In Microelectronic Circuits by Sedra/Smith, the authors perform small-signal analysis on a basic active-loaded MOS differential pair amplifier. In attempting to find the amplifier's short-circuit output current, they reason that it is appropriate to assume that the common source node of the input transistor pair is a virtual ground even though the circuit is not symmetric. They explain that because the drain of Q1 is connected to ground through the low impedance of a diode-connected MOSFET, the drain of Q1 is a low impedance node and therefore must be close to ground potential. This approximation makes the circuit symmetric and greatly simplifies the analysis. Reference pg. 639 for anyone who has the book.
So, the question is: under what circumstances are we able to claim that a low impedance node must measure nearly zero volts?