Pixelfed Hit 500K Users | Hacker News Reader

Pixelfed Hit 500K Users

fedidb.org

306 pointsdoener83 comments

I'm excited. continue to think in the long run, I'm more excited by fedi than AT protocol

I've never been clear on why we should care about AT protocol. It's a VC funded effort which should immediately be suspicious for anyone wanting to build federated platforms.

I'm excited about ATProto and I'm also suspicious, because they could eventually try to turn away from federation. Although so far, they seem to be moving in the other way, as they keep developing and launching more stuff that makes the services more decentralized, not less.

I think it's healthy to be skeptical and suspicious about it, I am too. But I won't let that stop me from at least playing around with it and see where they go. Maybe one day they take the wrong turn, maybe not. They haven't so far, and I tend to wait to judge stuff based on what has happened rather than what could happen in the future.

if what you care about is "federated platforms", you might not care much (or maybe you might realize ActivityPub scaling is problematic).

many other people care more about "usable (right now) platforms".

The lack of handling identity is a showstopper for ActivityPub. I cannot for the life of me figure out why they didn’t go the extra and make identity portable.

Couldn't you, in theory, provide a service using ActivityPub + DID to kind of make this happen today? It's just that the current implementations (Mastodon et al) didn't implement things that way. I think I've read people arguing that here on HN at least.

I would love to read more about this idea.

What makes you more excited about Fedi vs. AT protocol?

You can host your own instance on a single server with Fedi, but need 3+ separate ones to host your own on AT protocol and it’s more complex.

I don't know much about AT protocol, but afaik, it technically could work as fedi if everyone ran their own servers. Right?

No, because there is no message passing in ATProto. So if you want eg. to know how many likes one of your posts has, you need to see the whole network aggregated through the firehose. No-one will push that "I like your post" message to your PDS.

Doesn’t seem AT is advanced enough yet because nobody else but Bluesky is able to run a server.

Or- fedi's design encourages more self-hosting because it's so easy and straightforward. With AT, yes it's harder to run your own instance but there's also less reason to: personalized moderation/feeds are just built in.

Can someone explain how is it planning to achieve the federation of that massive amount of data and storage ?

Isn't Pixelfed using the ActivityPub protocol?

As far as I know, that makes it scale like email. When you post something, your host sends it to all recipients - everyone who subscribed to your posts.

That does not sound like a lot of traffic. Even if one has 10k followers, as the post would only go once to each of their hosts. So if the 10k followers are on 100 hosts, that means 100 messages.

It is indeed using ActivityPub but most people are on the main instance and there's not much federation afaik.

What is needed is an index for all the instances that are willing to be published. All server instances would have access to the public index via some gossip protocol.

Also are there any other server instances other than the php server? Seems like they ought have used the matrix protcal instead of active pub. More server implementations in different langs, and it has e2e build I. Don't think actpub has that.

Well, sure - if someone wanted to be the hoster for all mail traffic on planet earth, then they would need a lot of resources and a business model to finance it.

But if that is their plan, why would they have used ActivityPub in the first place?

So I guess they could just shut their door as soon as the traffic is too much for them to handle. And tell people to look for another host or host themselfes.

right now. Let's make it grow

I believe the plan is to not have a plan (right now): https://mastodon.social/@dansup/113887622931474663

There is now also: https://www.kickstarter.com/projects/pixelfed/pixelfed-found...

Pixelfed works like any other Fediverse server. Content is stored on the node where it’s posted by its author. The nodes of people following the account have the option to cache a copy of the posts or just keep a reference depending on their implementation but Mastodon for example would cache it.

Mastodon seems to be doing pretty good so far on scaling up. Pixelfed is smaller than Mastodon, so it will be fine.

mastodon is text, pixelfed is images/videos

As I understand it activitypub itself shouldn't matter there as the only thing going through it should be URL references to the image/video data which you can front with a CDN. Federating instances don't have to host each others' data in that context.

In addition to the other points, there was some discussion about more directly p2p exchange of content (like WebTorrent or something) to share the load more evenly. Spritely is a more methodical approach to it but it could be layered into ActivityPub as an extension without changes to the core protocol.

Pixelfed could make a hash of every blob uploaded and then store that hash in the ActivityPub message. Then the clients could peer using WebTorrent or Trystero and randomly ask peers for that hash until it discovers a peer that has already cached the image. This would reduce load on dansup's server within his own ecosystem anyway.

> the federation of that massive amount of data and storage

I'm not sure what you mean by that but the media blobs (photos and videos) are not "federated" (i.e. passed around instances) most likely, but hosted in one place (the instance of the author) and referenced by their URL.

it varies by activitypub implementation. mastodon for example caches media per instance, pleroma simply hotlinks.

briefly searching through github issues, i believe pixelfed does not cache remote media. discussion on this issue about remote media cacheing seems to indicate that pixelfed only caches avatars https://github.com/pixelfed/pixelfed/issues/4571

I was under the impression that federated nodes acts as caches. Is this not the case?

