This is interesting, when I studied CS at university I'd say there were 2 different types of people that got high marks.

There were people who were self taught , either before they started CS or once they learned some programming at university they identified other areas outside the course that interested them and they could apply their new programming skills.

These people typically got jobs in the software industry after graduation and become software developers.

The other group of people were people who were just generally high achievers and learned enough programming to pass the course with good marks but nothing much else. They got equally high marks because they were good at passing exams.

Most of them either retrained for a career in finance , went into academia or got management gigs at tech consultancy companies.

I can't think of anybody I know who is a working programmer who is not self taught to at least some degree.

> They got equally high marks because they were good at passing exams.

I don't understand how this makes sense, unless being "good at passing exams" means "cheating". Can anybody explain? I hear this said so often, and I usually chalk it up to the speaker rationalizing his own poor scores.

My CS exams were always hard, and the only way to "get good" at passing them was to learn the material.

Different types of learning. I got very high grades in all my courses during college, but a number of them were due to the fact that I am really good at cramming and figuring out what to study and what to ignore. That's what it means to be good at passing exams.

Most of my CS courses I actually spent the time to fully understand and internalize the material. That was the material I truly learned.

But when you write an exam, clues to your mental state are all throughout the material you write on the exam, and that mental state includes everything you know. Someone who's "good at passing exams" can extract clues to your mental state from the wording of the questions and figure out what kind of answer you want. On multiple-choice exams, they only need to come up with 2 bits of information about which answer to choose, and the proposed answers themselves give additional clues to the exam-writer's state of mind. 2 bits doesn't feel like mastery; it feels like an educated guess.

It depends on the subject area, of course, and the skill of the exam writer. But it's very common to be able to pass multiple-choice exams without knowing anything at all about the material.

Knowing the material, of course, helps you come up with the right answer --- but it also helps you a lot with "reading" the exam writer. And you don't necessarily need to know a whole lot about the material to get an advantage that way.

It's also often possible to get acceptable marks on exams by parroting rather than deep comprehension.

I think my test-taking skills were usually worth one to two letter grades' worth on exams when I was in school. I could usually get a D or C on exams where I should have gotten an F, and an A on exams where I should have gotten a B or a C. A little while back, I got 97% correct on the ai-class final exam without having learned more than half of the material. (In that case, though, I think the test also failed to cover most of the material.)

I think non-multiple-choice math exams are probably the hardest to "fake out" this way.

There are other people whose test-taking "skills" actually have a negative effect on their scores. First, they study the material in their bedroom or at the kitchen table, rather than the classroom, unnecessarily impairing their recall when the exam comes. Then, they show up to the exam exhausted and sleep-deprived from cramming all night, damaging their ability to think creatively or tolerate stress, and then they have an extreme stress response from the test-taking situation, further handicapping their ability to think. It's easy to imagine that someone like that could fail a test I'd get an A on, with the same level of knowledge.

"Someone who's "good at passing exams" can extract clues to your mental state from the wording of the questions and figure out what kind of answer you want."

I actually tried an experiment on this when I was in high school. I took the AP Comparative Government without ever having taken the course, or really having any sort of academic exposure to it (hey, it was free with the purchase of the AP US Government, and I was taking the day off from school anyway for the latter test). My only knowledge consisted of what I read in the newspapers, plus half an hour with a test prep booklet at breakfast that morning, plus whatever I could glean from the test questions themselves.

I scored a 3 on it. Not a great score, but passing. Pretty good, actually, considering that the test involved writing 4 essays on a subject I knew nothing about. So I figure perhaps 50% of the outcome of a test is knowing the material and the other 50% is test-taking skills.

Ironically, though, I think that the skill of extracting subtle clues to the mental state of the people around you, and the answers they expect, is far more valuable than any subject matter you learn in school. It's absolutely essential if you work in an organization, so you can understand who the decision makers are, what their priorities are, and what will really impress them without them having to tell you anything. It's absolutely essential if you manage people, so that you understand why they're working for you and what will motivate them to do their best work. And it's absolutely essential if you choose to strike out on your own and be an entrepreneur, because that's how you tell what customers want. They're generally indifferent to you and often have no clue what they actually want; there's no way you'll get them to tell you.

In my experience, some people are better are forcing themselves to be interested in things.

Interesting... if true, what this could mean is that virtually no-one who is taught programming at school ends up enjoying it. You have to be one of the ones who seeks it out before you are required to learn it, otherwise you're not the kind of person who would tolerate it as a job?

I don't think that's really true. I knew someone in college who had never touched a compiler before her sophomore-year intro CS course. She ended up graduating with high honors, went to work at MIT Lincoln Lab for a few years, and is now doing a Ph.D in CS.

I know several other people here at Google that went into college studying things completely different from CS, took a couple courses in it, and discovered they loved it.

I think the real determiner is what happens after college, after the academic support net is taken away. The people who go on to become really great programmers use it as a springboard to start seeking out information on their own. You can always tell whose these are once you hire them, because they'll ask you several questions about the system to orient themselves, get some code & documentation pointers from you, and then go about their merry business learning everything they can, including tons of stuff you didn't tell them (and often, didn't know yourself). The mediocre ones learn just enough to accomplish the task at hand and then ask you again as soon as they have to do something new.