Hi, my post is focusing specifically on YouTube since I observed the following categories have less intrusive solutions or privacy focused solutions, even if they are paid:
- Operating Systems (Linux, for example)
- Instant Messaging (Element, for example)
- Community Messaging (Revolt, for example)
- E-Mail (Proton, for example)
- Office (libreoffice, for example)
- Password Managers (Bitwarden, for example)
However, how do we distribute videos and watch them without data collection? I am NOT asking how do I use a privacy-focused front-end for YouTube, by the way, I am aware they exist.
I am wondering how we obtain a FOSS solution to something super critical such as YouTube. It is critical since it contains a lot of educational content (I’d wager more than any other platform), and arguably the most informative platform, despite having to filter through a lot of trash. During COVID, we even saw lecturers from universities upload their content on YouTube and telling students to watch those lectures. (I have first-hand experience with this at a respectable university).
I refuse to accept that there is nothing we can do about it.
That isn’t a solution, because YouTube provides discoverability as one of its primary draws. The primary draw for content creators is the promise of being paid via the ad network. So FTP doesn’t offer 1/10th of what people go to YouTube FOR. EVEN if your theoretical FTP server had literally every piece of content in the world; people would still go to YouTube.
Recommendation systems are well studied. I don’t think it’s unreasonable to add some form of recommendation layer separate from (or on top of) the content delivery. It doesn’t need to be up to par with YouTube’s, at least before there’s any major content.
Most YouTubers rely on sponsors or Patreon. Podcasters are doing the same - many of which are self hosting. So I don’t think an ad delivery system is that needed.
I don’t see how it would have to work much differently compared to how Pocketcast or Overcast already works.
The first problem is getting content to the platform.
There isn’t even a recommendation engine for video media that is open source and/or freely usable at this point in time that I am aware of. This needs to be a public database which is editable freely by everyone, in order to get every detail about media. Something like OpenStreetMaps; but for audio/video – and a suitable media recognition engine so that every variation of a media file doesn’t need its own entry. We also need laws so that services such as these are clearly protected by fair-use; because they currently aren’t, and get attacked at every chance by media conglomerates.
Recommendation systems don’t need to be that complicated. In its base form it’s just a list of videos you’ve watched (or content creators or topics). It can then be compared with the watching lists of other people to get an idea of what else you might be interested in. No need for any advanced video recognition.
Maybe this list is isolated within a single instance. Maybe it can be shared between instances. Different instances might use different recommendation systems.
Again, it might not work as well as YouTube’s, but I don’t think it needs to.