Intuitively, it is unlikely there's 500K individual servers set up.

We can then also observe other comments clarifying that no, there aren't 500K instances.

The other comments provide...tweets? mastodons?...from the maintainer also clarify that in practice, there's 1 instance.

People are questioning how that will scale, and the tweet from the maintainer was cited as part of that because the content of the tweet is that tl;dr there's a $2,600 month gap between Patreon and hosting costs.

Why does the number at the top say "Total Users: 519,185", while the graph lower down, titled "Pixelfed User Count", maxes out at 316,151?

My guess: Total federated users interacting with the instance, vs total users registered on that instance

I can't answer your question but in that vein I want to add that in general it's very easy to get sign ups on a fedi node. That's why I always preferred instances that closely vetted new users.

The more relevant number is often active users.

I ran one for many years and I quickly disabled open sign-ups due to all the bots I noticed.

Isn't it as easy to fake the sign up number as it is to fake the active users number? What makes one harder to fake than the other?

One fakes itself, while the other requires someone to actively fake it by generating user activity. When I was active all it required was a regular login.

For many years now only the active user count has been relevant in the fedi, so I assume the creator of pixelfed knows this already.

I'm saying all this to say that news about user milestones mean nothing compared to the warmth and engagement regular users feel when interacting in the fedi.

I feel like I misunderstand something. Why are fake signups happening automatically? And if they do, why isn't faking user activity automatic too?

Do you mean that "Active User" count is just based on login, so it's "fake" as in "the user isn't actually active just because they only logged in"? In contrast to the other number that is based on actual engagement?

I'm sorry if I seem obtuse, I'm trying to make sure I understand correctly.

As an instance admin I can only say that there was a steady stream of sign ups from robot-like accounts, until I required accounts to write a motivation for signing up and then got rid of maybe 80% of new account sign-ups.

So I'm only speculating that a lot of that 80% were automated sign-ups.

Why create a bunch of fake accounts on fediverse nodes? I don't know, maybe they're used as currency later, or manipulation if the platform takes off. Who knows. These are only my observations.

Honestly, none of the numbers on this page add up..

Any recommendations for people to follow? Never used it before

Nobody interesting is really on there yet. It's also pretty buggy/laggy. The main network/server tends to go down. The bugs are mostly related to traffic overwhelming the servers.

Sorry I haven’t been following along and I have only a fuzzy idea of how these Twitter competitors work. Do users and posts on pixelfed federate with Bsky?

I’ve heard of some people I know leaving Twitter for bsky. I never used Twitter myself. So if I want to suddenly become social do I have to join bsky to follow them etc or is there a whole constellation of alternatives?

And if there are alternatives, which servers do the programmers who don’t want politics and memes and stuff congregate on?

Bluesky talks ATProto, which is a different protocol than ActivityPub used by Pixelfed.

Pixelfed federates with anything speaking AP in "the Fediverse", though, like Mastodon, Lemmy, PeerTube so you can follow people across all those (https://fediverse.party)

Bluesky isn't really federated, https://news.ycombinator.com/item?id=42215410

See https://fed.brid.gy/.

> Bridgy Fed connects web sites, the fediverse, and Bluesky. You can use it to make your profile on one visible in another, follow people, see their posts, and reply and like and repost them. Interactions work in both directions as much as possible.

I fully support, including financially, this project. But I worry that it will become what Mastodon has become which is (IMO) a complicated, shrinking, Twitter alternative with HOA's running large parts of it.

How? It has as perpetual "No posts found!

The service may be temporarily unavailable.

Pull to refresh"

message on the home screen of the app. Been like this since I first tried it in July.

Any one here on HN actually consume content from pixelfed?

I use this instance, and never had any problems https://metapixl.com/

I've been using the main server but it hasn't been a great experience. It isn't bad, just it hasn't been able to deal with the traffic.

I've been using it without issues, it's not perfect but I'm excited for something like the old instagram

Between pixelfed and bluesky, there must be some juicy scaling issues that the devops teams had to tackle.

If any of those devops folks are reading, tell us your horror stories!

Blog about Bluesky architectural evolution: https://newsletter.pragmaticengineer.com/p/bluesky

This is incredible.

119 million posts, x ~10MB per picture,

is like 1,200 petabytes? 1.2 exabytes?

Am I missing something here... this seems very impressive.

I believe Dan said yesterday? It costs around 4k a month. Unsure if he’s talking cad or us. Honestly more impressed it’s only that much.

