Slashdot Mirror
Why Johnny Can't Code
GoCanes writes "Salon has an article named 'Why Johnny Can't Code,' an interesting examination of the dearth of line programming languages available today. At first I wanted to read this and say aha, here's a simple line oriented language that's available through open source, but after reading the article I couldn't find any. And being an old fart, I remember the days spent with edlin and basic."
I have been thinking it is due to a few reasons. First off, it seems that math education is sorely lacking in many college students. We are waaaay off the mark here with lots of remedial work being necessary for entering college students and this is having a major effect on peoples ability to develop algorithms. The next major reason is lack of communication skills and writing skills. When I was a grade school kid learning BASIC, we had it drilled into our heads that we had to comment our code and explain exactly what it is that we were planning. This was done to help us learn how to think through a problem, but also to get us to help communicate what it is that we were trying to do with our code. Finally, it seems that many students have gotten lazy and are simply looking for an easy way out and not wanting to code line by line.
Visit Jonesblog and say hello.
Please provide a URL that does not require signing up to crap.
I submitted the same story, and also made the following comment elsewhere:
It is worth noting that Microsoft ships Visual Basic for Applications
with most of their software (at least they did when last I checked), and
OS X and many Linux distros ship Python.
Having said that, I agree that the world has changed and getting started
with programming isn't as natural as it used to be. Part of the reason
for that is that our expectations have changed, but languages haven't
really caught up. Nowadays, we expect programs to present us with fancy
GUI widgets. If all a program does is print some text to the terminal
and read some input from it, we don't feel we've written a real program.
However, in the majority of programming language, creating these fancy
GUI widgets is much more difficult than doing terminal I/O. Thus, either
the rewards are less, or the barrier to entry is higher.
On the other hand, there is HTML, which makes it very easy to create
user interfaces, but the user interfaces are somewhat limited and adding
in program logic can be very tedious.
Fortunately, software vendors are not sitting still. Konfabulator, Apple
Dashboard, gDesklets, Oprera Widgets, KDE's Dashboard, and Mozilla's XUL
(which is supported natively in Mozilla browsers, in Opera and MSIE
through plugins, while native support is being added to Konqueror and
Safari) are all ways to make it easy to create visually appealing
programs and add functionality to them. They are all based around the
concept of creating the user interface in XML, then using a scripting
language to implement the behavior. Indeed, this makes creating GUI
software easy enough that beginning computer users can do it.
I think these widgets are the future, especially once they will be
treated like full applications that live on the same desktop (last time
I checked, Apple Dashboard actually was a separate desktop from the one
that contains the "real" programs).
Please correct me if I got my facts wrong.
What a load of rubbish.
apt-get install \
bash \
python \
gambas2 \
kturtle \
fp-compiler fp-units-base \
php5-cli
The reason children don't code (if that is even true, as it's a completely unsubstantiated assertion) is because they don't want to.
I started programming when I was ten, and I did it by hand-converting Z80 assembly language to machine code and then used BASIC poke commands to write them into memory. I had to work hard to scrape a C compiler from somewhere and that was heaven.
Today it is a million times easier to write a program if you wanted to. Blame ease-of-use culture; blame video games; blame stupid parents; but blaming the lack of access to programming languages is ridiculous to say the least.
Carpe Daemon
So, just in case some of our students read this (its always weird when your students say "I saw your comment on Slashdot yesterday"), I should have mentioned that there are exceptions to the rule (obviously) like the developers who are developing some of our image processing code right now here at the Univ. of Utah Scientific Computing Institute and our collaborators at the UCSB Center for Bio-Image Informatics. These folks are doing amazing things as is the neuroscience grad student in our lab who routinely amazes me with his ability to code. But these folks are in these graduate programs and environments because they are interested in solving innovative and new problems and they enjoy a challenge. The vast majority of students who are learning to code are not interested in asking questions or doing hard things. Interestingly, a large number of students it seems coming into CS programs at the undergraduate level want to program games, but have no idea how much math and algorithm development goes on when one is crafting new material. It's easy to use someone else's engine or physics models, but coming up with your own is harder and requires some talent and dedication to learning your craft.
Visit Jonesblog and say hello.
If you really want to play with BASIC, you still can. There is Freebasic, at http://www.freebasic.net/, a GPL'ed open source BASIC compiler.
