I wish that Code Year was 2013 and 2012 was “some smart people with good ideas and a lot of money built took the time to build great pedagogically-driven tool to really solve an existing problem for folks who want and need training in this area.”

So is it me, or is Code Year just Codecademy on a calendar? Oh, and they started to use Meetup. I don’t really see what their $2.5M round is buying yet, let alone how they’re going to make it back. Maybe I just need to be patient. But Code Year is already launched.

Zooming out, I see Code Year as a small (and perhaps overexposed) part of a larger overall movement that includes MIT’s OpenCourseware, Khan Academy, and the Stanford AI and ML courses. There are serious questions about how well this kind of stuff works. But it looks like the train is leaving the station, so we might as well start figuring out how to point the tracks not off a cliff.

Bloomberg says he’s going to do it:

My New Year's resolution is to learn to code with Codecademy in 2012! Join me. codeyear.com #codeyear—
Mike Bloomberg (@MikeBloomberg) January 05, 2012

And now the White House is partnering with Codecademy to create something called “Code Summer+” which I predict will yet again repackage Codecademy courses without addressing questions of instructional design and pedagogy.

Then there’s this inconvenient little point:

Today the White House officially endorsed the *for-profit* company codecademy. Well at least it's not Halliburton!—
  (@zedshaw) January 18, 2012

While the business model hasn’t been revealed yet, Union Square Ventures also hasn’t come out to say that the seed round was actually a charitable donation. It certainly does make things confusing.

But wait — I’m a public high school computer science teacher. Why am I complaining about any of this? I see why Zed Shaw might be bitter pissed, but I’m not trying to sell any books about programming. (Update: Zed Shaw clarifies his position and says Codecademy’s failings actually drive readers to him. Which is fine. My point is just that he makes money in the same business space as Codecademy, where I have zero financial stake in any of this, so why the hell do I care who is getting a boost from Obama?)

My answer echos some of Julie’s comments: I love that there’s an effort to make people more aware of CS and an effort to help more people get into it. I worry that these efforts significantly misrepresent the challenge of learning CS concepts, and put too much faith in people learning on their own. It’s not like the resources and reference materials haven’t been out there on the internet the whole time. Is the problem really that people need game mechanics to help them learn? Or is the problem that this stuff is hard to learn without someone in real life helping you?

Now the idea that folks might just start getting together, outside of traditional institutions, to help struggle through the hard parts is inspiring. We might really be able to move beyond the traditional model of education. But then how do we know the learning is real? As usual, Bruce Sterling called it a couple years out. He wrote a piece for Wired called Favela Chic education back in 2009 whose conclusion was basically, “Who cares about proving it?” So come on, let’s just start learning how to program already! We’re not going to get to the future by sitting around reading and writing blog posts!

OK, but one more wrinkle before we get back on our social networks to check if any of our friends have learned any new skills in the last 5 minutes: Justin Reich just gave a talk entitled “Will Free Benefit the Rich? How Free and Open Education Might Widen Digital Divides”. It’s worth considering the possibility that the efforts being made here, including the projects endorsed by the White House, may have exactly the opposite of the intended effect.

I can’t predict the future. But as a person who spends my days helping kids learn computer science, I can tell you this: People are not going to stop getting stuck on tricky concepts. They’re tricky. People are not going to stop needing expert help and guidance. They don’t know what they don’t know. Unless the online self-learning initiatives and businesses start to take this into account, start to take research seriously, and start to test solutions, a lot of venture capital and venture philanthropy money is going to end up burned. And after all that, we’re still going to need more programmers.

Obviously, this is a pointlessly stupid thing to do. In the year 2012, people have multiple ways to access the internet. And as boring as it is to quote the contract, it seems clear that the district can’t legally limit teacher use of resources:

6.2 Academic Freedom – The District and the Union agree that academic freedom is essential to the fulfillment of the purposes of the San Francisco Unified School District, and they acknowledge that fundamental need to protect teachers from unreasonable censorship or restraint which might interfere with their obligation to pursue truth in the performance of their job with the District.