More like 1 petabyte, no?

    iex(5)> mb = 119000000 * 10
    1190000000
    iex(6)> mb / 1000 / 1000
    1190.0

I’m making an assumption, but I would guess most of the images being shared are about a tenth of that.

Images and videos, though to be fair, the limit for both is 15 MB.

And I think it's a safe assumption there's some ffmpeg / imagemagick running somewhere that reduces the size even further before serving it to others.

Each instance can set it's own size limits.

The correct result of 119 million * 10MB is 1.19 PB.

Also the images on the first page of the author's own account average only 157kB, bringing your estimate to 18.6 TB...

pxscdn.com resolves to Digital Ocean, if he's using their S3-compatible storage, that's $372/mo + $0.01 per GB downloaded

You are off by some orders of magnitude.

More like 1MB per JPEG picture

Not even. Instagram photos are roughly 50kb-200kb for jpg’s.

if you compress it properly for a fast web experience, you'll probably be better 10KB-100KB

Those are transfer sizes, not sizes of the data at rest. Even transferred, the `srcset` for this image I picked at random includes a 178KB version; there's certainly a higher quality version stored, if not the original upload, something closer to it.

https://www.instagram.com/pet/p/C327eIguGRL/?img_index=1

Pixelfed active users: 27k

5% active users. Is that good or bad, compared to other social media sites?

Not sure if it's directly comparable to a service(pixelfed)'s MAU, but an engagement rate of 5% is considered quite good on social media.

Where do you see 5%?

At the top I see: 271,489 MAU and 532,524 total users (so 51%)

At the bottom (graphs) I see: 27,412 active and 316,151 user count (so 8.6%)

I don't know how to understand those numbers, wondering if you have a better source.

Just curious, is there an AT Protocol equivalent to Pixelfed?

Not right now, but it looks like someone is trying to build it on top of Bluesky: https://techcrunch.com/2025/01/15/bluesky-is-getting-its-own...

the one thing they need is better contacts/friends discovery tbh.

I wonder how many of those are bots.

How does moderation / censorship work on Pixelfed?

Exactly like it works in Mastodon - it's up to the server admins, plus some controls in the client. Choose a server where the moderation matches your preferences, or start your own if you're unhappy with the options.

The only thing I don’t like about the idea of having to pick the server / moderation is now that moderation meta conversation that is a potentially endless thing.

I just don’t know if back end meta is a really appealing feature to people in the long run…

Maybe it will be, I’m just unsure.

I mean, how is that different than choosing between commercial social media platforms today? You're tacitly accepting whatever their moderation policy is. If you're fine with that, then it really doesn't matter which server you choose, the generic primary Pixelfed server will probably be fine for you. If you do care, you can change your mind. Fortunately it's much easier to move instances and still be found than with any commercial social media platform.

I think moderation is a place where ATProto stands above the other options available. ATP gives the choice to the user, for moderation and algo feeds. Don't like Bluesky moderation, swap out the moderation service without changing apps or servers. Do your current moderation services miss something? Add another and stack them together. The design creates competition for both moderation and algorithms while removing the switching cost

https://bsky.social/about/blog/03-12-2024-stackable-moderati...

Among the big ones I find it hard to not be surrounded by nazis or other forms of eugenics enthusiasts regardless of their moderation. On Mastodon it's easy to do. Some say Bluesky has a fix for it but I dislike the people promoting it the hardest too so I didn't check.

Though I have to do a cleaning on Mastodon, shut away the Threads people that crop up on there now.

There are some (incomplete) bluesky mastodon bridges/merged clients, so it's possible to be majority in one and follow people on the other if your network isn't concentrated on one.

https://fed.brid.gy/

https://openvibe.social/

I’d say bridgy is quite an impressive bridge, incomplete isn’t fully capturing the extend of how well it work.

For me it is just the potential for endless meta discussion about moderating.

I agree that we choose one way or another, but putting that on you to pick between a bunch and then wonder what the difference is. It's a sorta "back end as a feature" but that is endless that is also a little disconnected from the experience.

I'm not explaining it well, it just seems like "more" work to some extent.

I think there's some truth to this, but I think we've just having a large meta conversation about social media moderation as a society right now that's largely unavoidable. That's not necessarily a bad thing.

On the other hand, I've largely avoided it because the fediverse allows me to just read the messages from the people I want to hear from, and put some filters in place if a few of those people occasionally talk too much about things I don't want to read about. I really have no interest at all in an algorithmic social media feed. If I want to read more about a specific topic, I head to a forum (hi, HN) or check my RSS feeds where the conversations are (generally) more focused and in-depth anyway.

Is there a default set of filters that is applied? My feeling is defaults end up skewing the entire platform.

You only see posts from people you follow or hashtags you follow, so my understanding is there are no default filters.

Reply on news.ycombinator.com