Steam boilers obviously need to be protected from the catastrophe of running out of water -- you then either get a meltdown if the boiler is fired unwatered for an extended period of time or a boiler explosion if the water returns before then, neither of which are good. Hence, automatic low water cutoffs are required on steam (and hydronic, but that's beyond the scope of this question) boilers in all modern boiler codes.
Current low water cutoff technology consists of two types of cutoffs: probe and float. Floats have the advantage of not requiring an external electrical power source to work, but can get hung up due to deposits (scale), causing them to stop working. Probes lack moving mechanical parts, but require external electrical power to work, as they use the conductivity of the water as part of the switching circuit, and can also be fouled by boiler scale.
However, there are other properties we can measure to determine how much water's in the boiler -- one of them being the static head (pressure) of the boiler contents. This can be done either as a partial head measurement, measuring the pressure of the water above the minimum waterline with respect to steam pressure and cutting out at 0 differential, or as a total head measurement, measuring the static pressure of all the boiler water with respect to the boiler steam pressure and cutting out at a differential equal to the static head of the water at the minimum waterline. In both cases, the pressure switch would have a trapped steam line piped to it as well as the connection to the boiler water volume, and it would be wired as is typical for low-water cutoffs, opening the fuel control circuit when it trips.
How well would such a cutoff scheme work at protecting a boiler? Would a pressure-switch based safety be more or less vulnerable to failing hazardously (due to fouling or mechanical malfunction) than the existing float-type or probe-type cutoffs? Are there other flaws in this scheme I should be aware of?