this post was submitted on 30 Jun 2023
19 points (100.0% liked)
/kbin meta
6 readers
1 users here now
Magazine dedicated to discussions about the kbin itself. Provide feedback, ask questions, suggest improvements, and engage in conversations related to the platform organization, policies, features, and community dynamics. ---- * Roadmap 2023 * m/kbinDevlog * m/kbinDesign
founded 2 years ago
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
I think the main way that could be achieved is if Kbin and Lemmy had a convenient "upload video" option that actually uploaded the video to peertube. Convenience is king. Back before Reddit offered image and video hosting (and you'd usually upload to a site like imgur instead), there'd constantly be people commenting that they didn't know how to upload their content.
That said, I'm personally cautious of PeerTube. Hosting small images is one thing, but video is something else. I don't really understand how PeerTube will keep running if it gets too much usage. Presumably, like most of these sites, it will depend on donations. I don't know if that will cut it for hosting video. My fear is that it'll be fine with low usage but as soon as it gets too high usage, we might see it going down (and taking a ton of content with it).
Peertube is inherently very scalable with relatively little cost due to an artifact of all social media platforms: Most of the traffic is driven by a tiny amount of videos/magazines/etc...
For services like youtube, you can use this as a way to quickly cache data close to the place it's going to be streamed: e.g. Netflix works with ISPs to install small servers at their locations to lessen the burden on their (and the ISPs) systems.
But with centralised systems you can only push this so far since ultimately everything is still concentrated at one central location.
Hypothetically, if you could stop this super-linear scaling for each user (you need to pay per user plus overhead generated from managing them at scale), you could easily compete against the likes of youtube simply because, at sufficient scale, all the other effects get ammortized away.
Peertube does exactly this by serving the videos as webtorrents: essentially this means that for every "chunk" of a video you downloaded, you also host that chunk for other people to download. That means that peertube itself theoretically only has to host every unique video once (or less than once since the chunks are in the network for a while), meaning you rid yourself of the curse of linear user scaling against users and only scale sub-linearly with the number of unique videos (how sub-linear depends on the lifetime for your individual torrents; i.e. how long a single video chunk stays available for others).
The costs that remain for every peertube instance is essentially the file hosting costs (and encoding the video, but that also only scales in the number of videos and could be pushed onto the uploader using WASM video encoders).
Storage itself isn't cheap, but also not ungodly expensive (especially since you can ammortize the costs over a long time as you platform grows with storage prices in a continual massive decline).
Platforms like Netflix and youtube cannot do this because
In general torrenting is a highly reliable and well tested method that scales fantastically well to large data needs (it quite literally becomes more efficient the more people use it)
@CoderKat It works on torrents, so the more people use it the faster it gets (in theory) https://en.wikipedia.org/wiki/WebTorrent
@NotTheOnlyGamer @McBinary