Teaching Kids to Code

by Shardul, 21 Jun 2017

This is probably a topic that has been extensively discussed by people much more qualified to discuss it than I am, but I still felt like trying to argue for my own take on it, hence this post.

What qualifies me to write this anyway? I started learning to code around the end of 6th grade and have been coding since then, learning computer science in a very fragmented way from multiple online sources in the process. In fact, most, if not all, of my programming knowledge comes from Internet tutorials and language documentation. In high school I have not taken (and do not plan to take) a single programming or ‘computer science’ class. And I don’t want to boast or brag here, but for many of my peers, Mandar and I were the go-to resources for programming help, which meant that I got a good idea of what sort of questions or problems they usually had and what thought processes generated them; I had a similar and more consistent experience when I informally coached a few of my friends for the AP Computer Science A exam in April 2016.

In this post I point out a number of things that can be improved with programming and computer science education, but I do wish to make it clear that the problems lie with the curriculum and not necessarily with the teacher. Our school computer teacher was a really good teacher—one of my favorites, even though I had hardly had a class with her, because she used to be involved in many of the outside-class programming activities we did—and it’s not her fault, or any teacher’s fault, that programming education is so bad. Who should be blamed is an entirely different question entangled in politics that I don’t want to talk about.

With that long preamble, I’ll move on to the actual topic.

Firstly, CS/programming isn’t for everyone. It is true that anybody can learn to code, especially with resources like the excellent Khan Academy programming course, Code.org (which I haven’t tried out but have heard is pretty good), Codecademy, and a whole host of easy tutorials for most popular programming languages. But learning to code often means learning a bunch of magic vocabulary and syntax rules that you mash together in different ways until you come up with something that compiles and produces the output you want for a couple of small test cases, if you bother to test it at all. There’s a difference between just programming and programming well. A lot of ‘anybody can code’ initiatives have unreliable ways to check whether true problem-solving abilities are fostered in students, the sort of abilities that let you come up with the concept of binary search trees on your own.

The skills required for successful programming are actually ones that you might have more success picking up in an undergraduate mathematics or computer science course: problem-solving, formal logic, reasoning with and about formal systems, logical reasoning in general, and on a more personal level, persistence and creativity. And while most people accept that undergraduate math courses are not for everyone, programming somehow should still be something all kids can learn. Working with computers means you need to understand how they function, why your incantations cause certain things to happen, and what precise things you need to tell the computer for it to do exactly what you want. Sorting algorithms shouldn’t be something that you read and learn from a book, but that you discover for yourself (at least some of them) from first principles, after you are introduced to computers and their inner workings. (Which brings me dangerously close to discovery-based learning which is a topic for another post. :P)

It is improbable that schoolchildren will learn these skills while learning to code. Most likely, they will have to wait until college to take hard courses in functional programming or operating system design to really wrap their heads around the hard parts of theory. But children with the aptitude for these skills will benefit from an education in programming, upon which they will build their skills later.

Assuming you have some way to pick out programmatically-inclined students in a class—which is certainly non-trivial; looking at their performance in the class would be a bad way to do it because you don’t know what factors are involved—we come to the second point of keeping them interested. Programming can be scary, intimidating, and un-fun if taught wrong. At this stage I feel that the age of the students becomes an important factor.

Teaching young children programming concepts is by far not a modern idea. In 1967, Seymour Papert, a mathematician, computer scientist, and educator working at MIT, designed the LOGO programming environment with other researchers specifically to teach elementary schoolchildren to code. LOGO has come a long way since then, giving descendents like Scratch, which I would recommend over LOGO today; Alice is built along similar lines and with similar purpose.

(The recommendations in the previous paragraph are backed by decades of research. Those in this paragraph are not.) For slightly older children, say, maybe 8th and 9th grade, Scratch and friends can quickly get boring and childish. However, I still feel that direct a direct association between writing code and seeing cool things happen on the screen is important, so my suggestion is to teach HTML/CSS/JS as one package. From an education perspective, none of these is too hard or scary (OK, maybe JS is, but not as much as Java for example), and since they’re designed to work together smoothly in any browser on any platform with no compilation or other programming hassles, they can be fun to tinker with. Results are visually appealing and immediate: “Hey, I just designed my own <X> with cool animations and colorful fonts and dynamic layouts!” And the practical benefits of learning HTML/CSS/JS are obvious because these aren’t some artificially constructed technologies, they run the entire Internet. Two solid years of learning HTML/CSS/JS in school (along with all the cool associated technology we have today, like JS frameworks and elegant NoSQL DBs) could possibly be enough to get you into basic professional web dev. (Not that I’m recommending it, two years could get boring.)

After that, keeping students interested is not an issue. The only thing you shoud not do is teach everyone Java which is, for some reason, what high schools do everywhere; Java is not a good choice for teaching either programming or computer science. (Unless you want to teach Android development later, but even that might now become unnecessary.) Some have harshly criticized Java for reasons that I mostly agree with, and made interesting points about why Java should not be taught in schools, which also I agree with. (If really necessary, a semester or two of Python—non-scary, doesn’t require advanced CS knowledge to fully grok, is useful to build simple things—could be taught as a ‘real’ programming language.)

Which brings us to the third and most important point of teaching CS theory along with programming. If the two articles linked above about why teaching Java in schools is bad haven’t convinced you of it yet, I reiterate: theoretical computer science education today needs more rigor. Both those articles talk about college rather than high school; I argue that if the plan outlined in this post is followed, high schools will receive talented students that already know enough about programming and computers to take on the theory right from 10th or 11th grade. This will no doubt require a carefully tailored curriculum that avoids all the pitfalls and possible gamifications that are too common in education today (again, maybe this can be the topic of another post) and this is well beyond the point where my competency ends, because I myself have not taken a college computer science course yet, not being in college. But I still believe that it is possible and will provide a wholesome, effective CS and programming education to high-schoolers who can do so much more after it.

“High-schoolers who can do so much more”: what an exciting phrase. Computer science, and especially artificial intelligence, is among the biggest fields for modern research. Programming constitutes an essential skills across disciplines, in research and the industry alike. And I’m not saying that what I’ve proposed for computer science and programming education is the best possible solution, but I firmly believe that something has to be done to improve on the current state of things today, and this is just an effort towards that.

Happy Summer Solstice!