Just on the face of this, it’s a cool idea, but it’s one of those
things that didn’t even seem possible within a magnitude, so of course
a few minutes of Googling was in order. Given the ecologically
destructive effects of oil and coal burning, there would be no way
that we would be doing it if solar were a reasonable option.

For the year 2000, the U.S. consumed about 3.8
trillion kilowatt-hours (kWh) of energy. You need about a 100 square feet
of solar panel to provide 1 kWh. The amount of energy generated
over the year depends heavily, of course, on where the panels are
installed, but let’s assume that we’re going to put them in Texas,
which means that we’ll get 1,800 kWh a
year.

So, working this out means that we get:

3.8×10^12 kWh / 1800 kWh * 100 ft^2 / 5280^2 ft

… which is about 87 square miles on a side. Still enormously huge
and enormously expensive, but just in checking the facts I’ve gone
from “Oh, right, there’s no way” to something that is within a
magnitude of reasonability. A magnitude, really, is just
an engineering problem. There is no doubt that it would be
totally unreasonable to make this many solar panels, there would be
shortages in gallium arsenide and other components used for solar
panels today. On the other hand, these figures are predicated on 10%
efficiency, whereas we have experimental technologies that are getting
in the range of 30%.

This is a story that is going to get bigger: Gregg Easterbrook
rails against Lilly^[1] in the New Republic, claiming
that they have led to the death of a healthy 19-year-old student in
Indiana. He makes some salient points, but is probably over the line,
especially by his last few paragraphs — see Corante for a good
analysis of the column^[2]. Nonetheless, I think the
Easterbrook does have a point.

A number of relevant facts:

• Why were they testing on healthy people? The FDA requires such
  tests to study side effects, specifically, the metabolization of
  antidepressants^[3]
• Why would students participate in such studies? For the extra
  money — Traci Johnson, the student that killed herself, was
  getting paid $150/day + meals^[6], which is really
  good money for a student
• From the New York Times article: Robert Smith, a Lilly
  spokesman, said the company did not believe that duloxetine, the
  drug’s generic name, caused the suicide. “This drug has been studied
  in 9,000 patients, in depressed and nondepressed healthy people, and
  we have not been able to discern any signal between duloxetine and
  suicide or suicidal ideation,” Mr. Smith said.. I could not find
  the studies totaling up to 9,000; the one study that I did
  find^[8] gave a rate of 1 suicide attempt per 115
  patient-years, which is not abnormally high for depressed
  patients. (I think that Easterbrook’s complaint about the suicide
  rate in test subjects (5/9,000) being too high is specious, as Corante
  points out; the question is, what is the general suicide rate among
  depressed patients? It works out to about 2% over the lifetime of the
  patient for patients that have been treated for
  depression^[11], but I have not yet found year-to-year rates
  by age group)
• Traci was one of 25 healthy students that were taking place
  in the study — given the small number of students, it makes the
  probability of some relationship to duloxetine that much higher
  — the average rate of suicide for people aged 15-19 is 9.7 per
  100,000, with a 5:1 male:female ratio. This means that the chances of
  1 out of 25 healthy students committing suicide, on the average, is
  0.25%; this would be even lower if you took into account the different
  suicide rates in females vs. males. So, statistically, something looks
  really suspicious about this.
• Antidepressants have been, in many cases, counterindicated for
  anybody under 18. The theory is that developing brains react to the
  drugs differently than adults. Traci Johnson was 19 at the time of the
  study

The Indianopolis Star claims that the suicide rate for patients was
only 0.097 percent, whereas the rate for SSRI antidepressents is 1.5
percent. This figure is suspiciously low (it’s two magnitudes lower!)
and I suspect a division error by the paper; it doesn’t match the results
in the Journal of Clinical Psychiatry^[8], which gives a
suicide attempt rate just under 1%^[9].

Financial implications. Lilly had total revenues of 12.5B
in 2003^[5], and estimates for duloxetine ranged as
high as $2 billion in 2008^[3]. Lilly’s patent on Prozac (fluoxetine
hydrochloride) is gone and they are getting quite a bit of competition
from generics, so they really want a new, patented medicine in their
array of drugs.

Conclusion. I think it will all boil down to one thing: Dr.
Joe Glenmullen, a clinical instructor of psychiatry at Harvard Medical
School, said that “patients taking antidepressants are most vulnerable
to suicidal thoughts three to 10 days after they cease the
medication.”^[3] (also mentioned in the New York Times
article^[6]). If this is true^[7], then the
question is: was Lilly conscientious in making sure that students
taken off the drug were tracked? If not, look for a major negligence
law suit.