Pixelfed's first plateau in progress
https://pixelfed.fediverse.observer/dailystats
Looks like Pixelfed's growth spurt is slowing down. Absent any new external stimuli I'm guessing it'll stabilize around 200K to 300K monthly active users -- over a hundredfold order of magnitude from what it was just a month ago.
You're viewing a single thread.
You are very optimistic about the user retention š
People might stay there more than on Lemmy. Pixelfed seems in a better shape than how Lemmy was during summer 2023, with the constant DDoS on Lemmy.world and the language bugs preventing comments...
I think you're on to something.
I also think there's a difference in where the network effect kicks in for different types of social media. IMHO, Lemmy has just enough activity to not feel empty, and even then I wish there was more comments to interact with and more niche communities. With Pixelfed, I feel like as long as there's enough interesting posts it makes sense for people to visit regularly.
You're 100% right about that. I've never ran out of nice posts to look at on Pixelfed. I think the medium & format is a lot less addictive and a lot more relaxing/positive, which might help to explain it.
It depends. I donāt feel forced to go on there because it has none of my friends. Obviously Iāll go once every few weeks to check out cool photography but Iām really not an active user there.
The Pixelfed app isnāt very good right now compared to instagram was when it wasnāt terrible. Hope that Dan hires an actual UX designer to update the app with his kickstarter money. And that he can integrate loops with it as well.
I've been trying to set it up for 2 weeks now, the docker containers. There are so many backwards decisions, it's pretty clear he doesn't understand docker unfortunately. The entire app could be a snap to set up, but it's such a convoluted setup that it scares a lot of people away. (It pretty much is scripted assuming you will run it on a VM, and a high cost one at that).
The fact that it's built on PHP and Laravel in 2025 says a lot, and the fact that he started Loops this year on the same architecture also says a lot. It just doesn't scale, it's locked to a single host, and he's finding that as the servers are tipping over.
I'm not a PHP fan but it scales better than Python or Ruby (Mastodon) does. I think Dan is a cowboy of an engineer, but blaming performance on his stack choice is a bad take.
Not php but Laravel. Like I said it forces workers to be on the same host with the same storage as the API, it doesn't allow scaling of multiple API nodes or worker nodes, and his docker containers require me running special commands left and right. No other docker containers of mine do that
Havenāt tried self hosting Pixelfed. Just mostly been trying it out on the main instance. Iām really shocked that itās a mess to deploy with docker if itās on PHP/Laravel.
I hate PHP these days for dev purposes but I think laravel ought to be able to scale enough to run most Pixelfed instances. Facebook ran PHP when it was much larger than Pixelfed.
Facebook had to basically rewrite php so it would scale though, it's called Hack now
Check out YUNOhost. I had a Pixelfed server up and running in 20 mins.
The entire app could be a snap to set up
I refuse to use snaps
Heh
Yeah not having stories is a huge shortcoming.
I don't even know the point is stories. Why not post... A picture or video directly?
People want transient videos that don't stay in the gallery.
Looking at my friends who use Instagram, it's basically for stuff that's either too many photos for a single post, so it's made into a themed story, or for stuff you wanna show off, even though it isn't quite good enough for it's own post, so it's on your profile for a bit, but not permanently.
user retention
yeah, hard to tell.... Lemmy peaked a little under 70k MAUs and is around 45k now... if pixelfed peaks at 300k it's reasonable to think it levels at 200k (i.e. a hundredfold increase from a month ago).
ofc every situation is different... e.g. pixelfed has tighter Mastodon integration (pro) but may depend more on a network effect (con). also iirc the lemmy MAU count methodology shifted at some point, from post/comment to post/comment/upvote/downvote which is a confound for the lemmy dropoff count...
Dans actions will decide user retention, I think.
What goes up must come down
