MIT Media Lab Making Programming Fun For Kids

An anonymous reader passed us a link to an article on the Boston Globe's website, talking up efforts by MIT to make programming a non-threatening part of grade-school education. MIT has developed a new programming language designed to encourage experimentation and play. Called Scratch, the project eschews manuals and high-level concepts in favour of approachability. "Efforts to make computer programming accessible to young people began in the late 1970s with the advent of the personal PC, when another programming language with roots at MIT — Logo — allowed young people to draw shapes by steering a turtle around a screen by typing out commands. But the path to mastering most programming languages has been strewn with obstacles, since students needed to figure out not only the underlying logic but also master a brand new syntax, observe strict rules about semicolons and bracket use, and figure out what was causing error messages even as they learned the program."

Lego Logo

I learned Lego Logo as a grade schooler in summer school. Great fun! Definitely one of the things that influenced my youth leading me into a CompSci future.

-Rick

Re:Lego Logo

Pfft. You whippersnappers have it easy! I had to wire my logic gates by hand, and that's the ways I likes it! That was after I mined the copper to make the wires! Plus, I had to use an exercise bike to generate the electricity to power the damn thing!!!

What?!

You mean BASIC isn't fun? It was fun for me... but maybe that's why I'm reading Slashdot now.

Re:

[Accidentally posted AC, reposting b/c I'm a filthy rotten karma whore]

After reading your post, I initially agreed with you. Then I remembered all the horrible, horrible crap I wrote in basic.

The problem with basic, as I experienced it, is that it never re

Re:What?!

I've not seen this MIT project, but Logo then Pascal is a good introduction to programming for kids. Logo teaches them to think in terms of processes, as well as teaching the importance of syntax. It also gives them a visible product that they can take pride in and appreciate... "Hello world" just isn't as inspiring to kids as seeing a colored square on screen that they made.

Pascal is more like modern programming languages, and while it has its problems, it's simple enough for a preteen to use.

As for your comment that BASIC gets slagged on slashdot -- I think typically it's VisualBasic that gets slammed, for giving people the tools to get a bit of programming done without making sure they have programming concepts down. People who learn to program in VBA learn a lot of bad habits, and if they start doing real development instead of basic scripts, they don't have the background necessary. It's not so much VBA that sucks IMO, it's the fact that so many VBA users learned how to write code without learning how to program.

Re:What?!

Logo teaches them to think in terms of processes, as well as teaching the importance of syntax.

No, Logo teaches them to push a turtle around the screen. It doesn't really convey a sense to young children that they're "programming" a computer. I technically had Logo before I ever had BASIC, and it took me years to realize that it was supposed to be an introduction to programming. Most of us saw it as an introduction to computer graphics.

As for your comment that BASIC gets slagged on slashdot -- I think typically it's VisualBasic that gets slammed, for giving people the tools to get a bit of programming done without making sure they have programming concepts down.

While Visual Basic is a poor tool to teach programming (most "programs" taught are simple GUI constructs with little to no code), the original BASIC regularly gets slammed because of Dijkstra's 1968 article, Go To Statement Considered Harmful [acm.org]. Dijkstra's core argument was that GOTO statements created spaghetti code. While this is unavoidable in assembler, his point was that it does not need to exist in high-level languages.

That paper had a profound effect on languages that followed, resulting in many modern languages doing away with a GOTO keyword altogether. (e.g. Java reserves GOTO, but does not implement it.) Taken by itself, Dijkstra had a point. Unfortunately, he went on to say: "It is practically impossible to teach good programming to students that have had a prior exposure to BASIC: as potential programmers they are mentally mutilated beyond hope of regeneration." This started the idea that BASIC is somehow the "wrong" way to teach programming.

The truth of the matter is that the design of BASIC will only limit programmers who are not interested in a long term career (or at least hobby) in computer programming. Most BASIC programmers quickly find the limitations of the GOTO statement on their own, and need little prodding to move to subroutines via GOSUB calls. From there, a programmer quickly learns the limitations of global variables. This makes the introduction to procedural functions much easier.

