Total 21 Posts

## Three Claims Function Carnival Makes About Online Math Education

Today Desmos is releasing Function Carnival, an online math happytime we spent several months developing in collaboration with Christopher Danielson. Christopher and I drafted an announcement over at Desmos which summarizes some research on function misconceptions and details our efforts at addressing them. I hope you’ll read it but I don’t want to recap it here.

Instead, I’d like to be explicit about three claims we’re making about online math education with Function Carnival.

1. We can ask students to do lots more than fill in blanks and select from multiple choices.

Currently, students select from a very limited buffet line of experiences when they try to learn math online. They watch videos. They answer questions about what they watched in the videos. If the answer is a real number, they’re asked to fill in a blank. If the answer is less structured than a real number, we often turn to multiple choice items. If the answer is something even less structured, something like an argument or a conjecture … well … students don’t really do those kinds of things when they learn math online, do they?

With Function Carnival, we ask students to graph something they see, to draw a graph by clicking with their mouse or tapping with their finger.

I’d like to know another online math curriculum that assigns students the tasks of drawing graphs and arguing about them. I’m sure it exists. I’m sure it isn’t common.

2. We can give students more useful feedback than “right/wrong” with structured hints.

Currently, students submit an answer and they’re told if it’s right or wrong. If it’s wrong, they’re given an algorithmically generated hint (the computer recognizes you probably got your answer by multiplying by a fraction instead of by its reciprocal and suggests you check that) or they’re shown one step at a time of a worked example (“Here’s the first step for solving a proportion. Do you want another?”).

This is fine to a certain extent. The answers to many mathematical questions are either right or wrong and worked examples can be helpful. But a lot of math questions have many correct answers with many ways to find those answers and many better ways to help students with wrong answers than by showing them steps from a worked example.

For example, with Function Carnival, when students draw an incorrect graph, we don’t tell them they’re right or wrong, though that’d be pretty simple. Instead, we echo their graph back at them. We bring in a second cannon man that floats along with their graph and they watch the difference between their cannon man and the target cannon man. Echoing. (Or “recursive feedback” to use Okita and Schwartz’s term.)

When I taught with Function Carnival in two San Jose classrooms, the result was students who would iterate and refine their graphs and often experience useful realizations along the way that made future graphs easier to draw.

3. We can give teachers better feedback than columns filled with percentages and colors.

Our goal here isn’t to distill student learning into percentages and colors but to empower teachers with good data that help them remediate student misconceptions during class and orchestrate productive mathematical discussions at the end of class. So we take in all these student graphs and instead of calculating a best-fit score and allowing teachers to sort it, we built filters for common misconceptions. We can quickly show a teacher which students evoke those misconceptions about function graphs and then suggest conversation starters.

A bonus claim to play us out:

4. This stuff is really hard to do well.

Maybe capturing 50% the quality of our best brick-and-mortar classrooms at 25% the cost and offering it to 10,000% more people will win the day. Before we reach that point, though, let’s put together some existence proofs of online math activities that capture more quality, if also at greater cost. Let’s run hard and bury a shoulder in the mushy boundary of what we call online math education, then back up a few feet and explore the territory we just revealed. Function Carnival is our contribution today.

## The Digital Networked Textbook: Is It Any Different?

Let’s speculate that before this year’s cohort of first-year teachers retires from math education more than 50% of American classrooms will feature 1:1 technology. That’s a conservative prediction – both in the timeline and the percentage – and it’s more than enough to make me wonder what makes for good curricula in a 1:1 classroom. What are useful questions to ask?

Here’s the question I ask myself whenever I see new curricula crop up for digital networked devices like computer, laptops, tablets, and phones.

Is it any different?

That isn’t a rhetorical or abstract question. I mean it in two separate and specific ways.

Digital

If you print out each page of the digital networked curriculum, is it any different?

The answer here is “sort of.”

