HOWEVER, upper-stages to move satellites to GTO are ejected out of earth orbit (for safety reasons - you don't want it to add to the space junk problem). These ejected stages are taken through an Earth-Moon Lagrange point, before being navigated on a low-energy path to a heliocentric orbit. There's no reason a microsat can't hitch a ride on an upper stage, and then wait until the ejected stage reaches the L-point before deploying itself, at which point it's downhill to anywhere in cislunar space.

People are currently working on doing this: http://www.lunarcubes.com/

Even if you were launching only microsats (without any other payload), I'm not sure the cost per satelite would be that high? Wikipedia (http://en.wikipedia.org/wiki/Comparison_of_orbital_launch_sy...) says the cheapest option right now is $2200/kg for LEO, so about $200 for a 100g phone. I guess going to the moon is about 10x more, but that's still not unreasonable...

You don't buy payload space by the kilogram. You buy it by the rocket, with the smallest launches costing millions of USD each. That per-kilogram figure is just the number that comes out when you divide the launch cost by the maximum theoretical payload capacity.

Of course. The proposal is to buy a rocket to send a large number of small satelites somewhere, and split the cost among the satelite owners.

SpaceX's Grasshopper project is likely to make the price come down a lot RSN.