Two-way voice with real-time video shows, as it happens, a visual sign that someone is speaking.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.