How do you know the hits aren't cosmic rays? It seems a few hits per hour could easily be cosmic rays, unless they are distinguishable somehow.
A typical camera target is about 3 mm on a side, for a total sensitive area of ~1e-5 m2. The muon flux at sea level is about 1 per cm2 per second, or 36 million per m2 per hour, or 360 hits per hour on a camera target, so even granted that the "active" area of the target is only a few percent of the physical area, muons more than account for the signal he is seeing, and muons deposit a lot of energy.
The electostatic concentrator is an interesting feature. Charged nuclei typically attach themselves to dust motes in the air, which are then swept along by the E-field. It isn't clear why he needs such high voltages to get 50 V/m over a few centimetres, and in my experience Cockroft-Walton generators are easy to build but hard to make work. With such tiny capacitors any stray leakage current would destroy the effect, and he does not anywhere measure the voltage produced.
The enhancement of the count rate could be accounted for by electrical noise from the CW driver circuit.
This is pure speculation on my part, based on a good deal of experience designing and building radiation detectors, but I'd like to see a lot more evidence before I said this was an effective radon detector.
An alternative, easier to debug and demonstrate, and more likely to work approach would be to use a commercially available activated zinc sulphide sheet with the camera target as the detector: http://www.eljentechnology.com/index.php/products/zinc-sulfi...
But I guess the main goal of the article is to show that, even though "radiation detector" sounds like a complicated thing to most people, or at least requiring some nontrivial components, it is actually possible to get started with DIY experimentation with very few and simple components and little effort. You might indeed get better results with ZnS:Ag phosphor sheets, but by showing that it is possible to build at least something with just a webcam that people already have you massively increase the likelihood that the reader actually starts to build and experiment. And when people already start experimenting, then many of these potential problems get incrementally noticed, figured out, and removed, and all kinds of cool projects can result. But yes, it would, indeed, have been nice if the article would have provided more information about testing the detector against various potential interferences and problems.
[0] http://spie.org/Publications/Proceedings/Paper/10.1117/12.20... (paywalled, though)
I think you are overestimating the cosmic muon flux quite a bit, see the particle data book for the discussion of cosmic muons.
http://pdg.lbl.gov/2012/reviews/rpp2012-rev-cosmic-rays.pdf (page 5)
Also ionization loss by fast muons is pretty low, it's in the "minimum ionizing particle" regime, and most cosmic muons we see "down here" are in the >1GeV/c area.
http://pdg.lbl.gov/2013/reviews/rpp2012-rev-passage-particle... (page 4 is muons on Copper)
But again, I think your doubts about the numbers is certainly warranted. With a experimental setup needing such a long exposure time to yield statistically usable numbers, I think one cannot really prove a lot. Certainly a detector more suitable to the task of counting radioactive decay, such as a calibrated large-volume geiger counter, would be better. At least one should run both setups in parallel for an extended period of time to be able to do cross-corralations.
[0] http://en.wikipedia.org/wiki/Uranium-238#Radium_series_.28or...
edit: re-reading I found this "In order to make it more sensitive, I added an electrostatic concentrator to capture one of radon’s alpha-particle-emitting daughter products"