Defining Vertical Learning

Happy new year!

I have attempted to define vertical learning multiple times, without success, in the past. Each time, my thinking gets a little bit clearer. Today, I feel like I am almost there.

In this post, I will be summarizing my thinking and linking to additional posts that take individual thoughts and expand on them. I will be doing this thinking and writing over a period of a few weeks, so this summarizing post will be evolving. I will be updating it every few days. If you read something that strikes a chord with you, and you’d like to know more, please let me know in the comments. My goal is to define vertical learning to a general audience as clearly as possible.

Starter Motor Analogy

Most cars today have an electric motor that is used to start up the car’s engine. When you turn your key in the ignition, the starter motor draws current from the battery and starts to spin. The starter’s driveshaft is connected to the engine through a series of gears, so when the starter motor begins to spin, it turns the engine. Once the engine is running on its own, the starter motor disengages from the engine and turns off. In this system, we are using a smaller motor to get the larger engine going.

A Focus on Outcomes: Integrity, Alignment, and Taste

Vertical learning is a starter motor. When building a starter motor, outcomes matter. It doesn’t matter how good of an electric motor we build if it doesn’t successfully fire up the main internal combustion engine. In this case, the engine is an internal feedback loop that drives us, as individuals, to strive for integrity, alignment, and taste.

We strive for integrity by identifying our core values, discovering who we are and the things that matter to us. We strive for alignment by making sure that we are living according to those core values. This means examining our own behavior, but also examining things in our environment. Taste is still a little nebulous for me — I feel like it has many aspects — but I think it is linked to integrity and alignment. When something isn’t quite right, you know it and it bugs you and you have to do something about it. You have a set of internal standards that you are constantly evaluating things against.

The Origin Story

Vertical learning was born out of a very personal journey. I had no ambitions to change the world or put a dent in the universe. I felt somewhat alienated growing up. I couldn’t find anyone who thought in the ways that I thought, so it was a challenge for me to connect with people and talk to them. I was drawn to education because I wanted to:

1. Understand how my mind worked
2. Help people to learn and think better
3. Create a community where we could learn and think together

It may seem like a stretch to talk about building a starter motor that can fire up the engine of integrity, alignment, and taste in everyone. It seems crazy to imagine that we could provide learning experiences that would put everyone, including that incredibly close-minded relative who corners you at family gatherings and seems to lack any self-awareness, on a path to enlightenment. But I’ve experienced it for myself, through my own personal growth and the growth I observed in a cohort of students that I had the pleasure of working with for three years.

Building Up, Drilling Down

So how does the starter motor work? What is the technology behind it? The core idea is that learning goes vertical when we learn by building up and drilling down.

To be continued…

Going Vertical with Differential Steering

In Teaching a Robot How to Dance, an introductory computer science course for high school students, we ask students to program a robot to dance to unknown music and with an unknown robot dance partner. To do this, the robot has to be able to track its dance partner and move accordingly. One of the sub-tasks that we ask students to complete while developing this dance behavior is to first figure out how to program their robot to track and simply follow an infrared (IR) beacon.

The robot in Teaching a Robot How to Dance uses differential steering. It has two wheels, one on the left and the other on the right. To move forward, you program the two wheels to rotate at the same velocity. To turn to the right, you program the left wheel to rotate at a greater velocity than the right wheel. In this video, Daniel (the instructor) explains how the drive system on the robot works:

And in this video, Amalia and Ariana (the two high school students) figure out how to use the drive system to turn the robot to the right:

Altogether, it takes less than a minute for Amalia and Ariana to understand differential steering well enough to complete the task of programming a robot to track and follow an IR beacon. By the end of the session, they are comfortably adjusting wheel velocities as they improve their design.

When we first started working on Teaching a Robot How to Dance, we had a number of discussions about how the students would drive the robot. In most introductory robotics courses, students program the robot to “move forward” or “turn right,” where “turn right” is a pre-programmed subroutine that turns the robot in place 90˚ clockwise. When Daniel worked at iRobot, they developed an application programming interface (API) for driving the Roomba using subroutines that drove the Roomba in arcs.

An API is helpful because it is much easier to think in terms of angles and distances when driving a robot than left and right wheel velocities. But for Teaching a Robot How to Dance, we wanted to stress that the robot is essentially sensors and actuators, with behaviors linking those sensors and actuators. In this case, the wheels on the robot are the actuators, and we did not want to abstract that away.

Once we agreed that students would drive the robot by setting left and right wheel velocities, we started to discuss how to build differential steering into the curriculum. I did not want to spend a couple of days focused on differential steering, asking students to just drive their robots around. Since this was a project-based course, I wanted to dump the students into the deep end and have them start working on complex behaviors, such as follow-the-leader, right away. I felt that differential steering was something that students could learn just-in-time as needed.

