Thursday, 8 September 2011

Don't study computer science

I've been helping out with a couple of admissions outreach events lately. This has brought me into contact with a bunch of bright young minds who are hoping to read computer science at my fair institution. Many of them already fit the industry definition of the hacker - not only a competent programmer, but possessing a deep understanding of, and playful delight in, the way computers work.

But I always feel a tension when advising at these events: I want to seize the best and brightest of these students, shake them, and shout, "don't do it!" Something in me feels like a computer science degree is a terrible waste of a hacker mind.

Of the best hackers I know, an extraordinary number started programming at a very young age - the mode appears to be about seven years old. It is no coincidence, or at least fitting, that this is just below the famous 8-year-old cutoff for natural language acquisition: these are people who speak code natively. They have been practising their craft for over a decade by the time they arrive at university. Even on the Cambridge course, the good ones can probably sleep through the non-mathematical parts of the entire first year, and a fair chunk of the second year too. [1]

Why are these people coming to such a brilliant, exhilarating place and then wasting their time? Wouldn't it be so much more fun, wouldn't you learn so much more, if you did something else for three years? If hacking's your hobby, you'll be doing it anyway, won't you?

What's more, you'd be missing out on a number of ways in which computer science sets you up poorly for the world.

Not least among these is the social component. While I stress that I am talking about trends rather than individuals, I will avoid the careful half-truths of the admissions advocate: as a group, the computer scientists at Cambridge are nerdier than just about anyone else here. One lungful (or noseful) from a CompSci lecture theatre at 12.00 on a weekday will provide a visceral reminder of that. Even the undergraduate mathematicians, those other perennial butts of nerd jokes, are 30% female. CompSci languishes around 10%. [2]

The most sociable ones, of course, will always find other sociable people - often elsewhere, through their college or shared interests rather than through their subject. [3] But this creates an evaporative cooling effect, by which the remaining computer scientists are an even less sociable bunch.

This gets in the way of the normal social learning process. When I observe that a room full of CompScis smells unequivocally worse than a room full of physicists, I'm not just taking a cheap shot at them. Personal hygiene is a baseline indicator of social competence, precisely because it's socially enforced. If that many of these students don't even have friends telling them, "Dude, shower", then how many more are missing out on the more subtle forms of social education that university traditionally provides?

This is very much not to say computer science students are somehow defective and need to be ostracised. In fact, very much the opposite. But an undergraduate course in computer science takes people who share a hobby that involves spending large amounts of time alone, many of whom weren't the popular kids in school, and then effectively ghettoises them together. This does them few favours.

Back to our particularly clever 17-year-old hacker, with 10 years' experience already under the belt. Should we be advising them to apply for a course with a lot of content they already know (or could learn in a weekend if they needed it), where their peers will be on average less socially competent? Or should we advise them to go do something like physics, or engineering, or maths, or cell biology instead?

I could just be saying that because it's what I did. After a school career spent hiding from boarding school behind the shield of a monitor and keyboard, my teachers widely assumed that I would apply for CompSci. But I also greatly enjoyed science lessons. And, while I already knew that computing would be with me for the rest of my life, I reasoned that I would have precisely one opportunity to study hard sciences at a world-class research university. So I applied for Natural Sciences, took in some physics and physiology on the way, and ended up as a neuroscientist[4]. I loved it.

But it turns out that an appreciation of this reasoning is already embedded in the industry. I remember confessing my degree plans to an established hacker at a Debian Linux meet-up I attended at the age of 16 or so [5]. I thought that I was being slightly subversive, even perhaps insulting him by disparaging his field, but his response was nonchalant: "Oh, that's what the best people do. NatSci undergrad, CompSci PhD if you do one. It's the stylish way."

Computer science, it seems, could easily slip into the same category as business: something whose best practitioners often learn by experience, and then perhaps catch up with the formalities at a postgraduate level.

As for the desirability of computer science as a research discipline, that's another rant for 3 AM on another morning.

[1] I actually had a friend from school who was sufficiently far ahead in maths when he arrived that he literally slept through the mathematical bits of his course for the entire first year. While I envied him his morning lie-ins, I suspect he was an exception. He was no fool, though: although a passionate and skilled hacker, and virtually certain to choose a career in the computing industry, he chose NatSci Physics to give himself a challenge.

[2] I'm aware that this sentence opens up a whole can of gender-role worms, and that I'm deliberately avoiding tackling them in this post. As a summary, I will say that in our society, the distribution of girls and women to subjects and hobbies seems to correlate fairly well with the perceived sociability and/or social acceptability of its participants.

[3] This is not an option at non-collegiate universities; I don't know what they do there. Possibly sports teams, or other hobbies with broad popularity, or perhaps it's just that fewer of them escape.

[4] I even nearly ended up spending four postgraduate years sticking electrodes into petri dishes. Vital work, and someone has to do it - but the more working biologists I know, the more I feel I dodged a bullet.

[5] Yes, you can still find a photo of me at that meet-up. No matter how hard I try, a spotty teenaged version of me is still one of the top Google Image results for my name.

No comments:

Post a Comment