I am just wondering
what is the (cost + maintenance)/(km sq cover * years in service)
starlink is stated to have a lifespan of 5 years
vs normal 200 -300 foot radio masts?
Only 5 years? Damn that's a waste then. These are literally just disposable satellites. Seems to me it would actually make much more sense to do radio towers. Do a network that's completely separate from regular cellular and has no caps but uses a similar tech. Heck to make it mostly autonomous and not rely on hydro, each tower could be solar powered just like satellites would be. The south side of the tower could have solar panels mounted on it that would power the equipment. One advantage of satellites is that in space the conditions are rather predictable, no winds or storms etc, but if they are only going to last 5 years anyway then that advantage is kind of moot.
The short life span is related to the exact concern that started this thread. Part of the design of the system is that the satellites will rapidly deorbit due to atmospheric drag if not actively maintained, this effectively makes the system failsafe in the sense that even totally dead satellites will automatically dispose of themselves. This also means that even satellites that continue to perform nominally will have a lifespan set by their propellant store. Also, the satellites are designed to be mostly 'demisable', meaning that they will safely disintegrate as they reenter the atmosphere, so there's little to no chance of debris posing a hazard to aircraft or anyone on the ground. Future designs are intended to be 100% demisable.
As to coverage area, consider that a standard cell tower has a range of about 5-70km depending on conditions. Call it 50km, and each tower can cover something like 7500km^2. That means ~1300 sites to cover just the US. Each site requires the purchase or lease of land, and the ability to access that site. Meanwhile, SpaceX say they can achieve "significant" global coverage with only 800 satellites. Moreover, each and every satellite contributes to service across a huge swath of their coverage area (possibly the entire area, eventually, depending on orbital period and inclination), so it's a much easier system to scale up. With towers, adding one somewhere only contributes capacity to that ~7500km^2 service area, but adding a single satellite potentially contributes capacity to a few million km^2 (possibly more once they add mesh networking, so that they can route around congested base stations).
As a bonus, the nature of satellites means that coverage will not be focused on densely populated areas, so Starlink et al could be revolutionary for all of the areas of the world that not densely populated (or wealthy) enough for terrestrial providers to consider building out with infrastructure.