Putting too much stock in coherence. Sunlight is not coherent, but throw enough of it at a target, and things happen. Little moments of constructive interference are enough for an effect.

From your article:

> Note that the thinned array curse applies only to mutually coherent sources. If the transmitting sources are not mutually coherent, the size of the ground spot does not depend on the relationship of the individual sources to one another, but is simply the sum of the individual spots from each source.

Sun, all the points are about the same distance from you. They all contribute about equally. Well, all the ones over the horizon, but anyway.

Satellites, well, first, most are below the horizon at any given moment, and unlike with the sun this doesn't just halve the number of sources visible. But let's pretend the earth is transparent. The satellite closest to you is four times brighter than the ones twice as far away as the closest, and the distant ones are about (525km/(12742km + 570km))^2 = 0.00156 times the brightness of the closest.

The point I'm making with the thinned-array curse in response to a comment about their phased array antenna is that the scenario is a thinned array, and that phased arrays, while useful, don't solve everything.

From the shell theorem, the sum of the effect of of all of them at any point inside their orbital shells can be approximated as* a constant value everywhere equal to the effect if you put them all in the middle and measure at the shell radius, so radius = 12742km/2 + 540km = 6911 km, so 8000 * (540km)^2 / (6911km)^2 ~= 48.8 times the brightness of a single satellite. So even all of them is not much.

* Approximation will diverge noticeably roughly when you're closer to the shell than the separation of sources in the shell