Basically, it's easy to provide a student with new tools when they feel the need for them. If you simply give them the tools without giving them the background, they will never learn to use the tools correctly. That's why I personally believe that classic BASIC is still an excellent teaching tool. Besides having simple syntax that any child can understand (one instruction goes after the other, see?), the interpreter environment allows children to play around with the instructions without having to write complete programs for each experiment. This invaluable teaching feature is lacking in modern structured programming.

Thus it is my personal belief that we need to STOP reinventing teaching languages, and just go back to what works. All we're doing with these new languages is giving them the CompSci version of "New Math". And all that "New Math" ever accomplished was to generally confuse children, and ensure that they never take up higher maths. Such is the result of providing highly structured coding tools to a child who wants to explore.

You can read more of my thoughts on this subject in this article [intelligentblogger.com].

Re:

When I started messing around with BASIC, my older brother simply told me to never use GOTO. "Don't ask why, you won't understand, just don't use it". I rarely did, and I didn't miss it. People can absolutely be taught programming in BASIC (in fact, I'd ar

Re:

No, Logo teaches them to push a turtle around the screen. It doesn't really convey a sense to young children that they're "programming" a computer.

Well, first, their cognizance of whether they are "programming" isn't as important as the concepts they are l

Re:

Well, first, their cognizance of whether they are "programming" isn't as important as the concepts they are learning.

Untrue. If they don't know what they are learning, then how can they ever apply it? Furthermore, programming courses attract those students

Re:

Yes, VB Script is a lousy excuse for a language.

But as far as tools go; sometimes you just need a crowbar, not a screwdriver, or a scalpel.

If you're writing an office suite, or 3d simulation, of course you don't want to even consider something like VB Scri

Re:

Especially ironic since the whole point of Pascal was to be a teaching language rather than a "practical" one. In how many other fields do students start out with the fully capable tools that a master would use?

Re:What?!

Yeah, Pascal was a stricter language but IMHO the rules were less confusing than those of C++ and Java. The biggest problem that I've had with using Python for teaching young students has been the opposite - what would be a simple error in a statically typed language can become a more subtle error in python, and the error messages that it produces are really quite poor compared to even a C compiler (although not nearly as bad as C++ STL errors :). Because of this the kids I've worked with have a harder time debugging python programs than with other languages. On the other hand the ability to type lines directly into the python console is very useful and encourages learning by experimentation.

Re:

BASIC generates ire as a backlash against its marketing. The language would be long gone from this world had it not been called BASIC, but the name makes people think it's easy or appropriate for beginners or something. In fact, by comparison to modern lan

Hell

Yeah make it non-threatening so that they won't even have an inkling of the Hell that is computer science.

Re:Hell

What's intrinsically hellish about computer science?
The problems I see with it are related to the entropy of the human soul. Gets especially painful when the entropy aggregates into organizational behavior.
I, for one, find reading Knuth a delightful escape from Perry Ferrel's observation: "...and the news is just another show / with sex and violence..."

Re:Hell

What's intrinsically hellish about computer science?

Hellish to non-coders. And I use "coders" there instead of the more generic "geek", because most people with a near-obsessive interest in something can qualify as some form of geek, while very few people can really code well.

You don't just need to know "the" language (sign #1 that coding doesn't suit a person - They want to learn C or Java for a few specific purposes, rather than "how to code" and "how it works" - The language doesn't matter, within reason). You need a particular type of personality (near obsessive). You need a clear mind (I mean that in the Zen way - In my teens I tried "meditating" a few times and always found it frustrating that the guides made no sense, with phrasing like "stop your internal monologue"; I finally realized that while most people apparently can't shut the voices in their head up, I have no internal monologue that needs silencing, and consider that a BIG part of what makes me a decent coder). You need the ability to think really, truly logically. The ability to sit motionless for hours at a time really helps. Finally, and perhaps most importantly, you need to break arbitrarily complex tasks down into atomic actions (which goes along with thinking logically, in the proof-theory sense).

All of those, to most people, sound hellish. Thinking in terms of formal proofs? Quieting your internal voice enough to think over it? Sitting motionless at a computer for so long your SO/family needs to remind you to eat ten hours later? Most people don't want that.