When I was in college, we started off learning different assemblers (PC and Mainframe), then we moved to C and C++. The same college, now starts teaching kids some RAD languages, Java, C# and whatnot. IMO, new students no longer learn the basis of the computer, how the CPU works, how compilers work and how interpreted code works.
One of my most memorable classes was the C++ class when the teacher started off by teaching how compilers actually compile your code. Directly from this you could easy see what code to write and what not to write. Those classes have been replaced by Java, where you don't even worry about resource leaks. Hey, garbage collector takes care of it!
To make good programmers, one has to understand the basis of it. Compilers, C and Assembler. If all they learn today, well mostly OOP stuff, they won't learn the basis well. And they will end up writing bloated code left and right.
Am I missing something here? Why doesn't python count? It's easy, has a nice "line enter mode", embraces many advance programming concepts from various programming paradigms(OP, Functional, Procedural, etc). And has a bunch of nice Graphical hooks for the eventual game programming your little one will end up doing.
Disregarding Python, what's wrong with Emac's elisp or a nice session with tcsh. You'd be hard pressed to find a computer that you couldn't run one of these languages on and I've just barely scratched the surface of possibilities
No I'd have to say that today's children are given an even richer programing environment to grow with than we were.
--Steve
Why line-oriented?
Because Djkstra was right. With whatever apologies might be due to the author, his early exposure to BASIC has damaged his mind so beyond repair that he cannot concieve of a good learning language as being anything other than line oriented.
Python and Squeek, each in their own way, are probably the best learning languages ever devised. They are both "real" languages that can be used for "real" work after you learn them, but also serve as good stepping stones to other languages, without instilling you with a line oriented mind.
Programming students don't need to learn line numbers, they need to learn number; mathematics; and logical structure. Line numbers do nothing other than give the illusion that a program has structure, whether it actually has any or not.
KFG
Devil's advocate time: Remember Computer science isnt about the plumbing; its about abstraction, algorithms and problem solving ability. Modern platforms allow modeling and thinking at a much higher level - instead of focussing on low-level arcana.
For example, lets take the author's wishlist - creating Pong. I'd wager that the game can be written in a lot more expressive manner in any higher-level language, compared to basic. Poking different screen locations arent a lot of fun; gorilla.bas gets old very soon. Kids like instant gratification(adults too!)- higher level languages are ideal for this.
From TFA:
Yes.
The writer of TFA is whining that computers no longer ship with a BASIC interpreter. That's been true since Microsoft shipped Windows 95--and at the time (which, er, was 1995) a number of columnists (including me) noted the loss, and wondered what impact it would have.
Eleven years later....
The earth still continues to revolve around the sun, and kids are still learning to write code. They're not learning to write code by typing in exercises from a math book--they're learning to write code to develop macros in Excel (or OpenOffice Calc); they're learning to write JavaScript to enable custom functionality in Adobe Acrobat files; they're learning to write HTML.
Not to get all pedagogical on you, or anything...
But this may be a circumstance where the Education Establishment is doing something right. Typing code into GW-BASIC produces...what? Typically, an output value. Read these input values, produce that output value. That's a CRT-based emulation of a deck of punch cards--and while I believe I benefited enormously from learning to code on punch cards, very very few people in the programming world agree with me. Think about what the newbie programmer's early coding experiences are like:
That experience--being able to compile and run a program in something like real time, was HUGE--in 1974. Nowadays it is so outdated that I'd bet most kids would not see any correlation between that and a computer program that they are familiar with. How do you explain to a ten-year-old that the BASIC exercise from the math book is essentially identical to the internal processes of a video game, a web server, or the embedded micro-processor that drives your microwave oven?
By contrast, a web page is a terrific introduction. Open a simple web page with "View Source", modify the text, and display the new page--it has changed. Iterate several times--add paragraphs, change colors, play with fonts--and the kid gets it. Playing with a text editor (which, incidentally, still does come for free) the student can go a long, long way in HTML. Depending upon his or her interests the student can pursue graphic design, animation, AJAX programming--all sorts of stuff. Key point: the difference between a "Hello World" HTML page and EBay is only a matter of degree--and that is immediately obvious to the student. Making the mental link between an ancient BASIC program and--for instance--a Windows application developed in Visual Basic.Net is not obvious at all.
A case in point...