Now, Amalia and Ariana only scratched the surface of differential steering. Their robot might be rolling forward at 200 mm/s when it suddenly needs to turn to the right. When that happens, the left wheel continues moving forward at 200 mm/s, but the right wheel is thrown abruptly into reverse, moving at -200 mm/s. The robot basically comes to a full stop in order to turn in place… and the transition between moving forward and turning can easily happen dozens of times a second. That might be acceptable for follow-the-leader behavior, but it doesn’t look very smooth for a dance partner.

At some point, students are going to want to program smoother and more complex movements into their dance steps, and they aren’t going to want to do it by setting wheel velocities. When that happens, the students will begin designing their own API for driving the robot, and the design of that API will be informed by the kind of movements they want their robot to do.

Imagine that we want our robot to have a dance step where it spins 360˚ in place. If that spin move happens in four beats, it needs to be programmed so that the speed of the robot adapts to the tempo of the music. If our robot’s dance partner moves away in the middle of the spin, our robot may want to drift slowly in the same direction while spinning instead of spinning in place. And if the dance partner gets too far away, our robot may want to break out of the spin after one, two, or three beats, and transition to a different dance move that will close the gap between it and its dance partner a little faster.

By the end of the course, the drive system of the robot would consist of multiple layers of software and hardware, and we wanted the students to design as many of those layers as possible. Sitting above the wheels, there would be an API for setting wheel velocities that we would provide. Sitting on top of that, there would be an API for driving the robot that would be designed by the student. This basic driving API would then be used by the student to program the robot with a repertoire of dance steps, and the design of the API would be informed by the kind of dance steps the student wanted to implement, and it would be redesigned as needed. Each dance step would share a common interface so that the robot could select and use dance steps interchangeably, and there would be a layer for choosing which dance steps to take. Ultimately, there could even be layers for learning which dance steps are better for a given piece of dance music or dance partner, or for tweaking parameters in the dance steps. Sitting below the wheel velocity API, there could be a layer for dealing with wheel slippage. If the wheels are set to rotate at 200 mm/s but they are slipping on the dance floor, the robot’s dance steps might be thrown off.

A key feature of Teaching a Robot How to Dance is that students are constructing and then climbing up and down their own abstraction ladders. You don’t want to have to think in terms of wheel velocities when designing dance steps, so you build an abstraction layer on top of that. You don’t want to have to think about custom interfaces for each dance step, so you modularize them and add another abstraction layer. The more well-designed the abstraction layer, the easier it is to build on top of it and make your own life easier. Are you running into some unexpected behavior in an edge case? You may want to drill down to a lower abstraction layer to see what is really happening. This design encourages students to apply and appreciate computational thinking.

Differential steering represents a tiny fraction of the content in Teaching a Robot How to Dance, but we had to make dozens of design decisions on how to implement it in the course. The decisions we made are highly case-specific; we would have designed things differently if we were implementing differential steering in a different context. But constructing and then climbing abstraction ladders is something that I am passionate about, and it is part of what I mean when I talk about vertical learning. This is something that all kids (and adults) can and should be doing.

Teaching a Robot How to Dance: A Case Study

In 2011, I began working with a small team to develop an introductory computer science course for high school students with little prior experience or interest in technology. Daniel and Tim were experts in robotics and software engineering who had come together to design a marine robotics course at M.I.T., and Lee was a professor teaching computer science and doing research in computational biology.

When I joined the group, I suggested that we engage in a backward design process and start by articulating the objectives of the course. While we shared a common desire to encourage more students to pursue careers in STEM, we were all coming at the problem from different directions and operating with different assumptions. To get on the same page as a group, we had to clarify our own thinking as individuals.

We agreed that the focus of the course should be on computational thinking. We wanted students to develop a conceptual foundation and habits of mind that would enable them to apply advanced computation in any field. Daniel and Tim wanted the course to involve robotics, but they didn’t want to focus on mechanical design and programming. They wanted students to experience robotics as a rich and engaging problem space driven by the interaction of hardware, software, and the environment. Finally, it was important for students to drill down to some form of ground truth when engaging in the engineering design process; we wanted them to have to consider and eventually learn about underlying mechanisms: what happens when the wheels in the differential drive system slip, the charge in the battery gets low, or you are limited by the sampling frequency of the sensor data and behavior arbitration system?

Designing a task aligned with our objectives took discipline. We needed a project where students would have to meet our objectives in order to complete the task. It would have been simpler and faster to design a fun and engaging task, and then layer our objectives on top of it, but it also would have been far less effective. After a great deal of thought and brainstorming, we came up with the idea of teaching a robot how to dance. Teaching a robot how to dance isn’t the same as programming a robot to follow a pre-scripted choreography to music. Students needed to figure out how to teach a robot to dance to unknown music and with an unknown robot dance partner, which means detecting the tempo of the music and the position and movement of the robot dance partner, and reacting appropriately. It is the kind of rich, complex problem that would engage and challenge an expert roboticist.

