This is frustratingly one-sided writing. Yeah, WebRTC has limitations, but relying on a standard buys you a lot of correctness and reduces long-term engineering cost. The fact that WebRTC is complicated does not mean it is wrong; it means real-time media over the public internet is complicated.
Also, networking is inherently stateful. NAT traversal, jitter buffers, congestion control, packet loss, codec state, encryption, and session routing do not disappear because you put audio over TCP or WebSocket. Pretending otherwise is not architectural clarity. It is just moving the complexity somewhere less visible.
Also, networking is inherently stateful. NAT traversal, jitter buffers, congestion control, packet loss, codec state, encryption, and session routing do not disappear because you put audio over TCP or WebSocket. Pretending otherwise is not architectural clarity. It is just moving the complexity somewhere less visible.