When I look at iBooks in the iBookstore from Pearson and McGraw-Hill or when I see HMH publish their Algebra Fuse curriculum in the App Store, I see lots of features and, yes, they require a digital medium. They have a) interactive slider-type demonstrations, b) slideshows that walk students through worked examples, c) stock video in the margins instead of stock photography, d) graded multiple-choice quizzes, e) videos of Edward Burger explaining math concepts and f) probably other items I’m forgetting. None of those features would survive the downgrade to paper.

So the question becomes, “Is it different enough?”

Are these offerings different enough to justify the enormous expense in hardware, software, and bandwidth? Do they take full advantage of their digital birthright?

I don’t think so.

Networked

“Is it any different?” here means “if you were hundreds of feet below the surface of the Earth, in a concrete bunker without any kind of Internet access, is the curriculum any different?”

Here, in September 2013, the answer is “no,” which is a shocking waste of very expensive, very powerful device.

Look at the apps you have on the home screen of your smartphone and ask yourself “how many of these are better because they have a large network of people using them?” Me, I have 12 apps on my homescreen and eight of them – Tweetbot, Messages, Instapaper, Instagram, Phone, Mail, Safari, Spotify – are so much better because of the crowd of people that use them with me. When I switch off my phone’s network connection, they get so much worse. Those are the apps I care most about also, the ones that enrich my life, the ones that justify the expense of a smartphone.

When you switch off the network connection, most curriculum stays exactly the same. It doesn’t suffer at all, which means it isn’t taking advantage of the network connection when it’s on.

More Different

Digital devices should allow you to:

• Pose more interesting problems using more diverse media types and fewer words. (eg. three-act-style tasks).
• Replace your textbooks’ corny illustrations of mathematical contexts with illustrations from their own lives. Students: find a trapezoid from your own life. Take a photo. Tap upload. Now it’s in your textbook.
• Progressively disclose tasks over multiple screens so students don’t have to look at pages full of questions and information like this [pdf] and can instead start with a brief video and single sentence.

Networked devices should allow you to:

• See all your friends’ illustrations from their own lives. The teacher should be able to see that gallery of trapezoids, promote certain illustrations, and offer comments on others that are visible to everybody.
• Start lessons with integrated, formative polling. I’m talking about Riley Lark’s ActivePrompt software built right into the textbook.
• Create student conversations. Use student data to find students who disagree with each other, pair them up, and have them work out their differences. All of that should happen without the teacher having to facilitate it because the device is smart.
• Combine student data for better, more accurate modeling. (eg. Pennies, where each student collects a few data points which are then instantly collected into a much larger class data set.)

There are other possibilities, of course, some of which we’ll only start to realize as these tools are developed. But don’t just sit around and wait for an industry as reactive as textbook publishing to start making those tools for you. Publishers and their shareholders react to their market and that’s you. As long as they can still profit by repurposing existing print curriculum they will. It’s on you to tell your publishing reps that the curriculum they’re selling doesn’t do enough justice to the powerful, digital networked devices they’re putting them on. It isn’t different enough.

2013 Sep 27. And here’s LA Unified buying a billion dollars worth of iPads and then wasting the network that might make that investment worthwhile:

By Tuesday afternoon, L.A. Unified officials were weighing potential solutions. One would limit the tablets, when taken home, to curricular materials from the Pearson corporation, which are already installed. All other applications and Internet access would be turned off, according to a district “action plan.”

Featured Comment

This is always a problem in the early stages of a new technology. The “Technology Adoption Life Cycle” has proven itself over and over for the last 20 years to be the gold standard in analyzing tech markets.

The “innovators” adopt a technology because they need to be the first kids on their block to have whatever it is. The “early adopters” see strategic advantages and uses for it — and they are willing to put up with what they perceive as minor inconveniences like limited optimized uses in order to gain the advantages they seek.

That moment of “crossing the chasm” into the mainstream is that moment when a technology catches fire because vendors have figured out a way to reach beyond the techno-enthusiastic “early adopters” who have sustained their businesses to the techno-unimpressed “early majority” customers who are the major “show-me” skeptics. These skeptics form the first mass market for a technology, followed only later — and reluctantly — by a “late majority.”

