Your smugness disagrees with me. In my personal experience, I covered the basics - analysis of algorithms, graph theory, the usual data structures, operating system architecture, parsing & state automata (DFA, NFA, PDA etc.) - long before I went to college. You risk tarring all autodidacts with the same brush.

And let's not forget, there are plenty of CS graduates who conversely aren't very good at programming.

Well I have a CS degree, and I value it quite a bit, but honestly it's a second order concern in industry. The first thing is, can you write and debug code in a production environment? This is something that university just doesn't teach you; as long as you have the intelligence to understand the theory, the professors will prod you in the right direction on the assignments.

Now, once you can code your way out of a paper bag, then the question is do you have the CSCI fundamentals to apply the best known solutions to difficult problems. Facebook has their pick of the litter, so it makes sense that they might spend more time validating CSCI knowledge, but for most companies that are hiring this would be a very poor metric to go by.

While this is true, I have never needed to explain CS stuff in an interview, it is also true that very often when faced with a tough algorithmic problem even the little CS foundation my few years of formal education in the field have granted me helps quite a bit. If nothing more than the fact I know what to google for.

It's surprisingly easy to google for solutions once you know what you're looking for instead of searching descriptively :)

Ha, you said it man. Looking back I began to realize that this was the biggest benefit of my degree.

Technically, a homeless person could dig in at a terminal in the library and teach themselves anything they needed to learn about computers. However, first they'd need to know what to look for and by God that was difficult before I spent 4 years talking to people who worked with computers all day. I think my time on Reddit actually enriched me there a bit, too.

You figure out what you need to learn by doing something, which is why working on a project is so nice.

In today's day and age, it's true that Google will likely lead you in the right direction. However there really is no more efficient way to get a mental lay of the programming land than a quality CS program. Dynamic programming algorithms for example, are something that very few will ever stumble into on their own. You might get lucky when Googling, but it's not the same as having a mental model of the breadth of CS fundamentals.

Sure, but I still think it's more efficient to learn while you work on something real. When you do this there is less waste on certain topics you never encounter, and because you need what you're learning it's better absorbed and understood. It's just this method is less practical for employers since they just want someone already equipped with what they need.

Throughout my career, I've met many developers who couldn't code their way out of a paper bag, network engineers who couldn't network their way out of a paper bag, sysadmins who couldn't sysadmin their way out of a paper bag, etc.

I propose a new interview question: How would you code your way out of a paper bag? You will be given several constants, such as the strength of the paper bag, difficulty of tearing, etc, and will be expected to write your own Tear() function and call it properly.

The best part is that when the recruiters/headhunters ask why you rejected a candidate, you can honestly say "he couldn't code his way out of a wet paper bag."

I have a degree... in philosophy.

Everything I know about theoretical computer science I picked up from books, at a price far below the tuition of a typical university.

Which you picked up from books after getting a degree at a university that most likely taught you (or at least improved) your ability to think and learn. I think it is a bit disingenuous to say that you learned CS for far less than the tuition at a typical university after you got a degree at a typical university.

Well, to be pedantic, I didn't get a degree at "a typical university", I got a degree at a really expensive private college (courtesy of a scholarship which made it not so expensive to go). I got there by being raised to read a lot, engage in critical thinking, etc., etc., which skills were then honed over the course of four years.

Anyway, college was a boost, yes, but my main point was that one doesn't need a CS degree (and, honestly, there aren't many CS degrees out there anymore -- just Java-programming degrees masquerading as CS) to be able to do this stuff. Any sort of good liberal-arts background, even one not involving an actual piece of sheepskin from a four-year institution, should be all anyone needs to pick up CS or just about any other subject.

But the merits of a real liberal-arts education are a whole 'nother thread.

I've considered this. (Getting a Phil deg. altough I'm a geek and want to work in a technical field). Any pointers maybe? Is this a viable route?

On the flip side of that, I once interviewed a candidate for a summer internship at my tech company in San Jose. Our first question was "What is your favorite operating system?"

The candidate looked confused and replied, "What's an operating system?"

The candidate was a 2nd-year CS major at a college in Silicon Valley.

Needless to say, we hired another candidate--the 16-year-old kid who built Linux boxes at home for fun. :)

At the risk fanning a flame war between the value of having a degree vs. not, I think one of the big advantages of having a technical degree in the field of something you're applying for is having a lower value on the crackpot index[1] from the get-go.

Whether you're an expert or not is irrelevant. Its that statistically you're probably more likely to be humble about what you don't know, so that way you can go and hunt down the solution if you need to when those problem situations occur.

[1] http://math.ucr.edu/home/baez/crackpot.html

It's not about the piece of paper, but about the theoretical background. You can't even have a conversation with someone about algorithms, performance tuning, optimization etc if they don't know Big-O for example.

Incidentally that's why CS undergrads "waste" their time learning all the sorts, not 'cos you'll ever need to write one from scratch ever, it's just a good way to introduce ideas about time and space complexity.

I have a degree, but it is 10 years old by now. I can't even remember much from it. I don't think you learn that much useful stuff - real learning is learning by doing.