6.2.1 A teacher’s academic freedom is his/her right and responsibility to study, investigate, present, interpret, and discuss all the relevant facts and ideas in the field of his/her professional competence. This freedom implies no limitation other than those imposed by generally accepted standards of scholarship. As a professional, the teacher strives to maintain a spirit of free inquiry, open-mindedness, and impartiality in the classroom. As a member of an academic community, however, the teacher is free to present in the field of his or her professional competence his/her opinions or convictions and with them the premises from which they are derived.

6.2.2 Within the bounds of Board policies and administrative regulations, as well as adopted state and district curriculums, teachers shall have the opportunity to utilize best practices in employing their teaching methodologies that address students’ different learning styles. Teachers shall also have the discretion to use supplemental materials and develop supplementary lessons aligned with California content standards and district adopted core curriculum.

6.3 Listening, recording, television, or other monitoring devices shall not be used in any part of the building to violate teachers’ rights.

Emphasis mine. I’ll also mention here that I’ve never used TeachersPayTeachers myself. I saw it mentioned in an edtech newsletter, tried to look it up, and found it was blocked. So this isn’t about a pet resource. This is about how and why policy is made. If a teacher finds or purchases materials through this (or any other) web site, and deems them to be relevant in the discretion of their professional competence, then the teacher can use those materials in the classroom. So what possible reason is there for blocking teacher access to this site at SFUSD schools? This looks like a ham-fisted attempt to prevent teachers from sharing or selling curriculum materials. And why would the district want to do that?

SFUSD Board Policy 4540, which at least two of our current school board members claim they don’t know anything about on Facebook (screenshot here), states the following:

1. Patentable or copyrightable materials developed by staff members in the course of carrying out their professional responsibilities on District time shall be the property of the District.

2. Such materials developed partially on District time and partially on the employee’s personal time shall be the property of the District and the employee pursuant to an agreement between the employee and the District.

3. Such materials created by an employee during personal time when not fulfilling contractual duties to the District are the property of the employee.

It doesn’t make sense to me that materials developed by government employees would be anything other than public domain, but this also isn’t about the ill-informed Board Policy 4540. Blocking access to a lesson-sharing or -selling web site on the school network is an overly broad measure that doesn’t do anything to enforce the policy. All it does is limit the ability of teachers to use or even assess the value of a resource at school. All it does is make teachers work at home in order to avoid these kinds of arbitrary and petty censorship attempts. All it does is send the message that although we are trusted to work with your children, we’re not trusted to use the internet. These kinds of actions are indefensible from an institution that claims to value access, equity, student achievement, and accountability.

This is why I’ve been asking for the past week for a reason that TeachersPayTeachers is blocked. Today, SFUSD’s CTO Matthew Kinzie replied to my email, writing:

“I am not aware of any political motivation to block this site. Unfortunately, we lack the documentation behind the original decision to black [sic] the site. John approved ITD unblocking this site a couple of days ago.”

How many other web sites are blocked without any documentation of the original decision? That list of sites, however long it may be, represents SFUSD going above and beyond what their contracted internet filtering vendor, M86, blocks for CIPA compliance. And now we learn that this extra layer of censorship is not connected to anyone who can explain why or how it exists. Why are we running our school district as if it were a repressive political regime?

The person that CTO Kinzie refers to approving the unblocking is Dr. John Rubio, Educational Technology Supervisor. How he ended up with the unenviable responsibility for making the school district’s censorship decisions is a question that only his boss, Executive Director of Academics & Professional Development Bill Sanderson, can answer. Ultimately, Superintendent Carlos Garcia has the power to direct how these censorship policies (or informal policy decisions, since none of them are ever made public) are set and implemented. Meanwhile, SFUSD continues to devote resources to limiting student and faculty access to the internet, while failing to provide even basic email accounts for students. It looks like it’s going to be a long 21st century for those of us working here.

Without going buck wild and consulting some kind of dictionary-like resource, I figure it means at least three different things. First, a link is a connection, as when we use the href attribute of the a tag to make clickable text that will take you to another site. Second, the verb form means either creating or following one of these connections. Third, and most confusingly, there’s actually a link tag, which we most commonly use to import CSS rules from an external file.

Think about this for a second: When you specify the href of an a tag, you’re giving the URL for the client to visit and display when the stuff inside that a tag is clicked. We usually call that whole structure a link. But when you specify the href of a link tag with rel="stylesheet", you’re giving the URL of a resource that should be incorporated into the current document. No wonder the kids are confused! Remind me to tell next year’s class this little secret.

