% liked)
Reddit Was Fun
6374 readers
1 users here now
Memorial to "rif is fun for Reddit" Android app, aka "reddit is fun", shut down after June 30, 2023
founded 1 year ago
MODERATORS
Ah gotcha.
Isn't that a hard barrier/limit to scale then (as well as support)? Would it even be possible to run say a 5 million user Lemmy instance with a single write postgres DB (I assume compute can be load balanced, you can utilize CDNs for media content, can heavily cache the API, and that it supports read replicas?)
Nevermind 10, 30, or 50 million user communities 🤔
Though at that point you're essentially just lighting your bank account on fire for infra costs.
Probably, but the idea is lots of smaller instances, if you run into that barrier, your instance is probably too large. The largest fedi instances have around 20K - far less than a few million. Most smaller instances, like the one Im on are around 1-1000 users.
Id strongly discourage people from letting their instance reach the million user level as costs would probably far outweigh what a donation or subscription based server can support.
The point is to avoid the diminishing returns of monolithic instances and spread the load and users across multiple instances - if I were lemmy.world, kbin.social and beehaw I'd put a hard cap at 50K users. Wether they do that though is up to them.
What about general fragmentstion of large communities? I suppose by the time this has that level of adoption similar communities across instances may already be a solved problem as far as readership goes.
If there are caps on users that's a wet dream for corporate entities. Imagine groups preventing themselves from organizing, self-limiting knowledge sharing, and generally preventing themselves from being a collective force in an age where the internet is an integral part of life & society. Pre-broken down into fragments too small to be affect change or progress.
Also considered the user experience aspect of it. Fragmented communities are a poor user experience, where do I go? Where do I post? Am I missing information here? Which communities do I partake in? How do I collect information from disconnected, unaffiliated, sources? Experiences like that drive people away, not bring them in, and concepts like the fediverse rely on momentum to become more than just a transient niche.
That's my concern. 50k people is... Nothing. Even if it was a single topic. Nevermind general instances. Even simply for sharing knowledge that's a cripplingly, ineffectively, small number in a world where knowledge and information is becoming increasingly critical 🤔
I think you misunderstand the point and reasoning behind the fediverse in general
The point of the fediverse is to spread the load across multiple instances so everyday middle class people can afford to run them out of pocket or via a donation based fund. It's not to grow the instances so large that they become a monolithic social platform in and of themselves.
The reason behind this is the mistrust people have in general of these big techbro social media companies and the problems they cause when they inevitably get enshittified - we're trying to avoid doing that. That's why you're seeing hundreds of instances pledging to block/defederate with meta - especially with prior knowledge of what happened when monolithic google talk federated with the XMPP protocol. Others are waiting to see what happens and what Facebook plan to do before taking action, and some are all for it.
As for measuring the knowledge sharing power of the threadiverse (Lemmy and kbin combined) and the fediverse in general, we shouldn't look at a single instance, but rather the verse as a whole. If the fediverse gets large enough, search engines will have to come up with a way to better index the knowledge that comes from them, otherwise they are missing out on a huge chunk of the internet. Sure lemmy.world currently has 20-30K people but across the whole threadiverse we have 100K+ members across known instances and thats only set to grow higher as time goes by.
Some people are already worried that lemmy.world has already gotten too large too quickly. Beehaw outright defederated with lemmy.world due to the fact their small moderation team struggled to moderate the huge influx of members coming from lemmy.world - and others are concerned about what happens if lemmy.world goes down and takes it's communities with it - In a threadiverse of 100K+ users, 20-30K users is a significant chunk and it gives lemmy.world a lot of staying and federative power. People don't want it to become too big, because as a result people will have to federate with it out of necessity.
Interestingly the defederation of Lemmy.world and beehaw have lead to people moving their communities off of those instances to ones that are still federating with both, which I think is actually a good thing, as it takes some of the power from Lemmy.world
This leads to why I think community fragmentation is actually a non issue - it's entirely possible that app developers could create multis so people can combine a list of communities into a feed. They could even set up an automatic name-based combination based on what communities your instance already knows about. In the end the actual maintaners of Lemmy could add the functionality to save these multis to your account in a future update. Also, people have a tendency to go where the people are - which is why Lemmy is becoming a viable option in the first place, big communities will get bigger, smaller communities tend to stay small.
These are all problems to be solved and can be solved - we just have to keep the point and reason behind the fediverse in mind when we come up with a solution.
Lemmy hasn't even reached the 1.0 release yet (some instances are on 0.18 and some still on 0.17) it's still a beta, maybe even alpha build. There is plenty of work to do, and I wouldn't consider the current version even remotely feature complete. There are still tonnes of issues and feature requests to sort out too. But due to the open source nature, you can not only contribute, but you can also outright fork if if you don't like the direction of the software or want to quickly include your own features - I know lemmynsfw has their own fork of the software to help cater for the nuance of NSFW posts.