Once the task was defined, I began designing the course, breaking the task into sub-tasks and creating sequences of lessons to develop skills and concepts. I introduced the group to Moodle, a learning management system (LMS). We planned to deliver the course using face-to-face instruction, but we wanted to post assignments and materials online. I also often use threaded discussions to increase accountability, enable students who like to process their thoughts longer or need more wait time to participate more actively, and make it easier to follow up and build on earlier points in the discussion.

The group was impressed with how I developed skills and concepts in a logical and cohesive progression, but they had concerns that students would be unable to progress at the rate that I anticipated. Even though we had agreed that the course would be project-based with a constructivist approach, there was some desire to pull back and spend more time constructing foundational skills before getting to the heart of the project. We decided to test components of the course with small groups of high school students to see how they responded.

We challenged students to program a robot to track and follow an infrared (IR) beacon, which is one of the critical sub-tasks in Teaching a Robot How to Dance. We recorded the second session, and I documented it through a series of videos on the Computing Explorations website. (I also posted a short introductory video on YouTube.) Initially, the two students in the session, Amalia and Ariana, spend more time trying to get through questions than answering them. But later, they begin reasoning things out and using the tools and strategies modeled by Daniel, their guide and instructor. By the end, as their capabilities and understanding have expanded, they are visibly more confident and engaged in the task, and they are even questioning and answering each other.

I enjoyed the development of Teaching a Robot How to Dance on multiple levels:

• I had the opportunity to immerse myself in a new discipline and to view the world through the eyes of a roboticist.
• Starting with a blank slate, I was able to take risks and push myself to design innovative curriculum based on constructivist learning principles.
• We took the time to design a course that was cohesive and fully aligned with our objectives.
• I learned about new technologies and was able to share what I knew about curriculum design and instruction.
• By collaborating, we designed a course that none of us could have designed on our own, and we developed common understandings that will impact our thinking moving forward. We also built a level of trust that enabled us to uncover our own assumptions, take risks, and challenge core beliefs.

I believe that this is the kind of process that all educators should be going through in order to develop the curriculum and instruction our students need and deserve.

The Joys and Tribulations of Thinking Different

The Land of Zero Convergence: The Keystone XL Pipeline

I’ve been curious about the Keystone XL pipeline recently. Every once in a while, something is in the news and I realize that I don’t know as much about it as I should. This bugs me and I want to learn more.

However, once I started digging deeper, I realized that the Keystone XL pipeline lives in two worlds: a pro-pipeline world and an anti-pipeline world. Depending on your political leanings, one of those is the real world and the other is some bizarro world where logic has no meaning. Both worlds have their own set of facts, experts, and published reports backing them up, and there is little to no overlap.

Whenever there is a program on the Keystone XL pipeline with a panel of experts, the same arguments and disputed facts are trotted out again and again. Even if a point is eventually conceded by the panel after exhaustively examining and evaluating all of the evidence, that same point is magically un-conceded in future debates and we have to start over again from the beginning.

If you go to the Wikipedia article on the Keystone XL pipeline, it dispassionately documents both worlds without trying to find any ground truth. It lays out the events that have occurred and the statements made by both sides, while pointing out how every independent, third-party report has been “debunked” by one side or the other because the report’s authors have received funding from one group or expressed support for another group in the past. It tells you what has happened, but it doesn’t help you figure out what will happen if the pipeline is or isn’t built.

Toward a New Normal: Reframing the Discussion

Trying to formulate an opinion on the Keystone XL pipeline is next to impossible without a body of facts that represent the ground truth. You can embrace one of the existing ground truths (pro or anti), spend months sifting through primary sources to curate your own ground truth, or sit on the sidelines. I suspect that most sensible people choose the latter, and our political system is the worse for it.

A few years ago, I began to think about systems for driving discussions forward. I realized that there had to be mechanisms in place to reach consensus. I don’t seek consensus just to get along; I do it because we will never get to the heart of an issue if we can’t even agree on some basic facts. And agreeing to some basic facts may just give us the skills to create a more substantial meeting of the minds.

We start by identifying what we know and don’t know. How many jobs will be created during the construction phase of the pipeline? How many once the pipeline is operational? What would happen if there is an oil spill in the Ogallala Aquifer? What percentage of the oil passing through the pipeline will be used domestically versus sold overseas? How will the pipeline affect oil production and what will this do to oil prices? Where would the oil go if the pipeline isn’t built and where would we get our energy instead?

