Not completely sure - BUT IT DOES!!!!
My theory on it.... The arcing contacts while not seemingly intermittent causes an imperceptible switching. This 'switching' causes the tungsten, or in the case of the POS CFL - to have recurring starting currents. These starting currents in the CFL is just destroying the POS little ballast in there. In the case of the regular lamp it constantly bounces the filament until it gives way by way of metal fatique... Well that's just what I think...
But proof be positive - getting better contact on the lamp itself with a clean firm contact - corrects the ridiculously short lamp life phenomenon. Go figure....
So we are saying that a 120 volt bulb will last 60 times longer powered by direct current (DC) than if it were powered by alternating current?
remember on AC it already turns on and off 120 times a second.
The filament in a lamp does have a lower resistance when power is first applied, but after it has reached the temperature of luminescence's not much would change until the filament were to cool off, this should mean that a lower frequency would be required for it to do this, very much lower in fact, but at the same time the filament would receive a very much lower RMS voltage, thus counteracting the extra starting cycles, many two position dimmers just use a diode to cut the sine wave in half, thus only providing a 60hz pulse instead of the 120 pulses the filament would normally see, as far as I know this has not affected the bulb life and maybe extended the life of the bulb in most cases, so I cant see where a higher or lower frequency would shorten the life of a bulb unless it so low that it allows the filament to cool below a certain temperature? which would mean it would be so low it would pulsate.
What I can see is a high resistance connection in the base causing a higher temperature on the wire holding the filament, this could cause a failure?