A bit paradoxal but it looks that all central platform (twitter, reddit, facebook…) are helping the spread of Fediverse. Recently we saw the impact with Twitter on Mastodon, myself I’ve discovered Lemmy even if I wasn’t a reddit user. And before that Facebook first spread friendica and diaspora. It looks next step will be around Youtube where Google try to lock more and more its user.
I think peertube is going to be much more difficult, videos require an insane level of compute/bandwidth to distribute.
I think peertube may have it’s day eventually, but it won’t be for much longer than link aggregators/microblogging
It’s honestly not that bad? Most of what’s being federated with PeerTube is just metadata related to the video, the video itself is hosted on whatever instance it’s hosted at, plus some peers spreading chunks of the data around.
I’ve been running PeerTube for years, and the compute / bandwidth / storage cost is way lower with things like S3-compatible Object Storage. It’s super cheap.
The biggest drain on compute power, weirdly enough, is transcoding video. Which can definitely suck if you’re running a big site, and everybody wants to have all kinds of different video resolutions. But even this can be delegated to external workers in an upcoming update.
The compute is getting cheap now too. By the end of this year, you’ll be able to slap a couple of these in a CoLo for a few grand, and then be able to transcode hundreds of videos at once.
https://www.servethehome.com/amd-alveo-ma35d-custom-anti-gpu-silicon-for-av1-era-video-transcoding/
You’d be able to compete with 2nd tier streaming services like Nebula or Floatplane at that level.
But yeah, it’s hard to get to Youtube/Netflix efficiency at a small scale, you just can’t get the specialized hardware needed to do all the hardware transcoding and DMA media serving that they do. But we’re getting there.
I feel that atm PeerTube has too many issues to make it viable and am hoping for a different platform to rise up. My issues with PeerTube largely come down to five things based on when I last used it (a few months ago):
UX/UI: It really suffers in this department. It looks and feels kind of old and unintuitive. This is the least of my complaints but imo it’s one of the worst mainstream fedi UIs.
Over-reliance on plugins vs native features. The devs have taken a strategy of “Feature doesn’t exist? People can make a plugin for it!” for several features, most notably stream live chat. In practice this means even more dependencies and even more components that can break with updates/changes. And since they’re not unilateral across instances it makes everything even more confusing than most fedi sites. IMO plugins should only be for slight changes that affect the local experience rather than major requested features.
Lack of a mobile app. As someone who doesn’t watch YouTube on my PC much but does on my phone, the devs statement that they don’t see mobile as a priority at all and have no plans to develop a mobile app really sucks. What does exist is exclusively for android and imo isn’t a great user experience either. This immediately kills my desire to watch content on a PeerTube instance.
Moderation problems. PeerTube has a serious lack of moderation tools for instances to where federating is largely not recommended due to neo-nazi and alt-right content slipping in even when you’re really trying to keep it out. Most federation features have notices saying that enabling them isn’t recommended. Yikes. When I was running an instance I had a very hard time keeping questionable content at bay, while I don’t have that issue on Masto or Pixelfed or really anything else.
Set up/maintenance can be a nightmare. Trying to set up a PeerTube instance involves a mess of dependencies and the upgrade process is not nearly as smooth as other software. I can say this as someone who has hosted PeerTube and had it break during an update about five or six times.
P2P based high-bandwidth uses seem really hard to implement on mobile. I can absolutely see why a mobile app wouldn’t be a priority for them since it is significantly harder to impossible depending on local mobile contracts.
That’s actually a great point, thank you for bringing that up, I hadn’t thought about it that way. IIRC, can’t PeerTube fallback to non-P2P video as well? Perhaps not ideal but might be a way of getting stuff to mobile users.
Oh! And another great example of plugins vs native features (since I didn’t have enough characters):
While running an instance, foreign language content kept coming in en masse, I was like wait why- my instance is set to English. When I asked on their Matrix they were like yeah it doesn’t have language filtering capabilities. I was like ok is that a planned feature? And they said no but someone can make a plugin for it.
Meanwhile, on Masto I can go into my preferences and filter my language in two seconds lol
The main issue is monetization. With social media, nobody cares about getting paid but Youtube or Twitch channels are the main revenue from lots of people nowadays. Not only the creators but also editors or writers.
I could see a patreon/floatplane style system where you only get the livestreams if you pay, or video releases are delayed for a week if you don’t pay. Video sponsorships also make more money than Youtube ads I think.
Peertube take care of bandwidth issue by adding a peer to peer streaming. I don’t know if big instances are having real bandwidth issue. I think Peertube or Owncast issue is more about the money / viewers locking of Youtube / Twitch.
I’m also of the understanding that the way other current federated protocols handle video is far from efficient, that google using activitypub for YouTube would be a nightmare.