Student answer: “Abandoning the original reference for another (yea like dumping a girlfriend for another one, so the original is up for grabs (or gc)”

Ria Galanos (Georgia CSTA co-founder and a fellow instructor in Google’s CAPE program) was planning to write something about all this on the CSTA blog — so I’m calling her out on the internet. I’m not just naming Ria to make her blog, but also to thank her for an idea she shared with me. I was complaining that I don’t know how to reach some of my students who get lost early on in the course. Ria said that in her APCS class, she has students write a method on paper every day and this formative assessment has cut down on the amount of nonsense code she sees. I decided to try it.

For the past two weeks now, my AP students have written the body of a simple method every day at the start of class. They have 5 minutes with no notes or other resources. The problems are supposed to be quick review — like CodingBat on paper. For example, one recent problem said, “The method charCount takes an array of Strings and returns the total number of characters they contain.” The prompt also gave two example inputs with expected outputs, and then asked them to write the method body.

Later, I sort the answers into three categories: “got it” which means the code as written would compile and produce the correct answer for all inputs, “close” which means there’s a minor syntax or logic error but the overall answer shows an understanding of a possible solution, and “way off” which means confusion about how to approach the problem or major syntax misunderstandings.

Here’s how we’ve been doing:

What does that mean? Well, partly that practice helps. In the first week, I gave problems that all required a loop controlled by an parameter. Students got better at solving them. In the second week, I started giving problems with an array parameter which confused a lot of them. But after a few more of those kind of problems, results improved again. Introducing a twist (like I did in the most recent question) reveals that some of the “close” answers are just mimicry or pattern-matching and not durable learning. Long-term impact or improvement is currently an unknown. You can download the PDF of my first 10 questions here.

Despite the unknowns, there are three things I like about this practice: One, it adds some routine to the day and establishes a space to talk about the little details of programming. Today we talked about the confusion that can arise from using “i” as a non-index variable name. Two, I like getting regular feedback about who is where in terms of their understanding. I can talk to individual students about their work without the external force of a grade weighing on us both. And three, I like the reality check — we’re not where we need to be yet as a class.

I wouldn’t have started doing this without Ria’s suggestion. Teachers of most other subjects have a whole department to talk to about what they’re doing and share their results. Since CS is a small elective at most schools, our professional learning community has to take place online. Good thing y’all know how to use computers!

In terms of learning curve, students need to have a solid understanding of inheritance and polymorphism before abstract methods make sense, and even then they sometimes wonder why they exist and why a class has to be marked abstract if it contains an abstract method.

It’s been suggested to me that teaching interfaces before even teaching inheritance may help this make more sense. That’s something I’d still like to try.

On a recent homework assignment, I asked, “Now we know it is possible to declare classes abstract. What does it mean to be “abstract” and why would we want a class to have this property?”

The range of answers says something about this challenge. Some students basically said they had no idea:

To abstract means to tell the compiler to stop the code that is running and we want a class like that because making sure that the codes are in the correct place is important.

Others knew the mechanism but didn’t see the utility:

Being abstract means that the class can’t be instantiated. We would want a class to have this property because that will cause the compiler to stop any code from making an instance of that type.

And some, of course, had a better understanding:

“Abstract” means that it is just a reference class that is used for polymorphism. We would want to have this property because it guarantees that the an object of that type will never be created.

I think Java is actually a nice teaching language since it includes some of these concepts without making them as complex and option-heavy as C++, and requires a strict object-orientation and strong typing unlike the syntactically simpler scripting languages. The hardcore among you will point out that the right way to understand all of these language features is to implement them yourself in Lisp, and while I tend to agree, it seems like there just aren’t that many students who can start there.

I read Ben’s great post about his students using Codecademy and felt I had to respond to explain a bit more about what we’re working on and how we plan to change the way CS is learned across the world.

Our Mission

Codecademy was founded in August of 2011 by Ryan Bubinski and I.  Ryan had spent four years at Columbia helping CS majors and others learn how to practically apply their programming skills.  Through the Application Development Initiative, an organization he founded, hundreds of Columbia students learned (and often taught each other) practical programming skills and technologies.

