CS Education, where was it? Where is it now? Where is it going?

Question

I'm planning on doing research into Computer Science education for my senior project and I could think of no better group of people to consult than the SO community.

So, what do you think about the state of CS education in the United States (and the rest of the world, though I had planned on focusing on the US)? What do you feel are necessary topics? What topics are superfluous? Objects first vs. Objects early vs. Objects later?

ALSO, if anyone has or knows of any blogs, studies, etc., those would be very helpful as well.

Answer 1

What do you mean by CS Education? Do you mean software engineering education or actual computer science (or rather pseduo-science)?

Answer 2

You might want to contact Andrew Petersen. He is the Faculty Advisor for Undergraduate Computer Science at the University of Toronto at Mississauga. He is particularly interested in what your question asks.

Now for my opinion:

1. Focus on how concepts are being taught. This is because there is a huge gap in what teachers knew when they first learned what they are now teaching their students and what these students now know. Also keep in mind that these students are learning things from their teachers at a different time in their life than their teacher learned these things.

2. What kind of questions do students ask? This reflects the gap in the students actual understanding of anything when they are being taught and what the teachers think is the students understanding.

3. What kind of assignments are they asked to do?

4. What can they actually do as first year students? as second year students? ...? as graduates? In my summer after my first year of CS education, I was at an internship that wanted me to do some organic SEO work for them. At the end of this project I was told that what I had just done for them would have cost hundreds of thousands of dollars a decade or so ago.

5. Talk to employers. Why do they hire students from this program over other programs? Is it strictly domain skillset or does it also have to do with general learning skills?

6. Is there a social aspect to this? How are geeks and nerds being look upon today? What is their social status while in school? Do they play sports now? Do they have friends? What are the gender ratios?

This is all I can think of right now, but I will post more as I think of more stuff

Hope this helps

Answer 3

I am an undergrad student, and the following is my opinion.

One major aspect I would change in my courses, is to make them depend on previous courses. I mean it literally. When you throw the code you wrote, you are just throwing everything you learned in that course. What if we had the chance to contribute to a class-wide project where everyone is assigned a task, instead of that everyone is assigned the same project or a similar project. Of course, in the next suitable course, you would improve the aspects you learned in this class. While that, I can be exposed to maintainability and other real-worlds skills that I feel lacking.

My point is that the curriculum doesn't lack theoretical topics, instead it lacks the exposure to real-world application of that theory.

Thanks,

Answer 4

We have trouble separating the Science from the Engineering.

Many schools turn out "Computer Science" degrees with skills that would more properly be taught in an engineering program. Many employers mistakenly require CS on a resume, when in fact they need an engineer.

This has led to an overabundance of programs calling themselves CS when they are, in fact, nothing of the sort. It has also led to an overabundance of Computer Science degrees.

Real CS programs are becoming more rare. Students enrolling in them are valuable, but often under-utilized. If you need someone to code Java in a cubicle for you all day, you probably don't want a Computer Scientist. If you need someone to design code that takes weeks to run on your supercomputer, you may.

One metric to tell the two apart is the languages they're taught in. If your course of study is all in Java with a focus on practical applications, it's almost certain that you're learning the engineering version. If you start with a lisp-variant, and work in a new language every semester, you're far more likely to be training as a scientist. If you design operating systems and are required to take a great many math classes, you're probably becoming a scientist. If you're taking CS because it will provide a dependable, well-paying job down the road, you're becoming a software engineer. If you're taking CS because you want a PhD, you're probably becoming a scientist.

Employers need to recognize the difference when hiring. Some already do. Many end up hiring non-CS degrees such as Cognitive Science for creative, out-of-the-box thinking. It is unfortunate that the mark of a creative, scientific approach to problem solving has become an avoidance of the named degree.

Edit: I'm not terribly surprised that SO is full of people with CS degrees extolling the real world applications of such.

My point here is not to denigrate Computer Science as a degree, but to point out that the definition has widened to the point that "Computer Science" no longer tells us what we want to know. I want software engineering degrees to be labeled as such, and to focus on valuable skills for that discipline.

In other fields, calling something "Science" means that the discipline and resulting job opportunities have a strongly research-oriented focus. You hire Structural Engineers and Materials scientists for different jobs. I'd like the same to be true of Computer Science.

Answer 5

This is a great question. for background I'm 10 years out of college where I got both and MS and CS in "Computer Science". Pauls post from above is quite good - he's correct in stating that we have trouble drawing a line between the science and the engineering. Most of this is because computers are a relatively new thing where every ten years many of the technologies have been reinvented.

In many schools you can choose between CS course loads – often there will be a CS course that came from the math department and one that came from the EE department.