I have three daughters. Growing up in a house with more computers than people--and a T1 connection to the Internet--they have had more exposure to computers than most kids. The oldest two are in college, and both had summer jobs this year at the electronics company where I work. Daughter #1 worked in customer service role--spending a lot of time with Excel spreadsheets. Somebody asked if she could program--she replied that her dad was a programmer, but she didn't really know how to. As she said this, she was writing Visual Basic macro scripts to automate a lot of manual key-entry for the customer service staff. What she was doing was very similar to the old-fashioned "get input, produce output" coding that punch card decks and BASIC programs performed. But the IDE (in this case, Microsoft Visual Basic for Applications) and the paradigm (a GUI) made it extremely easy to understand and do. So easy, in fact, that she didn't really equate it with programming.
In other words,
Don't mourn for GW-BASIC. Spend time with your kids writing HTML, JavaScript, and Excel macros.
You should know that the University of Utah has a long history of academic excellence in computer science. People like John Warnock (founder of Adobe), David Evans and Ivan Sutherland (founders of Evans and Sutherland), Robert Barton (principal architect of all Burroughs computers), Tom Stockham (created digital recording among many other things), Alan Kay (GUI guy at Xerox PARC, developed Smalltalk and is a fellow at Apple), Chuck Seitz (pioneer in asynchronous circuits), Ronald Resch (pioneer in computer art), Alan Ashton (founder of WordPerfect), Tony Hearn (developed the oldest algebraic math package, REDUCE), Duane Call (designer of the FPS-120 supercomputer and vector calculation specialist), Henri Gouraud (developed the Gouraud shading methods so important for all your games), Elliott Organick (Founder of SIGSCE and author of many FORTRAN and CS textbooks), Buit Tuong-Phong (invented the Phong shading method), Ed Catmull (computer animation pioneer and co founder of Pixar), Jim Clark (founder of Silicon Graphics, Netscape and Heatheon/WebMD), Henry Fuchs (founder of Pixel Planes and researchers in high performance graphics software), Martin Newell (object rendering and founder of Ashlar and co developed the Painter's algorithm for surface rendering), Frank Crow (famous for his anti-aliasing methods for edge smooting), Martin Griss (developed Portable Standard LISP), Suhas Patil (founder of CIRRUS Logic), James Blinn (Invented the first method for representing surface textures in graphical images. Scientist at JPL, where he worked on computer animation of the Voyager fly-bys), Jim Kajiya (Developed the frame buffer concept for storing and displaying single-raster images), Robert Johnson (Invented the magnetic ink printing technology used on virtually every check we write), Brian Barsky (Developed beta splines and methods to link computer graphics, geometric modeling, vision science, and optometry) and so on and so one are all graduates of our program.
Visit Jonesblog and say hello.
Comment removed based on user account deletion
first off, it seems that math education is sorely lacking in many college students. ... The next major reason is lack of communication skills and writing skills.
There's a story about Abraham Lincoln when he was a cogressman. He received a questionnaire sent to new congressman in which they were supposed to give background information about themselves. When he came to the question which asked for a description of his education he filled in a single word: "deficient".
In fact, Lincoln felt the deficiency of his education so keenly he began an independent study of Euclid's Elements, with the intent of sharpening his thinking. Reportedly he mastered the material through book 6, which means everything relating to plane geometry, but not number theory or solid geometry. I'd argue that Lincoln's formidable intellectual accumen, whether it stemmed from his study of Euclid or from his native faculties, is often missed in the sentimental haze that surrounds the man.
The point of these anecdotes is this. There are certain areas of knowledge where thoughtful persons would have to label their education "deficient", any accomplishments they may have in them notwithstanding, because they are bottomless sources of utility. You just named two of them: mathematics and communication.
The need for remedial classes at the university level doesn't come from a degeneration of the educational system. It comes from a social and economic change which makes a bachelor's degree a minimal requirement for decent employment. Fifty years ago, there were two kinds of people who went to college. The socially elite, for whom a "Gentleman D" was perfectly acceptable. Arguably the societies they joined and contacts they made were the most important reasons to go. And the intellectually elite. The middle class, non intellecutally elite student got a job, often in a factory, which maintained him in the middle class and did not require a degree.
The growing need to have a university degree in the last half of the twentieth century led to the perennial concern of declining college board admission test scores, from which it was concluded that primary and secondary education was deteriorating. However, it's important to remmeber that average scores are average scores of a population sample, and if the population being sampled changes, you can't compare the scores. Adjusted for demographic changes, there was no decline in scores. There literally couldn't be, because of the little known fact that the tests and scoring scales are continually recalibrated by the testing agencies to ensure that students in any given segment of the academic population score the same from year to year.
