Why does this happen? Gasses become conductive when they are converted into plasma. Plasma, also referred to as the fourth state of matter, is a fluid composed not of molecules but of positive ions (cations) and electrons. Plasmas are formed when energy is put into a collection of molecules faster than energy is lost by radiation when the cations and electrons recombine. Recombination of the cations and electrons causes energy to be released as light, this is how fluorescent lights and neon signs work.
In a simple atmospheric pressure spark gap, as the potential difference (voltage) between the electrodes is increased (or as the distance between the electrodes is decreased), electrons begin to be emitted by the cathode (negative electrode) and travel to the anode (positive electrode). As they travel, some of the electrons will collide with gas molecules, knocking electrons loose and forming cations and more free electrons. Near the electrodes, where the concentration of traveling electrons is highest, a faint glow caused by the recombination of ions and electrons will become visible. This glow is called a corona discharge and removes energy from the ionized gas at a high enough rate to prevent the formation of plasma. At higher voltages (or smaller gaps) the energy put into the molecules by electron collision exceeds the ability of the corona discharge to dissipate the energy and a plasma is formed. The electric field between the electrodes will then separate the cations and electrons. The electrons will flow towards the anode, while the cations will flow towards the cathode. When the cations impact the cathode, they recombine with electrons from the surface of the cathode, completing the electric circuit. In an atmospheric pressure spark gap, the flow of cations and electrons tends to be confined to a fairly narrow channel which is called a spark. At the point on the cathode where the spark connects, a large amount of heat is generated by the impacting cations, damaging the electrode surface (more current = more damage). Electrons impacting the anode surface do not cause much damage, since they are thousands of times lighter than the cations and thus have much less kinetic energy.
Before we can go further, we need to understand gases better, particularly how the rate of collision between gas molecules changes as the pressure changes. The rate at which molecules collide depends on the concentration of molecules (how many there are in a given volume) and the average velocity of the molecules. The temperature of a gas is actually a measure of the average kinetic energy of the molecules which make up the gas. The velocity of the molecules then depends on temperature and the mass of the individual molecules. The pressure of a gas is the amount of force the gas exerts against a unit of area. This is a combination of the kinetic energy of the individual molecules and the total number of molecules in a unit of volume. (This is the Kinetic Theory of Gases in a nut shell.) Now, if we consider a sample of gas at constant temperature and volume, it is clear that a reduction in pressure means that there are fewer molecules and fewer collisions. Another way to think of this is that the distance a molecule can travel (remember we have not changed the temperature, so the average velocity of the molecules has not changed) without colliding with another molecule has increased. This distance is called the "mean free path". In air at one atmosphere of pressure and a temperature of 0º C (standard temperature and pressure or STP) the mean free path is 1 X 10-7 meters (or 1 X 10-4 millimeters, 0.1 micrometers, 100 nanometers).
If we go back to the atmospheric pressure spark gap for a moment, we now realize that one electron, traveling a centimeter or so from the cathode to the anode, will collide with about 100,000 gas molecules. Whether or not a particular molecule will be ionized depends on how much kinetic energy the electron has when it hits the molecule. If the energy of the incoming electron is greater than the binding force of the electrons in the molecule (the "ionization potential") then one or more electrons can be knocked loose. Unlike the gas molecules, whose kinetic energy depends on the gas temperature, the kinetic energy of the electrons is determined by the acceleration caused by the electric field between the electrodes and the distance over which this accelerating force acts. The distance that the electron "falls" in the electric field (and hence its kinetic energy) is limited by the mean free path of the gas. Every time the electron collides with a gas molecule, it loses most of its kinetic energy. If the distance the electron accelerates through is very small, it will never have enough kinetic energy to ionize the gas molecules. This is why higher pressure requires a greater voltage before a plasma can form. Longer gaps reduce the electric field strength and consequently reduce the kinetic energy of the electrons which is why longer gaps require higher voltages to initiate plasmas.
So what is happening on the "left side of the Paschen curve"? As the pressure drops, the mean free path of the gas increases and the kinetic energy of the electrons will also increase, meaning that a collision with a gas molecule will be more likely to result in ionization. Now remember that the x-axis of the Paschen curve is not pressure, but the product of pressure and distance. What has happened is that the distance between the electrodes is now smaller than the mean free path of the gas. The electrons have plenty of kinetic energy, but they are no longer colliding with any gas molecules so no ionization occurs.
Finally, the pseudospark switch.
The patent literature contains a number of inventions for handling even higher currents with pseudospark switches. US Patent 5,126,638 "Coaxial Pseudospark Discharge Switch" from Maxwell Laboratories Inc. details the use of two nested can-shaped electrodes with an array of holes capable of handling currents of 1 MA. US Patent 6,104,022 "Linear Aperture Pseudospark Switch" from Tetra Corp. describes the use of narrow slots, instead of holes, in the electrodes to obtain currents of 50 KA with a single slot.
I've sketched out a plan for a pseudospark switch using a coaxial array of linear apertures which I think I can build. If I ever actually build it, I'll present the results here.