I filmed the ATV 3 as it passed over Leiden, the Netherlands, in twilight this morning.

The video can be seen here:
http://sattrackcam.blogspot.com/2012/03/footage-of-atv-3-passing-in-morning.html

The spacecraft is quite bright, easily visible naked eye in a bright blue twilight sky.

A problem for Mr Popovkin's blame-game, is that Phobos-Grunt did not pass over Alaska until several hours after it failed.

1) Along with the iPad's flat shape, the metal rod and the cylindrical thing (the GPS tracker?) underneath it acted to weigh and stabilize it. Look at the video: after some initial tumbling, it stabilizes and takes a flat, horizontal attitude throughout the fall. The bottom of the iPad is always level with the horizon.

As a result, the flat iPad starts to create it's own lift-effect, slowing the ascend down. It starts to glide, kind of like a falling leaf.

Analyzing the the video, especially the imagery of the last second before impact on a frame-by-frame basis, you can see that the ground upon impact is not smeared in individual frames, but visible in all detail. This simply shows that the speed upon ground impact was not high at all. It makes a relatively gentle, gliding landing with a speed of at best a few meters/second.

2) As others remarked, it did not drop from space. 30 km is nowhere near the boundary of space. Even (a few, military) aircraft can and do fly at 30 km altitude. It's airspace, not Space.

The authors propose a link in their paper to fragments of comet 12P/Pons-Brooks.

This is nonsense however, as pointed out here: http://sattrackcam.blogspot.com/2011/10/ot-1883-zacatecas-observation-of.html

The earth has its closest approach to the 12P/Pons-Brooks orbit near December 6th, not August 12th (see diagrams in the link above). Hence, fragments of the latter cannot pass close to earth mid-august (and they do not come even particularly close on 6 December, as the minimal earth to comet orbit distance is still 0.2 AU, i.e. the comet passes closer to the orbit of Venus than to the orbit of Earth).

The whole story has very little substantial fact behind it, and factual errors such as pointed out above do not promote confidence.

Actually, the human eye is still beter at detecting trails on plates than automated systems are, especially where fainter trails are concerned. Automated system also have serious difficulty discerning between real NEA trails and trails from cosmic ray impacts on the sensor.

Spacewatch used automated detection, nevertheless human inspectors discivered 43 additional asteroid trails on the images between 2004 and 2006: trails that the automated routines missed.

Interesting as it is, this is not a "first". The Spacewatch program "crowdsourced" the search for NEA by using volunteer plate inspectors between 2004 and 2006 (the Spacewatch FMO project) and discovered 43 new NEA this way (http://fmo.lpl.arizona.edu/discoveries.cfm). I personally discovered 2005 GG81, a small Amor asteroid, as a volunteer plate reviewer in this project.

It currently is in a 198 x 345 km orbit inclined 42.8 degrees.

Actually, many/most countries are signatory to the Space Treaty that specifically states (amongst other things) that any space debris landing on their territory has to be turned over to the country who launched it, if the latter wishes so. So yes: by international law, UARS remains US property.

But of course, the Russians may always decide that the last Soyuz currently attached is "safe" after February 2012 after all.