I hate how this topic usually boils down to the stereotypical us-vs-them, "Real coders do/don't"... But sometimes, you just can't escape the facts. Most people can't code, which doesn't state a temporary lack of training but rather an outright permanant inability.

Re:Hell

What do you do when you think, speak out loud without any thinking ahead of what you're going to say?

I don't claim that I don't think about things, or even daydream just like everyone else. I just don't do it in English (or any language that ever could exist outside my own head). I can even think in words - You probably gave the best example, when I think about how to phrase something, I do so "in" the language itself; Oddly, although I only speak one natural language (English), I do the same thing when coding - I "think" in an internal voice speaking C, for example.



Let's say you shut your eyes...do you notice anything around you?

Yes, of course - I don't claim myself in a coma. ;-)
But "conscious" doesn't mean "words". I meant more than I don't have, hmm, a narrator, I guess? As I mentioned, I found it quite a surprise when I first learned that most people do. As I understand it (second hand of course), most people would internally "say" something about almost all of the major things that pass into their awareness; I don't do that.



I've never talked to anyone who was absent an internal monologue.

Think of the smell of a crayon. Do words suffice to describe it, or did your first burst of thought contain a wave of sensory impressions and memories that include kindergarten, wax, some little girl's hair, pictures on a refridgerator, the sound of an ice-cream truck, and far, far more than that, all in one burst? Just typing that, I tried to touch on a few of the points of what the smell of crayons makes me think about, and found it incredibly restrictive. Imagine always thinking in terms of that initial burst, and you have the idea.

Re:

Boss, I don't see what any of your points have to do with computer science in the abstract.
As I noted in the preceding post, your pain points have to do with people, not CompSci.
Is this an example of Post Soul-Crushing Meeting Disorder? If so, you've m

Re:

That's IT, not Comp Sci.

Programming in and of itself is not Comp Sci.

I politely disagree: CS is beautiful.

IT is hellish bullshit.

CS is pretty, applied math. And the culture of computer scientists is creative, inventive, and intellectual. Hell? No!

(This distinction, others have pointed out before me.)

More, some exposure to CS teaches people how to think. Before I started to program, I was horrendous at math. Every standardized test I ever took told me I should be a writer. But by turning logic into play, the computer changed everything. Sure, I can still barely add. But I'm going for a Ph.D. in theoretical control -- which is essentially an applied math field. Because, give me a calculator, and I can do pretty cool stuff.

How many people "hate math" because they think it's all about adding up numbers? Tons! (Including, unfortunately, most of the elementary school teachers who teach math). That's not what it's about! Computer Science is beautiful. It changed my mind, and my life: That's no overstatement.

My first language, as a child? QBasic.

I still like logo

Drag and drop seems nice, but it is a significant abstraction from real programming. My kids have both learned a bit about programming from logo, and they are 4 and 5.

Re:

Drag and drop seems nice, but it is a significant abstraction from real programming.

It doesn't have to be all that distant from raw code. Another MIT project (StarLogo TNG [mit.edu]) uses drag and drop that has a pretty much 1:1 relationship to raw code, but is pre

Now if only they could make programmers

not frightening to children....or women for that matter :P

Re:

He's not talking about making "programming" less frightening to women, but "computer programmers" less frightening to women (i.e., pimply-faced male coders who cannot, for the life of them, get a date with the opposite sex). Of course, he either assumes th

Not Possible

I made a call to Michael Tiemann, author of the GNU C++ compiler, a few years ago to encourage him to create a programming extension to his work with gnu C++ by adding graphical symbols to C++ which would allow people, especially children, to program in C++ by manipulating graphical symbols the way that C++ programmers now manipulate text to create software.

He said it was impossible.

All that means, really, is that it won't be Michael Tiemann who authors or participates in this inevitable breakthrough.

It is a bad idea.

C++ is a very complex language, and whether it is represented by text or graphics you will have the same difficult concepts to learn. Most of those concepts exist either for performance reasons, or as an aid in creating very large programs (they trade-off more up-front learning and work for less problems later on). Neither of these are desirable for a graphical learning language, nor is it desirable to build off of a compiled language. If you did create a graphical representation of C++ it would be an overly complicated mess that was no easier to program in than textual C++.