However, we have been the beneficiary, first of the Sputnik scare, then the SAT scare. For most students, education is actually much more rigorous today than it was fifty years ago. I went to elementary school almost forty years ago. My school was unusual in putting a great deal more emphasis in critical thinking and reading skills. My children go to an elementary school where the reading curriculum is much more challenging, and the math curriculum requires that students be able to reason mathematically and recognize situations where various mathematical techniques need to be used -- a huge improvement over the education of my era, which graduated countless students who could regugitate the quadratic equation forumula but had no idea when it might be useful.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
I totally and completely disagree. The issue is one of pedagogy--which language can kids learn at an early age? I was coding 1000 line programs in BASIC at age 8, and I wasn't alone in that. BASIC is an easy language to learn. Is it limited? Definitely. Would I prefer today to code in BASIC instead of the Perl, Matlab, and Fortran that I now use? No. But the author's point is about _learning_ to code. Perl isn't terribly easy to 'learn' nor are c or c++ or java, at least not to an 8yo.
I wonder if this CS is a branch of mathematics approach also puts people off. Personally I think there is much less mathematics going on at the programming level than college lecturers like to think. It is like the old saw that a dog that catches a ball is solving a 2nd order differential equation in real time. No its not, it is catching a ball. Similarly a program is more a logical story than some mathematical adventure.
It may gall college profs who are still trying to foist formal methods on people but setting a high mathematical barrier to entry on CS courses and having a high maths content is a bad thing.
As an example I have a first in Microelectronics from a British University (a course which had a large syllabus covering 'C' and machine language) but only just scraped the 'C' grade needed in mathematics as an entrance to this course despite having 'A's in Electronics and Computing. I doubt I have used any maths much beyond British 'O' level standard since and certainly the maths knowledge required as entry to a degree level CS course is too wide and deep. It is all geared up to people going onto research rather than the real world. You don't need to be an expert in set theory to write the level of SQL required by most applications.
This is just a natural evolution. The same situation exists in hardware design. Very few young people can whip up a useful circuit with resistors, capacitors, and transistors. At my company we have 1 or 2 "dinosaurs" who invariably get called upon to solve problems in the hardware when it gets down to this level. Everyone is amazed when they throw together a quick little circuit with a breadboard and a few hand-soldered components. The simple fact is, more people don't learn to do this low-level type of work because there is not as much need for it anymore. But, and here is the rub, there will always be someneed. We'll always have to have a few folks around who understand the fundamentals. One could imagine as things continue to evolve, that these "low level specialists" could become quite expensive.
The more you regulate a company, the worse its products become.
Unfortunately, without understanding the basics, all your higher level abstractions get you is unmaintainable abstractions. Understanding the base concepts is necessary to know when to abstract at higher levels, because then you'll understand what that higher level abstraction does for you.
Does this mean you need to know about peeks and pokes? In today's world, I no longer think so (and there's that huge collective sigh of relief from students everywhere). But you should definitely understand how your language handles memory, even if you're coding in a language with GC capabilities. It keeps you from churning memory and slowing down your app to a crawl.
The cesspool just got a check and balance.
Before they did anything else they were writing programs. Simple ones that started the game. Next they were playing with loops:
And er,
Do kids today that are interested in soccer not think it's fun because they can't bend it like Beckham the first time they kick the ball? Do people who like woodworking stop because they can't build a perfectly constructed table and chairs, and instead can only build a crappy bird house? I don't think that's really the problem here. Sure there's going to be some kids who are discouraged because they can't program Doom 3 the first time they sit down and code. But those probably aren't the kids who would end up having a career in computers anyway.
Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
What does learning about the underlying system help with in programming?
.NET stable are more mainstream than C/C++/whatever.
.NET runtime and turned into CISC instructions which are sometimes even then converted into RISC instructions... what exactly should I be coding to? How can I know what's happening to my code after it's run through 4 or even 5 levels of automatic conversion?
Architecting beautiful code? Nope.
Learning powerful, high-level abstractions? Nope.
Programming efficiently to the hardware.
With computers getting ever-faster and more advanced, coding efficiently is less and less important than it used to be. Efficient algorithms are still pretty important, but that's more about using matrix calculations and streamlining progam flow than about the intricacies of memory-management and choosing where on the hard disk you place your data for minimum retrieval time.
