Because bandwidth is expensive and at a certain point so is storage.
Also because ISPs are terrible companies that will sabotage your attempt to stream video for no reason other than spite.
In theory you can just point a browser at a video and it'll play. The problem is, for a significant chunk of the world, that video will only play halfway through, or play barely if at all, or it'll take an hour to load even though you have an excellent connection.
Then, there's the discoverability problem. If you want to create a channel, you want people to know your channel exists. So do millions of others. Youtube and similar platforms use tag/content correlation to suggest videos (and then some AI bullshit to do the same but worse) to naturally grow your audience over time.
The web archive copy of his LibreELEC video is 860MiB in size. His recent RPi video got 156k views over 6 days. Transferring the 1073 TBit of video over those 6 days in the most optimal scenario would require a constant stream of 2070mbps. At the VPS provider I find most reliable, that'll cost him €124 just for the network traffic for the first six days. That's not taking into account the burst of tens of thousands of viewers just after releasing a video, or the fact that Jeff has tons of other videos on his channel.
Of course there are optimizations. You could transcode the video to even lower bitrates to save bandwidth, you could set up different quality profiles and write/buy a complex video player library to handle those automatically. Peertube tries to solve this problem by leveraging P2P video, but not everyone is a fan of exposing their IP address to every other viewer (bot, human, data collection tool) when they play a video. And then you need to kick out all the scrapers and download bots wasting your bandwidth for their personal gain.
Plus, you can't monetize the videos. Great for people consuming content already paid for, like government instruction videos or corporate productions, but terrible for people who use Youtube to fund the video creation process itself.
Even Jeff Geerling, someone with a sizeable audience of relatively wealthy audience (thanks, tech industry!), says he cannot maintain his channel through Patreon alone:
> I was never able to sustain my open source work based on patronage, and content production is the same—just more expensive to maintain to any standard (each video takes between 10-300 hours to produce, and I have a family to feed, and US health insurance companies to fund).
Youtube happened for a good reason and it can't happen again without wasting billions on making the servers, software, bandwidth, and content in general free for years before it can reach critical mass. Even for hobbyists, offsetting the network egress fees alone would be a challenge without monetization.
Linus Tech Tips is trying to spread their eggs across more baskets by setting up Floatplane, and there's a reason Floatplane is far from free. Sticking a webm file on a web server somewhere is a solution for videos that get a couple hundred views, but it quickly becomes unsustainable.
Also because ISPs are terrible companies that will sabotage your attempt to stream video for no reason other than spite.
In theory you can just point a browser at a video and it'll play. The problem is, for a significant chunk of the world, that video will only play halfway through, or play barely if at all, or it'll take an hour to load even though you have an excellent connection.
Then, there's the discoverability problem. If you want to create a channel, you want people to know your channel exists. So do millions of others. Youtube and similar platforms use tag/content correlation to suggest videos (and then some AI bullshit to do the same but worse) to naturally grow your audience over time.
The web archive copy of his LibreELEC video is 860MiB in size. His recent RPi video got 156k views over 6 days. Transferring the 1073 TBit of video over those 6 days in the most optimal scenario would require a constant stream of 2070mbps. At the VPS provider I find most reliable, that'll cost him €124 just for the network traffic for the first six days. That's not taking into account the burst of tens of thousands of viewers just after releasing a video, or the fact that Jeff has tons of other videos on his channel.
Of course there are optimizations. You could transcode the video to even lower bitrates to save bandwidth, you could set up different quality profiles and write/buy a complex video player library to handle those automatically. Peertube tries to solve this problem by leveraging P2P video, but not everyone is a fan of exposing their IP address to every other viewer (bot, human, data collection tool) when they play a video. And then you need to kick out all the scrapers and download bots wasting your bandwidth for their personal gain.
Plus, you can't monetize the videos. Great for people consuming content already paid for, like government instruction videos or corporate productions, but terrible for people who use Youtube to fund the video creation process itself.
Even Jeff Geerling, someone with a sizeable audience of relatively wealthy audience (thanks, tech industry!), says he cannot maintain his channel through Patreon alone:
> I was never able to sustain my open source work based on patronage, and content production is the same—just more expensive to maintain to any standard (each video takes between 10-300 hours to produce, and I have a family to feed, and US health insurance companies to fund).
Youtube happened for a good reason and it can't happen again without wasting billions on making the servers, software, bandwidth, and content in general free for years before it can reach critical mass. Even for hobbyists, offsetting the network egress fees alone would be a challenge without monetization.
Linus Tech Tips is trying to spread their eggs across more baskets by setting up Floatplane, and there's a reason Floatplane is far from free. Sticking a webm file on a web server somewhere is a solution for videos that get a couple hundred views, but it quickly becomes unsustainable.