We would publicly debate each of these questions and hammer out an answer. Comments that are insightful, informative, clarifying, synthesize multiple comments, or pose thought-provoking questions would bubble up to the surface. When consensus has been reached on a specific point, debate on that point is locked and the discussion continues with that point established as a given. If someone wishes to re-open debate on a point that has been locked, that can happen on a side thread, but not in the main discussion. Over time, consensus is developed around a set of key points and the discussion moves into new territory. It’s fairly obvious that the debate over the Keystone XL pipeline is a proxy for more serious debates… debates that we aren’t equipped for right now. Is it sensible policy to keep oil prices high to encourage the development of renewable energy sources? What is the right balance between short-term economic growth and long-term environmental sustainability?

Who decides which comments bubble up to the surface, which questions are debated, and when consensus has been reached? The community does, but your level of input is based on karma/reputation points that you acquire for the quality of your participation. Ask thoughtful questions, help others synthesize opposing viewpoints into a coherent viewpoint, uncover your own assumptions, provide expertise, help someone express themselves more clearly by reflecting back what they are saying… and those comments will be highlighted and receive more discussion, and you will gain reputation points that enable you to have greater input in the community. Be rude to others and continue arguing points that are considered settled… and you can keep posting those comments but they will be buried for few to read.

Why This Is Important: Obamacare

This level of discussion is labor-intensive. Without constant curation from highly skilled facilitators, the feedback loop breaks down and the discussion stalls. In order for the model to scale, discussions need to advance far enough to generate more facilitators. It’s similar to compressing mass-energy to ignite a sustained nuclear fusion reaction. Is the effort worth it?

At some point, the debate on the Keystone XL pipeline will end when one side “wins,” but that kind of resolution comes at a cost:

1. Resolving an issue only after it has grown into a crisis is costly and often results in sub-optimal solutions.
2. Letting a winner emerge instead of hammering out a compromise that both sides can live with can also result in sub-optimal solutions.
3. We never debate the underlying issues.
4. Ideas that don’t conform to the two-sided debate are never considered.

When President Obama laid out his initial proposal for healthcare reform, he did something interesting. He framed it outside of the existing two-sided debate. Instead of focusing on universal healthcare and providing healthcare to the uninsured, he focused on the rising cost of healthcare, which is rising much faster than inflation. The cost of healthcare is a huge drag on the economy. Individuals and businesses have to spend a greater fraction of their income on healthcare. In that context, most individuals and business leaders would agree that controlling the rising cost of healthcare is a top priority, and one that virtually everyone agrees on.

Rising healthcare costs is a much bigger problem than social security. Most business leaders would happily raise the minimum wage if it meant healthcare costs could be reined in. But we aren’t discussing healthcare costs at all! Why is that?

The Democrats and the Republicans are so far apart and their best ideas (single-payer healthcare and health savings accounts, respectively) are so unpalatable to most Americans, that any real discussion is impossible at the moment. President Obama hoped to changed that with a radical new approach.

In certain regions of the country, Medicare was able to drive down costs and improve the quality of healthcare (as measured by patient outcomes) at the same time. Better care for less. There were small scale experiments with preventive care and how doctors were billed. President Obama wanted to unleash this creativity on a wider scale. Think of charter schools, but where charter schools are intended to compete with public schools, “charter” health insurance companies would compete with private health insurance companies. But who would sign up for a new and unproven kind of health insurance? Health insurance companies need large subscriber bases to negotiate costs down. The solution was to subsidize and provide this health insurance to nonconsumers: the uninsured.

Unfortunately, to Republicans, the creation and funding of these charter health insurance companies was simply a ploy by Big Government to expand Medicare and get into the health insurance business. They couldn’t see it as an attempt to create competition for health insurance companies and use market forces to hold down costs. When this happened, President Obama chose to retreat and that portion of the bill was gutted. The only thing that was left was subsidizing healthcare for the uninsured, which fit neatly into the existing narrative of Democrats trying to expand entitlement programs. The most interesting part of the bill, using competition to hold down healthcare costs, was never discussed. And frankly, I suspect that President Obama was right to back down: we lack the tools to think out-of-the-box and have any kind of real debate as a community.

Thinking Different: The Joys and Tribulations

I don’t know if President Obama’s approached would have worked. I’m skeptical, but I would have liked to have discussed it and for our collective brains to study it. That didn’t happen and isn’t happening in many other areas. President Obama thought differently, but he couldn’t get enough people to do the same.

I’ve thought differently my whole life. On one level, it is very cool and kind of fun to see solutions where everyone else sees a lost cause. But it can be disheartening and lonely, too. I’ve chosen education as my battleground. I see potential in us and I genuinely believe that we can have the kinds of discussion that I dream about… and that those discussions can make us understand others better and think more deeply. More importantly, I have tested those ideas and I have seen them work on small scales. Would I love to test these ideas on a larger scale? Yes! Would I love to have a circle of friends to hash out new ideas, people who would push and challenge me to understand and think deeper myself? Hell, yes! I guess this is me putting the call out… if this resonates with you, if you yearn for the things that I yearn for, drop a line and say hello. It’s awfully nice to meet you.

