if the channel is still active, people can just go watch the videos on YouTube directly, no?

That’s one of the points of a decentralizing project; to move people off of youtube, and get many people to help share content that youtube/google currently controls. Creating torrents of their current youtube videos is an intermediary step… the end goal is if things like this get enough traction, content creators would move entirely off youtube, and to things like mastodon or lemmy communities, and post magnet links of their videos.

Therefore if one archives something, more often than not they’ll probably archive an entire channel, so it’d be easy to end up seeding a ton of videos.

As stated above sometimes this could be 500GB for some channels, and most users don’t have the HD space for that. It’d be better for videos to be more granular, and try to get the watchers of videos to participate in the torrenting, not just archivers with a ton of extra HD space.

https://github.com/jmbannon/ytdl-sub

Nice, seems very similar to TheFrenchGhosty’s scripts that are being used here.

Creating an individual torrent for every video doesn’t seem like a great solution, because the typical channel has lots of videos, up to thousands.

It’s def a tradeoff, and you’re correct that we could potentially seed entire channels.

But seeding single videos is IMO better for a few reasons:

• Torrent clients won’t have to select a subset of files.
• Its more future proof (ie you don’t have to make a new torrent after the channel releases a new video)
• Searching for individual videos is easier, since the torrent name can contain the youtube video_id
• Seeders won’t have to dedicate massive hd space to entire channels, especially since high quality videos nowadays can already be over 1GB. They just seed the specific videos they think are worth keeping.

The main problem I foresee, is streamlining the process of creating and uploading torrents, so that people actually do it.

And a browser plugin could potentially help.

The torrents-csv service has been around for a few years, and its one of the major components of this, since it can allow people adding torrents via pull requests.

People have been able to create torrents from youtube videos ever since yt-dl came out. The main difficulty is streamlining the process and getting people to actually do it.

I wrote up a preliminary thing cause I got bored today, lmk if you have any suggestions:

https://github.com/dessalines/torrent-tube

Like you said, it’s hash based, so the first time someone posts a video, you just download that from YouTube official and check the hash, no?

Yep, that’d be the case. But also for that service at least, it only stores currently active / well-seeded torrents. And the main issue is that people aren’t creating and seeding youtube videos as torrents. That would require some effort and a push from the datahoarder-type communities to start doing that.

When a user submits a new association between a Magnet Link and a YouTube tag, what verifies that that’s actually the appropriate file?

In one sense, its not really a problem, since torrents are static data. Incorrect files would not only look wrong (and not be seeded as a result), but have different hashes than the youtube ones. But the service which hosts / shares these magnet links would need to source the uploaders, so that their torrents could be removed if it turns out they’re spam.

I run a csv-based torrent service, but after many years, I’ve been the only one to really contribute to the data. There the process of adding data is tracked via pull requests. Any other service though which allows ppl to upload magnet links / torrent files would need to have a login system / tracking to prevent spammers.

I like this idea, but I wish someone would create a torrent-based project based on yt-dlp, and in particular a browser plugin that checks a torrent search site or index to see if that youtube video torrent exists.

Internet archive isn’t as fault-tolerant as torrents, and also wouldn’t be able to handle so many massive files.