The idea is to minimalize the current flowing through them, so that they are as close to equilibrium as possible. In reality, when you look at the three electrode system schematics, these are just resistances in parallel, so there is always some very small current flowing, disturbing the equilibrium. If the current is small enough it can be - for all practical purposes - ignored.
Just a random thought: you could probably add a small resistor in series with the reference cell, to minimize the current further. Ohmic voltage drop would be negligible, and at least in theory the electrode would be close to equilibrium. But I don't think it would in any way prolong reference electrode life not increase the accuracy, so probably not worth the effort.