“There are those that look at things the way they are, and ask why? I dream of things that never were, and ask why not?”
—Robert F. Kennedy

Moving the World

A few weeks ago, I thought about writing an article on Apple Pay and posting it on one of the tech-site forums that I frequent. People were using the forums to discuss the merits of Apple Pay, but these discussions were filled with both missing information and actual misinformation. This made it difficult to draw any conclusions, and arguments ended up going in circles.

I hate when arguments go in circles and nothing ever gets settled. It happens in tech-site forums. It happens in education. It’s an incredible waste of energy that can stall forward progress. My plan was to write down what I thought I knew and then crowd-source anything that I had wrong or didn’t know. We would debate issues in the comments until we reached a level of consensus, and then I would update the original post with the new information.

When I shared this plan with my friend Daniel, he was less than enthusiastic. He asked me why I didn’t just research Apple Pay on my own and then post what I had learned. Why did I want to crowd-source the article?

This got me thinking. I wasn’t motivated by a desire to write a comprehensive and meticulously-researched article. I enjoy reading those articles, but not enough to research and write them on my own. I also wasn’t motivated by a desire to organize and harness a community effort to write a comprehensive and meticulously-researched article. I like the idea of sharing resources to build something together that would be difficult for any one person to build alone, but that level of sharing and community-building isn’t enough for me. No, I was motivated by a desire to transform individuals through the process of organizing and harnessing a community effort to write a comprehensive and meticulously-researched article.

Tech sites compete to improve along a specific trajectory. Their goal is to attract and retain eyeballs so that they can sell those eyeballs to advertisers. Some sites compete by writing well-researched articles. They attract and retain eyeballs by being authoritative and highly credible. Other sites compete by trying to be your one stop shop for all your tech news or the site with a community of like-minded readers. But either way, they measure success with the same metrics: How many readers do you have? How engaged/loyal are they?

Many tech sites would like to improve the quality of their comments and forums, but they can’t afford to do it unless it also attracts and retains eyeballs. Unfortunately, circular and never-ending arguments are a great way to fire people up and get them coming back for more. This is why many sites end up adopting clickbait headlines and goosing their readers to argue back and forth. It’s reached the point where many readers and tech-site editors just assume that any attempt to clean up comments and forums is a hopeless battle.

Why do people post comments on articles and in forums on tech sites? I think most people do it to connect with other people and win points for being clever. Being clever can mean being informative, but it can also mean the perfect put-down. Forums are fairly well-designed for these jobs, but some people would love to hire forums for other jobs: learning things, having thoughtful discussions, or pontificating. Since most tech-site forums aren’t well-designed for those jobs, those people end up as nonconsumers.

As a crowd-sourcing site, Wikipedia competes on a different set of metrics. It looks at the number of people who use it as a source of information and the number of active editors and contributors. I’ve read that Wikipedia has been struggling to attract and retain contributors. There have been complaints that a small group of longtime editors have been driving new contributors away. Unfortunately, that same small group of editors is the backbone of the site, so Wikipedia faces a serious dilemma. Do they try to make editing more inclusive and bring in fresh blood (but risk killing the site by losing their most active editors) or do they stay on course and hope that small tweaks will turn things around (but risk killing the site as the pool of contributors continues to slowly shrink over time)?

If I ever created a site to crowd-source articles, I would compete on an entirely different set of metrics: I’d measure the growth of my readers.

On a recent Apple Pay article, two readers were arguing back and forth. They exchanged over two dozen comments in an hour. A few other people tried to jump in, but they weren’t able to help. The first reader was wrong; he was basing his entire argument on a single misconception. Since you linked your credit card to Apple Pay inside of an Apple app, he assumed that Apple generated the token that is stored on your smartphone. What he didn’t realize is that Apple’s app establishes a direct connection to your specific bank, and the bank that issued your credit card generates the token, not Apple. It was an honest, and perfectly reasonable, mistake. The second reader was correct, but he never identified why the first reader was mistaken. Instead, he kept repeating his argument over and over again. He actually did explain that Apple did not generate the token, but there was so much cruft around that statement that the first reader couldn’t hear it. If the two of them could have identified the source of the token as the source of their disagreement, then they could have either resolved their disagreement then and there or at least parted company agreeing to research that one specific issue further. Instead, both of them parted company pissed off and convinced that they were right and the other was wrong.

