Alan,

You're absolutely correct about the inertia thing, I use this word as a "lump sum" of all the various parameters that you described.

The change you are looking for that leads to filament failure is mostly in the mechanical dimensions of the filament. Over time, part of it vaporizes and the imperfections in it make it weaker until it burns out.
Take the time and check out this internet page, found googling for +light +bulb +failure: http://invsee.asu.edu/nmodules/lightbulbmod/burnout.html
Have a look at the pictures of the aged filaments - you can see the "corrosion".

Frequently switching a bulb on and off will accelerate this process, arcing is nothing else than a fast, more or less random switch that additionally switches to on - usually not at zero crossing but at substantial voltages (that can gap the bridge between the two contact surfaces) so the inrush current is even higher than usual, accelerating the evaporation process even further.

And, to conclude: In a lighting circuit, arcing should not produce high voltage spikes. Please PM me the $64,000 check [Linked Image] and thanks for the interesting discussion!