Yeah, it doesn't really make sense to report on "science" PhDs in aggregate. The market for CS PhDs is probably about as closely related to marine bio as it is to poly sci.
That said, I still think that even the "useful" STEM fields you mentioned are lagging behind the professions that can take (as the article pointed out) 1/4 to 1/6 of the time to completion, which much higher completion rates (and less rigorous prerequisites to boot).
By the way, I remain extremely enthusiastic about majoring in hard science, math, cs, or engineering. These majors set you up really well for entry into the workforce. You can contribute immediately in a high impact way, and I've found this notion that you'll be viewed as "merely a coder" to be greatly overstated. There are all kinds of great directions to go from starting out as a developer (including remaining in development). But PhDs clearly don't pay (not surprisingly, you see a much higher percentage of US citizens in undergrad programs in STEM fields, since this is the economically rational choice - at the grad level, the rational choice is to go to law, medicine, or MBA programs, or just get to work and enjoy a salary).