How can I teach computer science without using computers?

Asking how you can study computer science without computers is a bit like asking how you can study cosmology without telescopes. Sure, it's nice to be able to look at the things you're studying and it's often very helpful to be able to play around with things. But there's a whole lot you can do without access to a computer: in extremis, you could probably do almost all of a undergrad course with no computers.

In practical terms, access to computers helps reinforce a lot of what you learn in a computer science course. Programming courses are, obviously, much more natural with access to a computer. On the other hand, being forced to write code on paper does encourage people to think about their code and make sure it really works, rather than just running it through a compiler again and again until it compiles and then running trivial test cases again and again until the obvious bugs go away.

Topics that would be most natural without computers would be the more mathematical ones. All the background mathematics, such as combinatorics and probability. Computability, formal languages, logic, complexity theory, algorithm design and analysis, information and coding theory. Anything to do with quantum computation!

Many computer science theories are purely theoretical and can be solved using pen(cil) and paper. Getting information about them and communicating them to the outside world requires a computer.

There is the obvious use of paper and pencil. When I was taught signal processing the special signal processor was not directly available to us and we just handed in paper reports.

Note that the great computer scientist Dijkstra apparently did not use computers very much:

Almost all articles in this series appearing after 1972 are hand-written. Having invented much of the technology of software, Dijkstra eschewed the use of computers in his own work for many decades. Even after he succumbed to his UT colleagues’ encouragement and acquired a Macintosh computer, he used it only for e-mail and for browsing the World Wide Web.

^{Cited from Disjkstra's Memorial}

This shows that you can do great computer science without a computer!

Besides paper and pencil computer science problems can also be solved 'in the real world', e.g., demonstration of sorting algorithms with cards or - more unusual - with dance as done by the Sapientia University.

See http://csunplugged.org/ for some great ideas. Google is one of the sponsors.

From the site:

CS Unplugged is a collection of free learning activities that teach Computer Science through engaging games and puzzles that use cards, string, crayons and lots of running around.

The activities introduce students to underlying concepts such as binary numbers, algorithms and data compression, separated from the distractions and technical details we usually see with computers.

CS Unplugged is suitable for people of all ages, from elementary school to seniors, and from many countries and backgrounds. Unplugged has been used around the world for over twenty years, in classrooms, science centers, homes, and even for holiday events in a park!

You can, it's pretty obvious how you would do it. I'm not sure how useful it'd be to people, but you can.

But I don't even know how a would be teacher of computer science, can ask how you can teach computer science without a computer.

Because if you can teach computer science, then you would know how.

If you know that-

-Algorithms can be written on pen and paper, and traced on pen and paper.
-Flow charts are written on paper.

Studying normalization, the structure of tables of data for databases, can be done on pen and paper. (I only did that on pen and paper when I studied relational databases)

Computer science textbooks don't require a computer.

If you ever picked up a computer science textbook and studied it in a library you'd know you didn't need a computer to understand it.

But you should be aware that anybody that has sat in a computer science class can see that most people cannot program, it's just beyond them. Too abstract for people. That is even true at University level.

http://www.eis.mdx.ac.uk/research/PhDArea/saeed/paper1.pdf

A substantial minority of students fails in every introductory programming course in every UK university. Despite heroic academic effort, the proportion has increased rather than decreased over the years. Despite a great deal of research into teaching methods and student responses, we have no idea of the cause.

I would add that to call it a substantial minority, is an understatement. It's a small minority that -can- program.

I would note though, the cause of the decline that he witnessed.. As computers got more popular e.g. between 2000 and 2010 (he wrote that in the middle of that decade), more wollies were interested in studying 'computer science'. Before then, only the more fanatical people were interested. And earlier on, using computers was more technical, but as computers got easier for the average person to use, and more powerful in the applications they could run, then more wollies were using them and more people that were only interested in what they could do rather than how they do it and how to do it.

It WOULD be possible to teach the concepts of Computer Science without a computer. Of course without access to a computer, what good does learning Computer Science really do?

When I was learning Computer Science in high school, our first several classes didn't involve computers at all. Instead, we worked from a book called Karel++, which taught object oriented concepts. The made up language can be "run" on a whiteboard or blackboard pretty easily because all of the code takes actions on a graph. Our first few assignments were all handwritten and the professor would demonstrate the correct solution on the board at the start of the next class.

When we started discussing algorithms, our professor called us all up to the front of the room and told us to stand in a line. He proceeded to demonstrate a few different sorting algorithms - starting with something silly he called "monkey sort" where he would randomly exchanged two of us and then check to see if we were sorted every iteration, but then performing bubble sort and shell sort - by "running" them on us.