You are better off creating a your own language (like this or LabView or Squeak or the newer graphical Lego Logo) than to try and retrofit C++, or worse to call on someone whose strengths are in low-level machine language generation and optimization to do it for you.

Yes, more or less.

In the late 70's I began thinking about and, by 1986, created a way for people in restaurants to work more efficiently by manipulating graphical symbols on touchscreens. By doing this they could walk far fewer steps, stop having to create guest checks by h

Whats the point?

Are there courses designed to make neurosurgery less intimidating to kids or genetic research less complicated or elite forces soldering less dangerous or stressful? It always concerns me when I see a bunch of geeks trying to stick programming down the throats of kids rather than focus on teaching them the real skills they need at that age.

Re:

There are machines and methodologies designed to make neurosurgery less intimidating to neurosurgeons. There are courses designed to teach neurosurgeons how to use machines and methodologies less intimidating to neurosurgeons.

The kids will develop machine

Re:Whats the point?

Thank you! EXACTLY what I was going to post. Screw programming! Certainly kids who are interested in that should be encouraged, but it's a VERY small minority that have a true interest. There are far more important skills that we should be encouraging.

Such as? How about true art training? Studies (which I don't have a link to) have shown that kids that are taught to draw realistically tend to do better in ALL subjects, probably because of the quiet concentration that it requires. Kids as young as 4 or 5 can be taught to do realistic art, but even a lot of art schools don't do beginning classes until 8 or 9, and the closest typical schools get is just letting the kids slap paint on paper without any instruction at all. Only gifted people learn to play piano by banging keys, and only gifted people learn to draw by scribbling. Yet anyone can learn piano through instruction, and anyone can draw realistically through instruction as well.

Sorry for the pseudo-rant on art classes, but I've been looking for art instruction for my young children, and it's very difficult to find. I finally found great book [amazon.com] and I'm doing it myself. :) Note the picture on the cover that was done by a non-gifted five year old, BTW.

Not that I think it's a bad idea

but this is programming in the same way that updating your blog is creating a web site. Pedantic, I know, but important in view of how people feel about H1-Bs and lack of scientific/engineering graduates in the US. It will be interesting to see how much this acts as a gateway to more people taking up programing as a hobby or vocation.

Just think...

Just think of all the Microsoft patents these kids can now infringe!

Kid Programming tool - RoboRally!

Best way for kids to learn how to program is a simple game.

ROBORALLY! [wizards.com]

You "program" your robot with cards from your hand placed in a certain order. A turn proceeds and the cards are executed. If all goes well, you hit waypoints, and blast a few other robots to dust on the way.

Clearly

Logo and scratch aren't really relevant for kids to learn at a young age. This is what C and assembly are for.

Reminds me of Alice

Reminds me a bit of the 'Alice' project from CMU - they seem to have a similar visual programming metaphor:

http://alice.org/ [alice.org]

Programming is fun to begin with!

Being able to create nearly anything you want on a computer, thinking through puzzles, showing your creations to your friends, the peership of programmers, learning an endless stream of new things -- programming is tons of fun! I started programming when I was 5 years old thanks to an Apple IIe home computer and have never stopped since. When I first saw the BASIC and LOGO programming at elementary school, my impression was that they weren't do it in the 'fun' way at all: we were supposed to just copy down what they did and no there was no real opportunity for exploration. Having taught programming a few times since, it all kind of weaves together: learning programming is more of a journey of aided discovery than memorizing route information. I think there is a contrast between that and most teaching. It sounds like Scratch is more about the exploration, which is great. And, you know there are gazillions of CS students who would love programming to be more fun as well!

Re:

When I first saw the BASIC and LOGO programming at elementary school, my impression was that they weren't do it in the 'fun' way at all: we were supposed to just copy down what they did and no there was no real opportunity for exploration.

The purpose of copying a program designed by someone else (taking their listing in a book or whatever and typing it in) serves three purposes:

1) gets you familiar with the process of inputting code, so that when you write your own code, you'll already be

BBC Scratch Article with Video

BBC Scratch Article with Flash Video [bbc.co.uk]

hmm

what happened to logo?

