I worked a earth imaging system (EIS) back in the 1990s. I didn't have security clearance (almost got it and realized it had no benefit to me but a bunch of crap I'd have to do, or not do/say, so I passed).
I'm pretty sure that the satellites at the time could read 3 inch newspaper print under the right conditions which happened rarely.
I've got a tack sharp 600mm Canon lens and it's sort of useless for a lot of stuff. If there is any haze, heat, dust, whatever, in the air, all that expense glass sees is that instead of what you want to see.
That note aside, it was a fun project. I was the I/O guy, I did this work:
http://www.nfsv4bat.org/Documents/ConnectAThon/1996/bds.pdf
and I got Seagate to redesign part of their 9gb (I think, might have been 2 or 4g) barracudas. I've got a benchmark in lmbench that shows how the disk performs as a function of seek distance, looks like this:
http://mcvoy.com/lm/bitmover/disks/seek.gif
The lower edge of the band is what you get if you seek and get to the track just as the sector you want is to about to pass under the head;
the upper edge of the band is what you get if the sector had just passed under the head;
the height of the band is a rotational delay;
and those outliers? In this case I think they were either bad blocks or I don't know.
But when you ran this benchmark on two drives, mounted in a rack right next to each other, you got tons and tons of outliers which blew any chance we had of meeting the performance metrics. I bitched at Seagate and they hemmed and hawed and finally admitted there might be a problem with their internal mounts.
The problem was that their mounts were so useless that the vibration caused by one head moving vibrated the drive next to it enough that the other drive's head didn't settle properly and you blew a rev waiting for it to get where it needed to be.
Seagate redid the mounts.
Fun project.