I'm someone who has been educated in CS, as well as other fields. I both perform and manage industrial-scale programming and research programming, and I hire programmers of varying skill-levels for projects.

In my experience, there are a few caveats to your advice:

1) Strong programmers, fun side projects, and deep understanding comes from the pre-Computer interest which many of the die-hard CS people have. It is rarely, if ever, taught in academic programs and is something I specifically look for in my heavy-duty programmers whether or not their major is or was in CS.

2) My personal take on things is: The best programmers take innate talent and a lifetime of experience and turn it into super-star ability - don't expect to become one by majoring in CS. The second best programmers have a strong mathematics background (i.e. proof writing), because their code requires less debugging, which is the most costly part of software development. Any programmer can do better by analyzing their process and learning from that, but that form of software engineering seems to have become less popular, since it takes more work and makes you vulnerable to bad managers. Algorithm design is, of course, an entirely different matter.

In short, I'd suggest a CS/Mathematics double major. It's a good idea to be able to speak more than one technical language after all.

Oh and, my usual answer to:

When's the last time you heard a math person refer to some real-world situation as "a real elliptic curve"?

Well there was the time Heisenberg came back from his honeymoon where he had, of course, spent the entire time trying to puzzle out quantum mechanics. He was about to give up because of the non-commutative multiplication when he talked to a mathematician friend who said, "Wait, let me tell you about these things we call matrices."

Mathematics gives you very useful models, it just doesn't tell you how to use them. CS gives you the tools to implement solutions, but it tends to leave the solutions rather ad hoc. Bringing the two together is one good mix. I could suggest a half-dozen others - but this will do for now.

Edited to add: This is a mix if you want to achieve the goals laid out in the post. Not suggesting that it is either trivial to get such a pair of degrees or desirable for everyone to head in that direction.

Whoa... you are suggesting that everyone else do what you didn't and the reasoning is a list of ideas with no data? No offense, but how do I know this is not a grass-is-greener, I-wish-I-would-have moment?

I would beware of other-optimizing. As soon as I graduated with a CS degree I realized I should have been in philosophy the whole time. I love my CS major and agree that it is an excellent choice and do not regret the work I put into the process. I also agree with your points. But choosing a major is not a small thing that should be tilted by a post based on anecdotal evidence from someone who didn't do it.

After watching math and physics majors struggle in CS101 I realized that CS is not an intuitive field for everyone. They were not stupid or lazy people. They just didn't "get it". You can graduate with a CS degree and never "get it" and not know jack shit about coding, systems, or any of the other great things you mentioned. I learned about those things. But right now I am working with someone who graduated with a CS degree and did not.

This isn't meant to be a rain-on-your-parade comment; I just want to warn the bloke who reads this post and thinks, "God, math is that useless? I guess I will go CS." The key part in your post they may have skimmed over is this:

You shouldn't be considering a career in academia unless you're passionately in love with your field, unless you think about it in the shower and over lunch and as you drift off to sleep, unless the problem sets are a weekly joy. A lesser love will abandon you and leave you stranded and heartbroken, four years into grad school.

If you are passionately in love with your field, go for it. If you are not, find a different field. CS is not something everyone can fall in love with and think about in the shower and over lunch and drifting off to sleep. Neither is Math, Physics, Engineering, Psychology...

I also want to ask: How can one lead an intellectually active life besides reading books? Is it common for non-PhDs to publish in some fields?

I strongly disagree.

Statistically speaking, salt mines await most CS majors.

I regret my CS degree every day of the week.

Unless your grades are exceptional (or you have the business acumen and extroversion to deal with startups) a CS education dooms you to the career of a human robot.

I will also note that it is possible to be passionately in love with your field without also being passionately in love with homework or exams. For people like this, some of whom want to pursue a life dedicated to chasing knowledge possibly more than most of the academically excellent, universities offer only a savage self-esteem beating.

Seconded. Random bits I couldn't manage to form into a cohesive comment:

I'm a math ABD on the verge of dropping out (to pursue a software project, no less) . I actually am doing really good research and probably have a promising academic career ahead, it's just not exciting.

