This sounds vaguely similar to how electret microphones work, where the sound waves compress a dielectric and change the capacitance. If you apply a bias before

This sounds vaguely similar to how electret microphones work, where the sound waves compress a dielectric and change the capacitance. If you apply a bias beforehand to put some amount of charge q across the capacitor, q=cv so the voltage v measured accross it changes inversely with capacitance. This is probably best for capturing dynamics. An alternative method is to make the device part of an oscillator where its capacitance modulates the frequency. This would be better for precision measurements on a slower timescale. I wonder how the repeatability would compare with a hall probe.