Electromechanical relays have some lovely features: they can provide galvanic isolation, very high crosstalk immunity, enormous gain, very low on-impedance, and, in locking designs, bistability even when the power goes out.
However, they have some big problems: they’re power-hungry and their operating speed is limited to the kHz to tens of kHz and their operating life is limited to thousands to millions of operations, which usually limits them to average operating speeds in the millihertz or less.
The reason for the short operating life is contact oxidation. (In theory elastic metal fatigue or creep could play a role too, but those are easy enough to avoid.) Mercury-wetted relays are a common design to lengthen this life, and Paper/foil relays discusses the possibility of using contacts of carbon (like keyboard dome switches), silver, or gold instead.
In other sense-switch applications, a common approach to avoiding the oxidation problem is to use phenomena other than conduction to transfer the energy. The TRS-80 keyboard, for example, was capacitive (though using springs made of polyurethane foam, which degraded rapidly), as are modern touchscreens, touchpads, and some touch panels in embedded devices. And there are numerous inductive sensors for position, orientation, and so on.
It occurs to me that relays can work through these media as well.
For example, if you have two ferrite rods with one winding around each of them, you can make them into a transformer that efficiently transfers AC power from one to the other, from dozens of Hz up to several kHz, by completing the magnetic circuit with more ferrite. By moving this additional ferrite under the control of a solenoid, you have a relay, one that will never suffer contact oxidation, because the contacts are magnetic rather than electrical.
Similarly, although the circuits described in Paper/foil relays are the usual kind of dc-coupled contact circuits, you could use a similar design to bring one of the plates of a variable capacitor into contact with the dielectric from a distance far enough to drop the capacitance by orders of magnitude. This could easily have enough capacitance to efficiently transmit power at frequencies of 100 MHz and up, again without any electrical contact and thus no oxidation. Such capacitive relays could move smaller amounts of mass, and over shorter distances, than the inductive relays described above, and so they should be able to operate much faster.