I'm always dumbfounded by how much more interesting CS grad students seem (not necessarily their work though, the kings there would have to be the physical sciences).

Math has definitely restructured my brain in great ways. I program a lot too, and programming has done likewise. I think the best thing I've gained from math, though, is the perseverance necessary to do good research, which comes in handy elsewhere for sure.

My recommendation: if you major in CS, take some real math courses on the side.

Thanks for writing this article. It would benefit the software development community to know more about this difference. I have a little more in depth description of the digital divide between programmers and computer scientists on my blog.

As an overly stereotyped generality, here are the differences I see between "computer scientists" and "programmers".

The programmer is someone writes code on a daily basis but is ignorant about how the compiler actually works. He is very familiar with the tools he uses but doesn't understand how they work or how to build them. The tools he uses is his world and he uses them to go about his daily work. He is also blinded by them and can't see past them. "If all you have is a hammer, everything looks like a nail". They never think there is a better way and just go about working on their problem the only way they know how.

In contrast, the computer scientist is someone who lives in the more academic world and understands the computer at a more fundamental level -- a machine for performing computations. He sees problems and is able to build a model to solve the problem.

In summary, the programmer uses the tools and the computer scientist makes the tools. The problem is that often these two individuals do not spend enough time in the day to day world of the other. So the programmer is ignorant that he can do things better (more efficient, elegant, etc) and the computer scientist doesn't spend enough time in the trenches to really know what would benefit the programmer or simply doesn't have an interest in solving the problem outside of intellectual curiosity.

Lisp is a great example of this. I recently learned it and it is an amazingly powerful language. Yet I find myself not able to use it because while it is interesting in and of itself, the people who worked on Lisp have done little to make it practical to use on a day to day basis. The learning curve is steep, documentation is poor (relatively) and trying to figure out how to get it up and running on a server in an easy and reliable manner is a P.I.T.A. I eventually gave up on it and switched back to PHP even though every part of me wanted to use it.

Look at PHP. Even though PHP pales in comparison with Lisp. PHP's superior documentation and ease of deployment have made it one of the most popular web development languages.

If the two communities can get together and live in each others' world more, maybe we can all benefit more.

The problem I have is that I am a self taught programmer with a keen interest in computer science but find that living in "computer science land" takes away from my ability to get real work done and pay the bills. I wish it wasn't the case. To my credit though, my PHP code has become a lot more elegant as I am able to take the CS concepts I have learned and apply them to my day to day work.

CS majors learn to really program

I can't really point to anything I learned while getting my CS degree that was particularly instrumental to being able to "really program", and for that matter I don't think I could "really program" until some time after I had graduated. Practice counts for a lot here, especially considering how ungrounded and unreliable the current pedagogy of "software engineering" is. Practice and perfectionism did much more for me as a programmer than any course I took in college.

I grew intellectually through that study, but I've upped my practical effectiveness enormously in the last few years by working with great CS people and absorbing all I can of their mindset.

From my vantage point as a CS grad who maybe wishes he majored in math, physics, or chemistry instead, it looks to me like you're getting the best of both worlds. CS programs tend to be very focused on producing capable programmers, but programming is largely an operational set of knowledge that can't yet be reliably taught in a classroom. My wife is currently pursuing a chemistry degree, and I'm downright jealous of the information density in her courses compared to mine. CS pedagogy is simply immature compared to older fields, full of conflicting opinion, heuristics, and transient industry buzzwords. An exercise: Ask three randomly chosen CS majors what object-orientation is and why it's so great, and compare their answers.

I'm waffling on the regret because the degree did lead directly to an enjoyable, well-paying job. But most of the time I spent in CS classes feels wasted in comparison to the maturity and density of math and science classes.

Simply put, most people who I know who major in CS, or especially who pursue a PhD in it, come to regret it.
As a rationalist, the plural of non-selective anecdote is data... OTOH, non-selective data within me may be selective data to you, since if I didn't object to this post so strongly I might not have commented.
I'll try to correct by commenting with simple enthusiastic agreement some time soon.

Just to add my own two cents. The market for CS majors is excellent. I guarantee you; you won't have any trouble finding a job if you are a CS major. Nowadays, there isn't a company that doesn't have software people.