My goal was to question the way we do things. Game streaming, for example, works differently. The game runs entirely on a remote server. The server has a GPU/CPU combo that renders the game in real time and user input is sent back to the server via UDP/WebRTC or a TCP tunnel.
Also, native mobile apps work quite differently from both video and game streaming.
Bottom line: we've seen some great innovations over the last 35 years, and I'm hoping to see plenty more. The most important thing is to question everything we do, to have that "yes, but why?" mindset.
Video streaming is still HTTP and the data is chunked. The reason data is chunked is so that you can have adaptive bitrate. Ie if one chunk takes too long to download then the next chunk will be the same time slice but at a lower bitrate. All served over HTTP.
HTTP itself has evolved massively over the years too. HTTP3 is actually a binary protocol built on top of UDP.
You’ve got the right attitude questioning why things are. But you’re also not alone. Which is why modern HTTP doesn’t look anything like its 90s predecessor and why so many public facing systems have adopted HTTP.
You do see proprietary video streaming protocols too. But the reason they’re not generally used for public-facing services is the same reason HTTP is everywhere: firewalls and routers work fine with HTTP because HTTP is ubiquitous. Whereas proprietary formats, and particularly ones based on UDP, don’t always play nicely with routers and firewalls.
In fact UDP in particular is a troublesome datagram for routing.
I’ve worked in the broadcasting sector for a while but it’s been a long long time since I’ve worked in the games industry. So I don’t know what is commonly used for game streaming. But for video streaming, they moved away from proprietary protocols to HTTP. As latency drops and bandwidth increases, I would expect to see the same trend with game streaming too.
The binary is a compressed artefact and the stream is a TLS pipe. But the principle is the same.
In fact videos streams over the web are actually based on how HTTP documents are chunked and retrieved, rather than the other way around.