Now, imagine if someone had jumped in and helped the two men identify the source of their disagreement. With a little research, they would have found a credible source explaining that the bank, and not Apple, issues the token. With consensus achieved, an editor would then add that information to the article so that other readers would not have to debate the same point. If this happens enough times, the first reader will learn not to be so sure of himself when he doesn’t actually know something, and that he can learn more by asking questions instead of arguing. In fact, if he asks enough questions, soon he’ll be the one helping the new guy out. Meanwhile, the second reader will learn that he can contribute more if he listens to people who are mistaken and can identify the source of their confusion. Instead of endless arguments, the forum provides an opportunity for growth.

My crowd-sourcing site is fundamentally different than Wikipedia because we compete on different metrics. At Wikipedia, people contribute to articles. When there is a disagreement, an editor comes in and makes a final decision. There is no attempt by the editor to help the contributors resolve their own conflicts. At my site, an editor is a mediator. They model how to resolve conflicts so that contributors can resolve more of their own conflicts, and many of those contributors will eventually begin mediating for others and become editors themselves. The debate has value; it’s not just a necessary step on the path to a well-sourced article.

Is there a business model for the kind of site I’m describing? I have no idea. I do know that there are nonconsumers who would love to hire the kind of website forum I’m describing. It’s also potentially disruptive because it changes the metric for performance. I also know that it is possible. I can envision it, even if I don’t know how large the market is. Imagine a site where people learned to resolve their differences, and then applied what they learned to other sites, to their face-to-face relationships, and then in the public arena. Instead of discussions about Apple Pay, we could grow to discuss global warming, universal healthcare, immigration, and sensible tax policy. Just imagine it. That’s the world in which I’d want to live.

Removing Our Blinders

When reading Disrupting Class, I detected four anomalies that might cause Christensen’s prediction that schooling will be disrupted by student-centric technologies in the form of computer-based learning to fail. The first anomaly is that the technology of public schooling is immature. Since schooling in this country is over two hundred years old and it has been steadily improving, that should not be the case. The second is that educational research is descriptive, but not predictive. Once researchers have described schools, they should automatically move to the next stage, which is to test and improve their models through predictions. That is not happening. Third, schooling may be improving, but instruction is not. According to Christensen, personal tutoring represents the state-of-the-art in student-centric instruction, but personal tutors today are no more effective than personal tutors twenty years ago. And to raise test scores and meet targets established by No Child Left Behind, schools are changing everything but instruction. Fourth, instead of studying outliers who may be doing something different to get extraordinary results, we assume they have a ‘secret sauce’ that cannot be codified and will not scale.

Just because these anomalies exist, it doesn’t mean that disruption won’t occur; it just means that disruption may have to proceed a little differently. It may be possible for schooling to continue improving without ever improving instruction. It may be that we are all playing a giant game of chicken, and once all other avenues have been exhausted and improving instruction is the only way forward, then we will improve instruction. But I’m skeptical. Personally, I believe that, to disrupt schooling, we need a disruptive innovation in the technology of instruction, and we won’t develop that innovation until we have identified and addressed the root cause that is causing these anomalies.

In earlier posts, I have postulated reasons why instruction may not be improving (core beliefs about what we can learn and getting stuck at local maxima). I don’t know if those are root causes or only additional symptoms of a deeper root cause, but for some reason, it is common in education to see what we believe to be true instead of what is happening right in front of us.

For example, many educators believe that the key to improving schooling is to motivate students by appealing to their interests. Christensen writes, “When there is high extrinsic motivation for someone to learn something, schools’ jobs are easier. They do not have to teach material in an intrinsically motivating way because simply offering the material is enough. Students will choose to master it because of the extrinsic pressure. When there is no extrinsic motivation, however, things become trickier. Schools need to create intrinsically engaging methods for learning.” Motivation is essential to learning. There is no question about that. But is motivation on its own enough? Sometimes I think that people fixate on motivation because it allows them to not think about the other components in learning.

Christensen writes: “We believe that a core reason why so many students languish unmotivated in school or don’t come to class at all is that education isn’t a job that they are trying to do. Education is something they might choose to hire to do the job—but it isn’t the job. While we continue our research to understand this crucial issue, we hypothesize that there are two core jobs that most students try to do every day: They want to feel successful and make progress, and they want to have fun with friends.” “Furthermore, when we use the phrase ‘want to feel successful,’ we do not mean the kind of surface-level idea of success that constitutes praising a child no matter how she performed on a given activity under the mistaken idea that building ‘self-esteem’ in this vein is a good idea. Instead we mean true success, where the student in fact accomplishes and achieves something real and makes progress.” If you listen to and work closely with children, you will know that this is exactly right.

