On Dec 1, 4:41*pm, wrote:
"The Pentagon loves its giant satellites — the
bigger and more expensive, the better. The
culmination was Keyhole -12; at an estimated
twenty tons, it’s believed to be about as large
as the massive Hubble Space Telescope. But
these orbiting behemoths are increasingly
coming under criticism by those who believe
the job can be done better, faster, and cheaper
by constellations of small satellites. The new
catch-phrase is Operationally Responsive
Space Access. This is the source of the
Army’s Kestrel Eye spy satellite program,
which is exploring the possibilities offered by
smaller sats."

See:

http://www.wired.com/dangerroom/2009...sats-of-their-...

Is this a good idea? *Or would it be better to invest
the money in more (and better) UAV's?

Well they do make tha argument that

"As with other small satellites, the key is to make it cheap. Kestrel
Eye is supposed to cost around $1 million per satellite. That’s a
bargain, considering a Predator drone is about $5 million. Even a
constellation of thirty of Kestrel Eyes will cost a fraction of a
large satellite."

Of course that begsthe question of whether they include launch costs
there. Maybe one rocket with a whole bunch of these, like a giant
shotgun pack style carrier bus? Or piggybacking on other launches?

Either way, not sure about the launch cost, but one edge they have
over a Predator is they are already up, and instead of waiting until
one gets flown to a position or worrying about fuel state, one can
just aim the camera to a certain location, and depending on orbit
patterns, have it on the point of interest quickly and for a good
duration of time. If you have enough, you can keep passing off to the
next one in line too.