This is quite cool actually. It reminds you that if your device has a sensor, it can communicate.
It should be possible to communicate through a webcam and a screen when the airgapped devices are on the same room. It could be possible to communicate by accelerometer(macs has these) and inducing vibrations using the HDD when the devices sit on the same table.
accelerometer: is very low bandwidth (think ~10-1000 measurements per second). The noise is ~100μg/√Hz - with 12 bits samples I'm not sure you'll pick up HDD vibrations (but rather easy to test).
Someone should lend the people researching this an oscilloscope.
No, not QR codes since it would require the webcam capturing the screen directly. The screen may be used as a light source for the transmission and the webcam can capture the light reflecting from the walls...
Certainly devices can "communicate" via sensory data, but that shouldn't cause your device to randomly execute code after receiving that data. Unless the programs that read from sensors are really poorly sandboxed...
It should be possible to communicate through a webcam and a screen when the airgapped devices are on the same room. It could be possible to communicate by accelerometer(macs has these) and inducing vibrations using the HDD when the devices sit on the same table.