Imagine that you have a medical issue, but your doctor doesn’t listen to you and appears to be utterly incompetent. To get what you need, you end up scouring the internet for information, performing your own diagnosis, and essentially telling the doctor which treatment you’d like to be prescribed. This may be better than nothing, but if you were paying for these doctor visits out of pocket, you’d be very angry and very unhappy. You hire the doctor to guide you as a medical expert, not to function as an expensive prescription pad and referrer. Students don’t hire schools to purse their own interests and engage in pastimes. Schooling is way too expensive for that. We can do that in our own time and there are other activities that can do that job much better than schooling can. We hire schools to help us be successful and to prepare us for the future, and we expect schools to have some expertise in those areas and to give us expert guidance. Students may settle for pursuing interests in school because it is better than nothing, but in the long term, it is de-motivating. “All students are likely to be equally motivated to feel successful. For some, school is a viable candidate to hire for this job. This group likely includes those whose parents provide a clear link between academic achievement and career success; those whose intellectual capacities were honed through repeated, sophisticated verbal interaction with adults before the age of three; and those whose way of learning or passion matches that of their particular teachers. The students who do not hire school to feel successful are not unmotivated to feel successful. They just don’t or can’t feel successful at school—often it makes them feel like failures.”

The evidence that students want to be successful and not only pursue their interests is right in front of us, but too many people ignore it. Christensen sees it, but then he has his own blinders on. Christensen starts by building the case that “schools need to create intrinsically engaging methods for learning,” but then he leaps to the conclusion that student-centric technologies will make that happen. “When prosperity has removed this source of [extrinsic] motivation, the solution must be to make learning intrinsically motivating. Student-centric learning will play a key role in addressing this challenge. If children are motivated to learn, and if we enable each one to learn effectively, we will have an education system with a great performance record.” How does student-centric instruction—teaching students according to how their brains are wired—cause an activity to be intrinsically motivating? It is certainly plausible that, if an activity is more accessible to our way of thinking, it will be more interesting and engaging to us and we will be more successful at it. Instead of proposing a mechanism that only increases a student’s motivation (appealing to student interests), Christensen is proposing a mechanism that increases a student’s motivation and makes it easier for that student to learn. That is certainly a step forward, but it is still only a theory without any supporting evidence. Student-centric technologies may enable more students to learn, but will they enable every student to learn?

Is there an activity that enables every child to learn and be successful? Christensen believes that there is. “When a parent engages in extra talk—speaking 48 million words to an infant in its first 36 months of life—many, many more of the synaptic pathways in the child’s brain are exercised and refined. This makes subsequent patterns of thought easier, faster, and more automatic. This means that children who have been lavished with extra talk have an almost incalculable cognitive advantage compared to those who have not been. Their brains have been ‘wired’ to think in much more sophisticated ways than those of children whose synaptic pathways have not been extensively developed and lubricated through use.” And “when children whose cognitive capacities have been expanded as described above confront and succeed at the initial academic challenges they encounter in school, their sense of self-efficacy—their excitement and confidence in their ability to succeed at difficult intellectual tasks—can blossom.” The correlation between extra talk and cognitive development is strong enough to suggest that children in every demographic can be successful in school if they are exposed to enough extra talk at an early age.

What else does the evidence suggest? That interest and student-centric models are not the primary drivers for learning and success. Extra talk had the most impact in the first twelve months of a child’s life, before the child can verbally respond to the extra talk. This indicates that the content of the extra talk was irrelevant. Extra talk appears to be successful because the child and the parent are both highly engaged by it. Children strive to connect with their caretakers as an evolutionary survival mechanism. That connection makes it more likely that a caretaker will care for and protect the child. A second crucial factor in extra talk is its level of sophistication. “Hart and Risley observed two sorts of conversations occurring between parents and their infants in their study. Parents they described as ‘taciturn’ often limited their conversations with their children to ‘business.’ Business conversations with infants are not rich or complex; they are simple, direct, here-and-now conversations. The words that truly matter are spoken in a posture that Hart and Risley term ‘language dancing,’ where the parents engaged face to face with the infant and speak in a fully adult, sophisticated, chatty language—as if the infant were listening, comprehending, and fully responding to the comments. It is deliberate, uncompromised, personal adult conversation.” This suggests that learning is most effective when it fulfills the job that it is hired to do and when it is rich enough to establish new sophisticated pathways in the brain, leading to increases in intellectual capacity, self-efficacy, and curiosity.

It almost doesn’t matter what we do to improve schooling if we don’t give children tasks that cause them to create new pathways and think more sophisticatedly. Right now, schools aren’t doing that. Some project-based schools are trying, but if you look at their curriculum, you will see that tasks are sophisticated for children, but not sophisticated for adults. That won’t work either. Vertical learning is the approach that I’ve been developing for the past two decades to enable students to work on increasingly sophisticated tasks and feel successful. I describe this type of learning as vertical because thinking, performance, and achievement build and accelerate over time. If you’d like a hint of what this kind of learning looks like, take a look at a 90-minute learning session we did at Computing Explorations. If we are going to make meaningful improvements in schooling and instruction, then we need to take off our blinders and focus on what really works instead of fixating on the things we believe should work.