At the same time, I was working at a bunch of startups (drop.io before it sold to Facebook and GroupMe before it sold to Skype) working on product and business.  It became increasingly frustrating that I didn’t know how to code so Ryan helped me chart a path towards programming literacy.  I took an introductory CS class at Columbia and read several books, along with a healthy complement of videos and tinkering.  Yet for some reason, programming never clicked the way I wished it would.

Ryan and I started Codecademy with the mission to make sure others didn’t have that same problem.  We’re trying to teach the world to code in a way that’s more fun, exciting, and motivating than what exists now.  We’re firm believers that code will be part of twenty-first century literacy, and we’d like to be one of the biggest educators of twenty-first century citizens.

Startups and Education

There’s a bit of a problem for education startups – the lifecycles and timing of both don’t quite match up.  We’re a small team (and released our first iteration when it was just the two of us) and we’re used to building and changing our product quickly before it’s perfect.  Hierarchical education is a bit different – lesson plans are established and teachers have set methodologies of teaching.

As a startup, we think a bit differently.  We tweak our curriculum every day depending on feedback from users and data we receive every time an exercise is completed.  We see where our users are stumbling and we get hundreds of comments and emails a day.  We’re constantly refining our curriculum.  That means the first iteration that Ben saw is different (as he mentions in his addendum) from the one on the site today.  We think using data and feedback from our students is the best way to create the best, most informative lessons.  That said, lessons are almost never perfect (just as they’re rarely perfect in the classroom).

Different Contexts

We’re thrilled to see that teachers and educators have used us in their classrooms.  We’re excited about becoming a resource for students both inside of and outside of the classroom.  But while we’ve been happily surprised by how quickly this use case has grown, it’s also not one we had planned on from the beginning.  We know we have lots of work to do before Codecademy works right out of the box for all ages, grade levels, and skill sets out there.  Ultimately we’re looking to create a good educational experience for everyone – including teachers and students.

Help Us Out!

As I said, we’re always learning.  We’re starting to work closer with our community by acting on their feedback and, most importantly, beginning to incorporate their lessons.  What we’ve built is merely a platform for others to create lessons on.  Think there’s something wrong with our introductory JavaScript class?  You’re welcome to create your own with our course creator tool.  We can only get better with your help.  Thanks!

It’s great to know that Codecademy is watching their Google Alerts and paying attention to what people are talking about. I was hoping for a bit more in this response on the specifics of the issues I raised — most notably, the problem that completing an exercise immediately pushes you to the next exercise without a chance to see your correct solution in context. It’s nice that they’re open to feedback, but I have a feeling most people will be wary about donating curriculum-improvement efforts to a for-profit company unless there’s a clear open-content commons-style agreement in place.

While I don’t doubt Codecademy they can turn the technology they have into some kind of business (perhaps after a pivot or two), I’m not sure if the actual lessons on offer are any better than, say, Learn Python the Hard Way or Eloquent JavaScript. This is not to put down anyone’s curriculum, but just to say that this subject is hard to present. People have been trying for a long time and no one’s really figured out how to dial in the difficulty, pace, and number/type of exercises that will work for a wide range of learners.

I tried out Codecademy with my 10th grade class last week. The students have some JS experience but are mostly new to programming overall. We started with the first course written by Ryan Bubinski, one of Codecademy’s founders. All was smooth sailing until Lesson 4, where some of my students stumbled.

Although they had just apparently learned (or at least typed in stuff) about variables in Lesson 3, some students couldn’t complete the instruction “Declare and set a variable named number to the value 42.” What was wrong?

There was a hint available, but they couldn’t understand it: “To declare and set a variable in a single line, you use var variableName = variableValue, where variableName is the name of the variable, and variableValue is the value you want to assign to the variable. For this exercise, the variable name is number and the value is 42.”

Many of the students who got stuck at this point were trying to literally enter, var variableName = variableValue. In fact, they had been directly copying all the other code up to that point with some success.

That hint seems reasonable when you just read it, but if a student needs a hint at this point in the lesson, maybe the hint should be, “Go back to the previous lesson and do it again while paying attention to the meaning of the symbols you’re typing.” And as long as we’re presenting this as a web app, why not offer a click-through for that or even require completion of more exercises on the earlier concept?

