What's the "simple verdict" on it--is there simple math?
As 1km^2 = 1,000,000 m^2. So taking the power usage, we divide by solar flux, by number of hours of high solar input, by 200 W / m^2, by area conversion, and get approx 3,000,000 km^2. Since the USA is 10,000,000 km^2, we would need to cover 1/3 of the country with solar panels to generate enough electricity, mostly down south where the sun shines the most. Alaska would be spared.
Since we use power 24 hours a day and can only expect to generate at peak rate for 6 hours, we need to store the power in batteries. Not to mention that not every day is sunny. It sometimes rains, panels get covered in dust and need cleaning, stuff breaks down, etc.
A Tesla PowerWall stores approx 120KWhr / m^3. So 4^15 / 120^3 = 33^9 m^3 just for the batteries, not including cabling, racking, etc. To get an idea of how this compares, Trump Tower in Chicago is approx 80,000 m^3. So we'd need 400,000 of them to house the batteries and this only for one day's worth of power.
Of course, none of the foregoing considers the power required to make the solar panels, the mounting hardware, the batteries, the housing and new power-lines. The math just gets more complicated. But if you were to consider every step from beach sand to working solar panels you will find that the lifetime energy produced by solar panels is less than that which was used to produce them in the first place. I read an article to this effect once, but can't find it. Google is your friend if you are interested to research further.
For battery space, the same factor of a few hundred still applies. Also I'm not sure where you looked up the volume of the Trump tower but wikipedia claims a floor area of 240,000 m^2 and I doubt people can walk in an area a foot tall. With a more realistic ceiling height of 3m, you'd be off by another factor of 10 there.
--> Closer to 1/1000 of land area for coverage and 100 trump towers for storage.
Your claim on it taking more energy to make a solar panel than it produces is likely untrue, but I don't have enough scientific background to quantitatively refute it. (http://www.nrel.gov/docs/fy99osti/24619.pdf claims a breakeven period of 2-4 years)
Yellowstone national park is 1/1000 of the US' surface area.
From your document:
"Purifying and crystallizing the silicon are the most energy-consumptive parts of the solar-cell manufacturing process."
"To calculate payback, Dutch researcher Erik Alsema reviewed previous energy analyses and did not “charge” for the energy that originally went into crystalizing microelectronics scrap."
The way I understand it, the calculations don't charge the most expensive part of the process by arguing that, being scrap of the semi-conductor business, its already been paid.
My most pressing question is how the capital costs are factored in. China built a huge PV capacity and arguably that made the price of panels collapse. Also, the semiconductor industry had been indirectly subsidizing the PV industry by upgrading their old fabs.
I also really don't understand how PV can be cheaper than Concentrated Solar Power < literally just a bunch of mirrors aimed at a normal thermal plant.
Lot's of funny numbers, lots of questions, few open source gitted models available to play with.