Although I agree with nearly everything he says in this essay, it's worth pointing out that he has written several other pieces that would make most normal people question his grip on sanity. At the very least, this guy is
eccentric. Also, the numbers he cited are from 1999, and are getting a bit dated. NIH post-doc stipends are now roughly $40k/year (including in the most expensive cities, like San Francisco); grad-student stipends are around $25k per year.
Everything else is true, in my experience. If you start a PhD in the physical or biological sciences this year, you'll spend 5-10 years in grad school, barely making ends meet at $25k a year. Then, if you're incredibly lucky (think 95th-percentile lucky), you'll get a stellar paper that will set you up for a post-doc in a superstar lab, which will give you a roughly 50/50 shot at a tenure track position somewhere (don't count on living in a city, or even a place where your significant other can find a job, however, because you're now living at the whims of the academic system.)
Your post-doc will take 3-4 years, and will pay you less than you could have made right out of an undergraduate CS program, in exchange for more risk and longer hours. You'll be expected to work nights and weekends, and in case you lack the drive to do it, there will be some guy from China with a J-1 visa sitting next to you. He's probably living in a boarding house right next to the university with ten other post-docs, and he's willing to work 100-hour weeks for a shot at a job in this country. That's your baseline.
If you're astronomically lucky, you'll get a faculty position -- hopefully not at Podunk U -- where you'll be paid significantly less than that undergrad CS major (who is now probably making over $100k/year, buying a home, getting married, etc.) You'll be competing for grant money in an insanely tight market (the funding rate for NIH research grants is well below 5% now), and evaluated by established faculty who will torpedo your proposals if it helps their own chances. At the end of six years of insane work, you'll get a shot at keeping your job. If you get tenure, you can settle down to a lifetime of below-market wages, while working relaxing 50-60 hour weeks.
Of course, if you choose to leave academia after grad school, the prospects are equally dismal. The rare position that requires research experience is essentially set aside for friends (you hear a lot about "networking" on the science job boards, but little practical advice on how to actually do it), or for people who have been poached from other companies by recruiters. Meanwhile, once you have a PhD, you're untouchable by most employers -- considered too theoretical and expensive for "practical" work. Ultimately, most people with advanced degrees in the sciences spend years teaching courses on contract, or jumping between dead-end, low-paying jobs, hoping to start something that resembles a career.