The next stumbling block I saw broadly in the class was substrings. Everyone can complete the exercise that says, “To get the substring, you can type “hello”.substring(0,2). Try that now.” Again, they all just type in what is says to type, sometimes without comprehending the text at all.

The next exercise says, “This time we’re going to get the first three letters of your name. See if you can save those letters to a variable named three using the substring() method from above.” Whoa! Huge leap there. Not only are we hoping for them to have understood the concept of substring, we’re also hoping they remember how to assign a value to a variable. And as if that wasn’t enough, it seems we’re also hoping to gloss over the part about how a statement gets evaluated to a value before assignment takes place.

Some kids literally have no idea what the instructions at this point mean. “See if you can save those letters to a variable named three using the substring() method from above.” They haven’t ever seen an example of a function’s return value being assigned to a variable. If they don’t get it, should we keep making them guess?

Either way, they guess away and eventually get it or get someone to help them do it. Then, right at the very moment of success, the interface jumps them on to the next exercise. It’s hard to know what you did that was correct. Particularly if a student has been working and working away in the editor at something, why should it disappear as soon as it’s right? There’s no time to reflect on the knowledge they just gained, nor an opportunity to try variations or improvements on the solution. The interface encourages the very sort of “click and forget” behavior that works against learning.

While this pedagogical problem with the site is pervasive, Ryan’s initial course does a pretty good job of using the interpreter as a learning environment. I can’t say the same for the next course offered: JavaScript Quick Start Guide. Now to be fair, this is targeted at folks who already know how to program. But it might as well just be an HTML file, because it doesn’t take advantage of the interpreter at all.

Many screens are just reading with no typing code at all. We hit enter, and the system assumes we understood, assumes we could contextualize the information.

Then we get to Lesson 3, Exercise 3: “Bang Operator”. The instructions say “Work through the example and then hit enter to continue.” And this is what’s in the editor window:

function sad() {
  // If a value is not true, return the string ':('.

}

Anyone have any idea what to do here? Yeah, me neither. This is my best guess:

function sad(s) {
  // If a value is not true, return the string ':('.
  if(!s) return ':(';
}

But that wasn’t the desired answer. So my top students gave up there for the day, me along with them.

Update 11-9-2011: They fixed the bug in Lesson 3, Exercise 3 to match what I assumed was the original intention. The structural problems with reinforcement remain. In the new DOM and jQuery course, Lesson 2, Exercise 4 it asks you to select the elements from a particular class. But you can select the elements from any class — as soon as you push enter it drops you right into the next exercise, which isn’t an exercise at all, just a demo of adding elements to the DOM. All you can do there is click “Run” and you’re now two lessons away from where you got confused in the first place.

And while I celebrate this opportunity for millions of iPad users to casually buy an app that opens up the world of programming, I’m confused as to why Apple allowed it in the iTunes store but last April rejected a Scratch app. Dr. Mitch Resnick (MIT professor, developer of Scratch, and winner of the 2011 McGraw Prize in Education) wrote at the time:

“We’re disappointed that Apple decided not to allow a Scratch player on the iPhone or iPad (as part of Apple’s policy against apps that interpret or execute code). As we see it, there is nothing more important than empowering the next generation of kids to design, create, and express themselves with new media technologies.” – Scratch Team Blog

Yet it’s clear that user-written Lua code is being interpreted in Codify. Searching for Lua on the iTunes store also brings up apps like iLuaBox, Lua Coder and Lua Console. These all seem to go beyond the limited case of apps that happen to rely internally on non-compiled code.

As of a year ago, the Scratch app was still a non-starter. At that time, the Lua community itself seemed to believe that, “[Y]ou cannot expose a plugin system to the end user, nor can you give her the ability to write her own scripts.” What changed?

I notice that Codify has no provision for uploading/downloading projects and I wonder if that was the deal-breaker for Scratch rather than the idea of interpreted code. Apple seems to allow sync with DropBox and other file hosting services. So is the rule just that you can’t do interpreted code and upload/download in a single app? Or perhaps Lua is the magic word for iTune Store acceptance, and simply naming the Scratch app to LuaScratch would do the trick.

Regardless, it seems to me that if Codify and the other Lua programming apps meet Apple’s requirements, then a touch-based version of the Scratch development environment should be allowable as well.