It's not webbrowser based but live torrent streams like acestream work really very well. Unfortunatly acestream is pretty shady and closed-source and associated with malware distribution and illegal sport streaming - but it looks like a modified python bittornado implementation. Porting this to webtorrent would be interesting as it would at least allow to share the bandwidth for live-streams. I assume from a technical point of view this is not impossible.
I'm not aware of any open source implementations for the browser.
People really need to give up on the idea that the browser is a good foundation for robust p2p protocols. There's too many restrictions and the way users conventionally interact with web pages means that running instances of the client tend to be very ephemeral, even if you design it as an SPA. WebRTC kinda works as a transport but it's still a misapplication of the technology.
What PeerTube should do now is have some kind of opt-in peering between instances to mirror each others videos and provide secondary sources to share the load, and then later build some kind of desktop client for user mirroring of content.
They've had this bit for a while:
https://docs.joinpeertube.org/admin/following-instances#inst...
It worked really well at least at the small scale I was testing with :)
The webtorrent always had a static http seed which is how videos played fast, but that web seed was usually more than capable of serving all the traffic. More so, it _must_ be capable of doing so because not all your clients will be able to support webtorrent effectively.
In order to use a webtorrent, you needed someone else to be watching the same exact video as you at the same time, talk to the tracker, create a p2p connection, negotiate chunks and hope the other side has those chunks.
This just never happened. Even when I tried to force ideal scenarios it was just too slow to be usable.
I did some similar experiments with IPFS and results where much better but got worse every time I revisited the tests in the months after. Then protocol-labs completely scrapped all the libraries I was using and it doesn't work at all.
At the point I'm pretty uncertain if a p2p web browser video app can exist and be usable enough to have people actually want to use it.