Witness the rise of interpreted languages like Perl/Ruby/PHP. Look at how semi-compiled bytecode languages like Java and the
Don't get me wrong - I first learned BASIC, then C, then C++ and a bit of assembler, and I wouldn't trade this knowledge for the world. Nevertheless, now when I write C# code which is compiled to MSIL/.NET bytecode, which is then interpreted by the
These days kids are learning how to use high-level tools first, and only learning the lower-level stuff as they get better and better. Sure this means an awful lot of nasty, bloated beginner-level PHP and VB.NET, but by the time they're ready to tackle compiler design they've generally already picked up the important bits as they go.
When we were learning code we'd write crappy code in BASIC/C, and it'd sit on our home PCs and never go anywhere. Now when a kid writes crappy PHP or VB.NET it's generally posted to a forum or used to run their website, so of course it's more visible.
We all wrote crappy code when we were learning, irrespective of the language. Certainly, I know I wrote some BASIC/C code that makes me cringe when I think of it now.
Everything in moderation, including moderation itself
You managed to get David Brin's point terribly, horribly wrong. It's not that BASIC provided a good basis for future programming, but that BASIC awoke a desire to learn more programming. It was good for non-programers to grasp the concepts behind programming.
I think the best candidate for a BASIC replacement was HyperCard, but alas, that too is a dead project. It was cool, and its scripting language - HyperTalk - was revolutionary in that it was geared to resemble "natural English". Sure, it was limited in many ways, but the original Myst was made with HyperCard, and those of us that fiddled with it were ready when the Web hit prime time, already armed with the concepts of event-oriented design. HyperTalk scripts were easy to read, so easy that commenting became scarcer.
I can't find a worthy successor to HyperCard, though. IMHO Apple should release it under some open source agreement, so that it can be ported
You're missing the ^G to make it beep after every line.
Amateur.
If it's for-profit but free, you're not the customer -- you're the product (e.g., the Slashdot Beta's "audience").
Because Johnny is an idiot:
* Johnny is the one at your company that has been working there for 10 years
* No one fires him becuase he "knows a lot of things about the process"
* Johnny talks about his technological exploits
* Eveyone else has to debug and fix Johnny's bugs
* He never asks the other developers any question on how to do something, for fear he will look inferior (too late Johhny!)
* AND WHEN YOU NEED A FREAKING IP ADDRESS FROM HIS PC???
You: "Johnny, can you get me your IP Address?"
Johnny fumbles around his desktop, going through all the Control Panel Options
You: "'ipconfig' Johnny"
Johnny: "EYE PEE CONFIG?"
You: "Yes, you know, the command prompt?"
Johnny: "What?"
You: "Never mind... what the status on your code module?"
Johnny: "Module??"
You: *head explodes*
Funny that kids today are pretty good at video games that are longer and more complicated than the ones we played, eh?
.doc files
What kids are after is fun. I did a ton of programming when I was a kid, and I never seriously considered doing it for a living until after I graduated from college. It was just fun.
What's missing is that programming languages used to be built into the computer. For me, BASIC was there when I booted up the computer. Documentation aimed at complete beginners like me was the norm. It would have been hard *not* to program.
These days the easiest way to start out would be with Python. It seems simple to us, but...
1 - know it exists
2 - know how to download the right version for your computer (not a mac or linux version)
3 - find the installer and know to run it (and know that it's safe)
4 - figure out how to invoke the installed program (ok, that one's easy)
5 - find documentation written for beginning programmers
6 - figure out why python won't run the programs you saved as
Can you imagine an eight-year-old Windows user with clueless parents doing all that by himself?
I didn't have to download anything or know anything about operating systems. I don't remember having to sort through tons of titles like "Advanced BASIC Beowulf Architectures in 2 Minutes for J2EE Certified Hardware Astroengineers" at the bookstore to find one that was right for me. In fact, I'm pretty sure the book I used came in the same box as the computer. Plus, all the computer magazines had program listings in the back that you could type in.
Microsoft would earn big brownie points with parents if they included an extremely simple IronPython-based (or even Logo-based) "Learn To Program" IDE with every copy of Windows. Twenty years ago, people knew how to write documentation for beginners. (Not children or computerphobes. Not people with congenital learning disabilities. Just beginners, of all kinds and ages.) I'm sure it could be done again.