So, I've been wanting to write a bit about why grad school and academia in general isn't necessarily a waste of time. Unfortunately I'm now a white male in a power position that depends on me sucking good people into grad school and then exploiting them, so this may end up sounding entirely self-serving... not much I can do about that, though ;)

Grad school, for me, was not the way obnoxious gits like Phil Greenspun and Mencius Moldbug portray it. I didn't do computer science -- my PhD is in biology -- so that may be the difference; but Phil's argument seems to be that money not only is but should be inextricably linked to job satisfaction, and I can tell you that I'm not planning to get paid all that much anytime soon. Mencius (that can't be his real name!) says, "...if anyone with a forehead steeper than a halfpipe has ever described their work as, for example, "grant-savvy," I am a Cape buffalo" -- but this is so breathtakingly stupid that it must be a straw man setup. Mencius also says "...in the first couple years of grad school [... ] your basic purpose in life is to learn all the things they forgot to teach you as an undergrad." I think this is an interesting point, but I don't know of any PhD programs that are like that; presumably he's talking about Masters?

Where I think Phil and Mencius are both missing is this: grad school and beyond are useful places for deep learning of the type that you simply will not get by working for a business, even Google. Now, like any statement about people, that's a 95/5 statement: 5% of grad-school-worthy people will do fine if you parachute them into Google at the age of 15. The other 95%, though, will miss out on the variety of technologies, opinions, approaches, and (most importantly) hard work that you will be exposed to in grad school.

Let's focus on that last bit, the "hard work" bit, for a second. I only know of a few companies in the world where you might be regularly pointed at unsolved, and often unsolvable problems, and then be told to make progress on them. This happens to every graduate student that goes to a good grad school. And, while I know this is an unpopular opinion, you can only really learn with the kind of in-depth hard work you are forced to do as a graduate student. That's part of deep learning: being challenged in a serious, systematic way.

The other part of deep learning is being exposed to a variety of hard problems from a theoretical point of view. Every day online I read about fascinating software developments that originate in deep CS soil; that people don't recognize this speaks more about their limitations than about the irrelevance of CS research. (That's a topic for another blog post, I think.) You cannot easily pick this up by reading blog posts by ADHD millionaries like Phil Greenspun, whose company failed in part because he himself didn't know the first thing about developing maintainable software (another research interest of mine -- but I digress).

So, I agree in principle with Leah Culver when she says that a computer science degree is useful.

Personally, I'm most excited about something else entirely. I used my undergrad math degree and my biology grad school training to get into two of the hottest and neatest research araes around, comparative genomics and metagenomics. There is shit going on in these fields that is so bleeding edge and interdisciplinary that I can't even seriously contemplate explaining it to people who don't have PhDs in biology, 3l33t skillz in programming, and a solid background in computational science.

And I'll pay you (poorly, it is true) to come work for me on it.

And, fundamentally, that's the bargain you make when you come to grad school: poor pay in exchange for in-depth larnin' on topics that you probably never even heard of before your first day in grad school.

I think it's a fair exchange.

YMMV.

--titus