He drew diagrams on the board, explaining the basics of how computer memory and the stack work. He demonstrated variables and pointers (it was a C++ class) and from there to linked lists and trees...

I could probably add to this if I went over the notes I kept from that class, but that's all I remember now. If these suggestions are interesting, I can expand on this answer.

I did a computer science degree and thought we were taught much programming (in various languages) we also were taught many mathematical aspects (and some hardware too).

The maths parts were more important to the first year - think matrix mathematics which is essential to 3d graphics (how do you calculate a surface normal for lighting a vertex?). We did a lot of statistics - think job throughput analysis, what rate of jobs can you pass into a system before it clogs up? That's pretty important for many centralised systems, and networking.

You can teach many algorithms, one poster here mentioned giving students a number each and then getting them to sort themselves using various algorithms (that's active and informational). In fact, any algorithm that can be described in an animated image on the internet can be taught in this way. Similarly the fundamentals of computing - program counter, accumulator and registers can be taught in the same way, or on paper. How about student-based multithreading - set a group of students a card game with a stack between each of them in a classic multi-threading exercise (or the 1 shared spoon problem). Or design patterns, I can think the blackboard architecture is perfect for students to learn how multi-process communication can be implemented, or a n-tier architecture described as a sort of chinese-whispers game or finite-state machines would suit students acting as modules in a 'living computer'. You could teach binary maths, or stack-based (ie reverse polish) execution.

Of course if you have access to electronics you could build an analog computer out of op-amps. They're cool.

Unfortunately all this will seem more like maths than computing, but much of it will stick and one day they'll understand why they were taught it (assuming they go into the computing industry). Remember that anything a computer can do, a person can do - it'll just take him a lot more time and boredom to achieve the same result. You just have to take simple examples to demonstrate what the computer would do, and imagination to describe computing fundamentals in a less dry way that I was taught, way back :)

Asking how you can study computer science without computers is a bit like asking how you can study cosmology without telescopes.

I am mostly answering just to react to this statement and because I strongly disagree with other answers. The correct comparison would in my opinion be

Asking how you can study computer science without computers is a bit like asking how you can study cosmology without a cosmos.

Computer Science is about computers, not just using them. True, a lot of Computer Science is based on mathematics, so the mathematical parts can be studied without computers (thus you're teaching mathematics then) just like instead of cosmology you can study physics which will be useful once there is a cosmos (or have access to telescopes), but in the end what you're studying is still physics.

Now, what everybody is doing in the answers here is note how bachelor level computer science is mostly theoretical and doesn't require computers, the thing those people are IMHO forgetting is that the same people have used computers extensively up to that point. Without a natural understanding directly delving into the theoretical stuff will just cause many students to either give up or have a terribly mismatched view of computer science. So teach some programming, understanding of what computers are first and only afterwards get into the theoretical stuff.

Like if I'm teaching students in Africa that have no knowledge of a computer besides general-level know-how but who aspire to learn advanced computer science and devote most of their time to this hobby?

Depends on budget:

• Nearly no budget If you've the money to travel to Africa grab a bunch of arduino clones (there are functional arduino clones for $13, check this list for example) , some output components (such as LEDs) and one laptop. Start with some basic programming, teach it to them on paper and have to input it on the laptop. It's not the most efficient way, but it will teach them at least a basic basic understanding of what computers practically are.
• Low budget Buy \$750 dollars worth of \$50 android smartphones and keyboards (microusb keyboards or bluetooth keyboards, check ebay.com or aliexpress.com for both the smartphones and keyboards). Yes they are crappy in a million ways, don't even tell me about it. Still though, you can input data on them (though don't even consider using the onscreen keyboards) and there are simple coding apps for android. Have multiple students per smartphone and that can work quite well.

So let me be clear, I am not saying Computer Science can't be taught without computers, just that you need to be used to computers and know them fairly well before you should start considering such a thing.

Are you looking to teach programming?

When I was a kid, we went on a camping holiday and I took the computer manual with me. I was able to write a simple game using pen and paper.

This was my first real program I guess, so a lot of learning was involved. I knew what I wanted to do, but there were a lot of false starts while I figured it out. But eventually I was able to write out the whole code.

I guess what I'm saying is start with pseudo code and\or flow charts to plan the project and then translate into code. I was able to debug a lot of my code just by reading it. You could also use peer reviews here.

To echo to the previous answers: there is a lot to be learned just from reading and doing pen and paper exercises. If you can work the problem on paper you have the hard part down pat. However, I also wish to mention the existence of fairly cheap ways to expose students to CS. Are you familiar with the Raspberry Pi Project? Even if several people has to share or you can only afford to get 1 or 2 units it might still be worth considering as a way to expose your students to the technology.