There seems to be an assumption that with enough tech, you could count the hairs on somebody's big toe. As with ground-based astronomical telescopes, the two things at work to ruin the image are atmospheric refraction and aperture. The bigger the aperture, the better the theoretical resolution, but the more likely a convection cell in the atmosphere will blur the image. In practice, resolution of a few arcseconds is the best that can be achieved on all but the most exceptional days. The theoretical limit is defined by the wave nature of light (diffraction) but is rarely achieved directly.

Even so, let's assume these spy satellites have diffraction-limited optics and are looking down through a vacuum. Let's assume a 20-inch objective. Dawe's rule states that the theoretical resolution is

4.6/20 = about 1/4 arcsecond

Now, what does this translate into in terms of sizes of things on the ground? Assuming a vertical lookdown from low orbit (150 miles), we have

Circumference of circle centered on the satellite = 2 * pi * 150 miles = 5 million inches

1/4 arcsecond = 1 5 millionth part of a circle

So in theory, things as small as an inch across can be seen from 150 miles up with a perfect 20 inch telescope with no distortion from the atmosphere.

In practice, this is probably more like half a foot. Pretty good, but not enough to read the paper.

1) 10 cm = 4 inches :)

2) The Hubble would not work if pointed at Earth. It's extremely sensitive to infrared radiation. Plus differential heating would ruin the performance.

3) 20in is a *big* telescope. I'm assuming military satellites use Maksutov-type closed-end optics. Putting up, say, a 40in telescope would be an enormous undertaking on the level of the Hubble scope. I doubt a corrector plate that big, that did not go geometrically apeshit when moving in and out of the Earth's shadow, could be made.