Seems to me that we are still very much in an “early adopter” market in the race for digital textbooks. No one knows the “killer app” for digital curriculum is going to look like, but we do know it might bear some slight resemblance to the analog textbook. But this will not

As Steve Jobs always used to say, the “killer app” for the iPhone was making a phone call. But it was all the supporting infrastructure tht was built in (seamlessly integrated contacts, e-mail, texting, reminders, calendar, notes, & management of the technology) that transformed the act of making a phone call.

## [Future Text] Math Cache

a/k/a Great Moments in Digital Networked Math Curricula

You Should Check Out

Math Caching and Immediately Useful Teaching Data from Evan Weinberg.

What It Is

Evan has his students working on some practice exercises. As they complete their exercises, they use their Macbooks to submit a) an answer (which is nothing new in a world driven by quantitative machine-graded data) but also b) a photo of their work.

The images are titled with their answers and then start populating a folder on Evan’s computer.

Why It’s Important

Mistakes are valuable. Student work is valuable. This collects both quickly.

Mistakes are valuable for starting conversations, for prompting to students to construct and justify arguments, for asking students, “What different question does this work correctly answer?”

Most machine-graded systems hold back students with wrong answers and let them advance once they’ve corrected their errors. But this essentially sweeps clear the brambly trail that led to that correct answer when there’s so much value in the brambles. Those systems don’t tell you why the student had those incorrect answers. They don’t allow the teacher to sequence and select incorrect student work for productive discussions later. Math Cache does.

Here’s Evan:

I didn’t need to throw out the tragically predictable ‘who wants to share their work’ to a class of students that don’t tend to want to share for all sorts of valid reasons. I didn’t have to cold call a student to reluctantly show what he or she did for the problem. I had their work and could hand pick what I wanted to share with the class while maintaining their anonymity. We could quickly look at multiple students’ work and talk about the positive aspects of each one, while highlighting ways to make it even better.

Somewhat Related:

A main assumption that I work with when doing these [student] interviews is that children do what makes sense to them even if it seems like nonsense to me. My job is to figure out what makes sense to them and why.

2013 Oct 2. Pearson’s research blog picks up this post and argues that I’m too pessimistic about machine-graded data.

## [Future Text] Des-man

a/k/a Great Moments in Digital Networked Math Curricula

You Should Check Out

What It Is

First, you had Fawn Nguyen’s assignment where students created a face using the Desmos graphing calculator.

Students reviewed conics and domain and range. That was a blast for a lot of reasons. Now Desmos has created a system where the teacher can quickly see the creation of the faces in real-time and use filters to sort quickly through student work in productive ways.

Why It’s Important

Des-man explores the potential of networked devices in math class.

You could download the Desmos iOS app, flip off your iPad’s Internet connection, and still have a good time creating your Des-man. If the experience of using a digital math curriculum doesn’t get any better when you turn on the Internet, it is wasting the Internet.

With Des-man, an Internet connection lets you see all your friends’ Des-men instantly, as they’re being drawn. It lets the teacher see the Des-men quickly too and then select and sequence them in productive ways.

We have here a math activity for networked devices that doesn’t waste the network. That shouldn’t be noteworthy, but it is.

## [Makeover] Penny Circle

TLDR: Check out Penny Circle, a digital lesson I commissioned from Desmos based on material I had previously developed. Definitely check out the teacher dashboard, which I think is something special.

This is it, the last entry in our summer series of #MakeoverMonday. Thanks for pitching in, everybody.

What Desmos And I Did

Lower the literacy demand of the task. The authors rattle off hundreds of words to describe a visual modeling task.

Clarify the point of the task. A great way to lower the literacy demand is to convey the point of the task quickly, concisely, informally, and visually, and then formalize, expand, and verbalize that point as students make sense of it. Here, the point isn’t all that clear and the central question (“How can you fit a quadratic function to a set of data?”) is anything but informal.