Oh great

Now all they need to do is ship this on the OLPC, to make sure all US programming jobs are obliterated 10 years from now.

Looks a lot like...

...the original programming "language" for Lego Mindstorms. That one got me so frustrated with its limitations that I got stuck into NQC ("Not Quite C"), a "real language" for Mindstorms, as soon as I possibly could.

Some will never push the boundaries of Scratch, never discover its limitations. But for those who do, those limitations could well be exactly what drives them to try "real programming" - maybe using Javascript and CSS to push things around on a page. Who knows where they'll go from there?

bit like squeak

which my kids use. Squeak is based on smalltalk and is a gentle introduction to object oriented programming concepts

Re:bit like squeak

Although Scratch doesn't seem to make this clear, Scratch IS Squeak (well, an even easier to use wrapper anyway), and it's listed as a Squeak project http://www.squeak.org/Projects/ [squeak.org]
Try for yourself... download Scratch, drag the included image onto your Squeak VM and it'll open fine (although, at least on the Mac version I can't find a way to quit properly).

Hackety Hack

Similar to Scratch,

why the lucky stiff [whytheluckystiff.net] has started an amazing project called Hackety Hack [hacketyhack.net], in an attempt to solve the Little Coder's Predicament [whytheluckystiff.net]. It's a development platform designed for the younger coders and beginners, with an emphasis on sharing, community, ease-of-use (lots of built-in functionality), and cute cartoon characters. Currently it teaches Ruby in a series of fun lessons, but _why has stated that it might teach other languages in the future. A slick help interface comes bundled, as well as a Ruby cheat-sheet.

Come and join in the public beta testing. The forum is active and the people are nice. And don't forget to share your exciting hacks with the rest of us!

--

Eli

Logo? Meh.

Efforts to make computer programming accessible to young people began in the late 1970s with the advent of the personal PC, when another programming language with roots at MIT -- Logo -- allowed young people to draw shapes by steering a turtle around a screen by typing out commands.

From what I remember of Logo, few people in the class "got" it. Everyone in CS harps on and on about how great logo is, but most of my classmates in grade-school just laughed when the "turtle" did stupid things, and asked the teacher for help (ie, to fix it for them.)

To say teaching Logo "teaches programming" is akin to saying that having your kid watch you inflate your tires is "teaching car repair."

Re:Logo? Meh.

From what I remember of Logo, few people in the class "got" it. Everyone in CS harps on and on about how great logo is, but most of my classmates in grade-school just laughed when the "turtle" did stupid things, and asked the teacher for help (ie, to fix it for them.)

Yes, one of the big failings of Logo is that although it had the potential to help make kids smarter, it couldn't do anything about the teachers.

Disclaimer: I wrote Logo for the C64, Apple II, and Mac.

Tomorrow on Sesame Street

Grover, "Hey kids! The word of the day is... Recursion! Brought to you by the color #CCCCFF"

Welp

I just downloaded Scratch and in a few minutes made a picture of a pig move around the screen while rotating and making fart noises. Honestly, this is all I really want out of any programming language.

Re:Python as a starter language

I think these researchers should use Python and form a child friendly language derivative. It has clean syntax, and makes it easy to express a lot of hard concepts. Plus it has a live interpreter, which is like Logo. This way, they can learn programming in a easy environment and when they build confidence to do something more complex, they will have an excellent language to start from.

I've used Python to teach elementary kids and, while it was mostly great, it really lacked a good graphical/audio system. We tried turtle and Pygame, but neither of them were even as easy and fun as the old setpixel, drawrect procedural style of BASIC, pascal, etc. I wish Python had a nice simple drawing module that can with the standard build (and Tk doesn't count imo.) Did I miss it? :)

Re:

Pygsear [nongnu.org] is supposed to simplify teaching programming via Python and graphics. It's implemented as a layer on top of PyGame. The author is writing a textbook [easthighschool.net] for a course using it. I haven't used it, so I don't know how effective it is, but it seems to i