As for which dept your course load comes from, the stinker is that both are important and you can’t do one well without the other. My course load focused on algorithms, math, and the science. This background was GREAT if you're looking to stay in a research environment. Trouble is to do REALLY COOL STUFF you can bet the size of the algorithms will grow as you pile on code.

Once you start piling on lines of code the job changes to how do you architect your code so you can refactor, track problems, layer, decouple, and grow the code base …so instead of using a calculator you’re using project management / engineering techniques.

The good news is that there is need for both in our economy – you’re lucky. Even better is they really go hand in hand, you don’t get a great piece of engineering without good science, and good science requires the engineering discipline to scale.

Here’s what I’d do – get a project, something math heavy, something practical, whatever. Then get a free copy of Visual C# Express and a copy of the book “Introduction to Algorithms” by Cormen/Lieserson/friends. This is a very famous text used by almost every CS department in the US. Here is a URL: http://www.microsoft.com/express/Windows/. I’m pushing C# because it’s a clean language and VS is free and integrates debuggers/compilers/editors. You could also use C++/Java/C.

Good luck!

Answer 6

Wow. Where to begin?

• Treat it more like engineering.
  Pick any other branch of engineering - Electrical, Chemical, Mechanical. It is considered "applied science". That means there are underlying physical laws discovered through science and math, those laws are quantitative, and the engineering consists of applying them to solve real-world problems. To my mind, the basic underlying laws of computation has to do with computability theory and information theory, in their various forms.

What we have in "computer science" is some of that, not much, accompanied by a lot of band-wagons. Certain things are supported by proofs of their benefit, like algorithmic complexity, and Huffman coding. Other things are not, like OOP and functional programming. And it seems like the less concrete justification there is for a concept, the more dogmatically it is taught. It is as if medical students were taught to believe (under pain of a failing grade) that monoclonal antibodies were the answer to every illness, and when they graduate they carry that religion into the outside world. NO - they are taught that every situation has multiple interpretations, multiple possible solutions, and to weigh the pros and cons of all the different courses of action.

To me, one clear indicator of this problem is the emphasis placed on inventiveness. To find a new and interesting way to do something, even if it has flaws, is highly prized. In CS it is discouraged in favor of doing things "the right way".

• Have quality-control.
  A curriculum consists of courses that build on each other, so if the professor of CS202 feels that the professor of CS101 is not adequately preparing the students, they can hammer it out. In an engineering school, there is interaction between industry and academia, so that if industrialists think the graduates should know something, they can bring it up with the professors.

In "computer science" the attitude of professors is that it is their job to teach what they consider most interesting, and that it is the "real world's" job to teach them useful mundane techniques. So if you're hiring graduates, you see people who can tell you what a friend class delegate or a virtual destructor is but don't know how to develop design alternatives. On top of that, the graduates are not of a mind to learn practical skills from industry. Rather, they are of a mind to spread the gospel of the latest nifty bandwagon taught to them as the end-all of software to all the ignorant old-timers in industry.

So, long and short - treat it like real engineering, and have quality control via two-way interplay with industry.

Answer 7

Well, you probably won't like my answer.

The enemy is accreditation.

Going back to early computer science education days, the best schools quickly had good computer science programs: MIT, UC Berkeley, CalTech, etc. These schools had a lot of experience with new fields and many multi-talented professionals to slot in quickly. And then there were some outliers, like California State University at Chico (CSUC).

Now a state university in California is not a top tier school: those are the Universities of California. This small, middle of agricultural area school quickly became an excellent school for computer science. The pressures were immense on students. Excellent faculty were lured by rural settings and reasonable weather. The goal was to take a student, teach them for four years, and turn out a graduate equivalent to a professional with five years experience. Hewlett-Packard would routinely hire large fractions of the class.

Then, in 1986 or so, the department decided to become an accredited computer science school. Several courses were split into two semesters instead of one. Calculus was added as a requirement, displacing one of the several required writing classes. Students were allowed to write the code from home instead of being physically in the cavernous labs. The requirement that students would code on a different computer and different language each semester were dropped.

The pressure dropped. The unreasonable drives for six 3,000 line programs per class dropped. The frantic collaboration in the crowded labs as people debugged each others code went away. The quality dropped.

Today, the Computer Science department at Chico is all but dead. There are less than a dozen professors left. Companies no longer clamor for its graduates. Oddly, there are four other computer degrees at Chico in more demand (Business Information Systems, Management Information Systems, Computer Engineering [firmware], and Computer Art and Animation.

There is no need to be mediocre.

Answer 8

With all due respect to CS schools, I would recommend against the academy, while so much exciting research is going on in the industry. Computer vision, for example, looks unthinkably promising - and there're companies doing wonders in the field.

Also, the industry is a much more "honest" place. Making something work is a more convincing achievement than academic papers.