Add intellectual need. The task poses modeling as its own end rather than a means to an end. Models are useful tools for lots of reasons. Their algebraic form sometimes tells us interesting things about what we’re modeling (like when we learn the average speed of a commercial aircraft in Air Travel by modeling timetable data). Models also let us predict data we can’t (or don’t want to) collect. We need to target one of those reasons.

The most concrete, intellectually needy question, the one we’re going to pin the entire task on, pops its head up 80% of the way down the page, in question #2, and even then it needs our help.

Lower the floor on the task. We’re going to delay a lot of these abstractions – tables, graphs, and formulas – until after students know the point of the task. We’re going to add intuition also and ask for some guesses.

Motivate the different abstractions. The task bounces the student from a table to a graph to a power function in five steps without a word at any point to describe why one abstraction is more useful than another. Students need to understand those differences. A table is great because it lets us forget about the physical pennies. A graph is great because it shows us the shape of the model. And the algebraic function is great because it lets us compute. If those advantages aren’t clear to students then they’re only moving between abstractions because grownups told them to.

Show the answer. We tell students that math models their world. We should prove it. The textbook does great work here, asking students in question #2 to “check your prediction by drawing a circle with a diameter of 6 inches and filling it with pennies.” Good move. But students have already drawn and filled circles with 1, 2, 3, 4, and 5-inch diameters. I’m guessing they would rather draw and fill a 6-inch circle than do all that math. The circles need to get huge to make the mathematics worth their while.

So here’s our new central question: how many pennies will fill a really big circle? We’re going to pose that question by showing someone filling a smaller circle, then cutting to the same person starting to fill a larger circle. It’ll be a video. It’ll take less than thirty seconds and zero words. It’ll look like this:

We’ll ask students to commit to a guess. We’ll ask for a number they know is too high, and too low, asking them early on to establish boundaries on a “reasonable” answer. The digital, networked platform here lets us quickly aggregate everybody’s guesses, pulling out the highs, lows, and the average.

Then we’ll talk about the process of modeling, of looking at little instances of a pattern to predict a larger instance. We’ll have them gather those little instances on their computers, drawing circles and filling them.

Those instances will be collected in a table which will then be aggregated across the entire class creating, a large, very useful set of data.

(Aside: of course a big question here is “should students be collecting that data live, on their desks, with real pennies?” Let’s not be simple about this. There are pros and cons and I think reasonable people can disagree. For my part, the pennies and circles are basically a two-dimensional experience anyway so we don’t lose a lot moving to a two-dimensional computer screen and we gain a much easier lesson implementation. However, if we were modeling the circumference of a balloon versus the breaths it took to blow it up, I wouldn’t want students pressing a “breathe” button in an online simulator. We’d lose a lot there.)

Next we’ll give students a chance to choose a model for the data, whereas the textbook task explicitly tells you to use a quadratic. (Selecting between linear, quadratic, and exponential models is work the CCSS specifically asks students to do.) So we’ll let students see that linears are kind of worthless. Sure a lot of students will choose a quadratic because we’re in the quadratics chapter, but something pretty fun happened when we piloted this task with a Bay Area math department: the entire department chose an exponential model.

Eric Berger, the CTO at Desmos, suspected that people decide between these models by asking themselves a series of yes-or-no questions. Are the data in a straight line? If yes, then choose a linear model. If not, do they curve up on one side of the graph (choose an exponential) or both sides of the graph (choose a quadratic)? That decision tree makes a lot of sense. But the domain here is only positive circle diameters so we don’t see the graph curve on both sides.

Interesting, right?

All this is to say, if you’re a little less helpful here, if you don’t gift-wrap answers like it’s math Christmas, students will show off some very interesting mathematical ideas for you to work with.

Once a student has selected a model, we’ll show her its implications. The exponential model will tell you the big circle holds millions of pennies. We’ll remind the student this is outside her own definition of “reasonable.” She can change or finish.