Gazing Into the Crystal Ball

Christensen predicts that, through the development of student-centric technologies, computer-based learning will soon disrupt public schooling in the United States. So far, “computers have not increased student-centric learning and project-based teaching practices. The implementation of computers has not caused any measurable improvements in achievement scores. And, most important for the purposes of this book, computers have made almost no dent in the most important challenge that they have the potential to crack: allowing students to learn in ways that correspond with how their brains are wired to learn, thereby migrating to a student-centric learning environment.” “But as is the case with all successful disruptions, if you know where to look—competing against nonconsumption—computer-based learning is methodically gaining ground as students, educators, and families find it better than the alternatives—having nothing at all.” “The data suggest that by 2019, about 50 percent of high school courses will be delivered online. In other words, within a few years, after a long period of incubation, the world is likely to begin flipping rapidly to student-centric online technology.”

Why is Christensen betting on computer-based learning when its track record has been so poor? “First, online learning will keep improving, as all successful disruptions do. It will become more enjoyable and take full advantage of the online medium by layering in enhanced video, audio, and interactive elements. Currently, according to reports, online learning works best with more motivated students; over time, it will become more engaging so as to reach different types of learners. A second driver of this transition will be the ability for students, teachers, and parents to select a learning pathway through each body of material that fits the learners’ needs—the transition from computer-based to student-centric technology. The third factor that will likely fuel the substitution is a looming teacher shortage. The fourth factor is that costs will fall as the market scales up.” Christensen knows that current models of computer-based learning are not very good and “largely mirror the dominant type of learning method in each subject.” But in the industries he has studied, products inexorably improve.

Since Christensen made his prediction in 2008, we have seen the growth of massive open online courses (MOOCs). In the case of MOOCs, “online technology provides accessibility for those who previously would not have been able to take the course. It provides convenience for a student to fit the course into his or her schedule at the time and place that is most desirable. To varying degrees, it is simpler because it offers comparatively greater flexibility in the pace and learning path. And when it is software-based and online, it can scale with ease. Economically, it is often less expensive than the current model, even at today’s limited scale.” But the completion rate of most MOOCs is less than 10%. Most students who sign up for a MOOC never finish it. Christensen believes that “layering in enhanced video, audio, and interactive elements” will make online learning “more engaging so as to reach different types of learners,” but there is no reason to believe that multimedia is going to make these courses intrinsically motivating to most students. Students don’t hire schooling to be entertained; they hire schooling to “feel successful and make progress.” The students who are most successful in MOOCs are those who are already most motivated to learn. To be disruptive, computer-based learning cannot just improve average performance and be more cost-effective, it has to help those who are least motivated to study core academic subjects be successful at them.

Christensen is essentially asking us to do something in software that we don’t know how to do in person. If we assigned every child a personal tutor, and every tutor was an expert in instruction and had access to the best instructional resources in the world, many children would continue to struggle and feel unsuccessful. Average performance would go up and the United States would be more competitive, but we would not be any closer to the goal of every child in every demographic becoming proficient in all core academic subjects. This is not something that we can do with the instruction we have today.

Unless there is a disruptive innovation in the technology of instruction, computer-based learning will be a sustaining innovation that improves schooling, but does not disrupt it. There is actually little or no resistance to computer-based learning in schools, and once it becomes good enough, schools are likely to hire it for four different jobs. First, many courses will include some form of blended learning, where teachers integrate online coursework into an existing course. This is how schools will plug computer-based learning into their interdependent systems. Second, schools will offer online courses in place of electives and other specialized or advanced courses that they can’t afford to offer themselves. Third, schools will convert some existing courses into online courses for their most motivated students. By hiring computer-based learning for these two jobs, schools will be able to channel more resources to core academic subjects and their neediest students. Fourth, schools will use computer-based learning to provide additional instruction to struggling students. Note that, in my prediction, schools are not using computer-based learning in place of teacher-led instruction for struggling students, but to increase the total amount of instruction these students are receiving.

Schools have responded to No Child Left Behind not by improving instruction, but by giving students more of it. If math class used to be 50 minutes a day, it is now 70 minutes a day. On top of that, struggling students go to a “math lab” or “intervention block” twice a week where they work with a math teacher in small groups. The use of small groups is not to provide customized instruction, but to increase extrinsic motivation through heightened accountability. This is basically the same model we have been using for special education for decades. Since there have been no innovations in instruction for two hundred years, schools have had to find other areas in which to innovate. If computer-based learning is simply a platform for delivering the instruction we already have, then it will be crammed into the same role.