What MIT Should Have Done

Ten thousand students have just taken the final exam in MITx’s course “6.002x Circuits and Electronics.” The sheer size of this course (120,000 first registered back in March), the high-wattage backing of MIT for a certificate of completion to all those who make it through to the end, and the free, open access nature of this MOOC (a “massive open online course”) seemingly ushers in a fundamentally new paradigm in higher education. When coupled with the recent headlines about similar ventures such as Coursera, Udacity, and MIT’s own new partnership with Harvard to form edX, the policy world has been positively aglow: David Brooks calls it a “campus tsunami”; John Chubb proclaims it an “historic transformation”; Thomas Friedman writes, simply, “welcome to the college education revolution.”

In one respect, such adulation is completely understandable. MITx has given anyone, anywhere, the chance to learn from a world-renowned professor at one of the top universities in the world and receive a certificate of achievement for so doing. And it’s just a matter of time before edX (or someone else) solves the logistics of user authentication to turn this certificate into transferable credit for a nominal fee. And as edX and others expand their computer-driven and fully-automated course offerings this fall into the humanities and social sciences (Coursera already has more than 9,000 students enrolled in its fall poetry course), the notion of place-based education seems no longer secure. As Kevin Carey has written, the “monopoly has begun to crumble. New organizations are being created to offer new kinds of degrees, in a manner and at a price that could completely disrupt the enduring college business model.”

Indeed. If I can take the vast majority of my general education requirements from top-notch professors at famous universities in the comfort of my bedroom, why should I ever again have to pay $400 per credit to get dressed, drive 20 miles to my regional college, search for a parking space, and attend a lecture in a massive auditorium with 300 other students listening to an adjunct I don’t know go through a PowerPoint lecture I could have scrolled through at 1.25X speed in the MITx course? In one fell swoop, every introductory college course in the country has been put on notice.

This can be seen as an obvious outcome of a decades-long flowering of the Internet: The Open Courseware movement has been thriving for more than 10 years, with MIT’s own platform getting a million visitors a month; Khan Academy’s extremely popular videos have helped spawn an entirely new lexicon of the “flipped classroom” (see, for example, how the TED talks have recently jumped on board); and the for-profits (such as Straighterline and New Charter University) have used the efficiency and scale of the Web to break the higher education cost barrier such that a semester’s worth of classes is by now cheaper than your monthly cell phone plan.

George Siemens has glossed such developments as the potential of the “four Vs” of the Web to transform education: The sheer amount of data out there (volume), offered in so many different modes of delivery (variety), available anytime and anywhere (velocity), and at different levels of data depth, accessible differentially from novice learners to expert researchers (variability). This has turned online education from a sideline industry just five years ago into an assumed and expected part of how 30 percent of all postsecondary students learn. Online courses are now front and center in the “disruption” of higher education. As MIT’s provost (and just-named next President) summarized in a letter to the faculty, MITx has the potential to “dramatically improve the productivity of education and the access to quality education worldwide, and will transform the nature of our residential learning environment.” Amen.

But there is a problem. A fundamental problem. MITx, and all such similar initiatives, are still delivering a Learning 1.0 product in a Web 2.0 world. They have replicated all of the problems of the traditional industrial-age model of lecture-based teaching and testing that has minimal linkage to student outcomes. The eight percent retention rate from the course’s start to now makes it clear that it is the course, and not the student, at the heart of their conceptual paradigm. This is why the blogosphere is wrong. There will be no online MOOC-driven revolution so long as MITx’s discussion board continues to be littered with stressed-out students worried about making deadlines and solving too-difficult equations, students complaining about the repetitiveness of the lectures or celebrating their midterm scores. So long as such courses continue to be teaching-focused rather than learning-centered, the only transformation will be that students online will fall asleep from boredom much faster than those sitting in the cramped lecture-hall seats.

MIT could have done so much more. They should have done so much more. In fact, I want to suggest that there is indeed a real revolution in the making, but it has little to do with the size or scope of such MOOCs. Rather, what MITx has stumbled into is the opportunity to create a never-tiring, self-regulating, self-improving system that supports learning through formative on-demand feedback. Formative “just in time” feedback (rather than summative “end of course” testing) is the holy grail for learning theorists because it turns unidirectional teaching concerned mainly with delivering knowledge into a recursive guide and springboard for learning. If MIT had done that, they would have changed just about everything about how we think about higher education. But let’s take it a step at a time.

On a deeply practical level (and all of these examples are currently available and demonstrably successful), what MITx should have done is created an adaptive testing model in which students are presented with a question about a lecture topic at an appropriate level of difficulty based on their previous correct or incorrect answers to previous segments. Thus as students demonstrate mastery at specific levels, they can move forward; else they are sent to subprograms to test for and support key background skills and knowledge. Such subprograms, in turn, could make use of a wider set of resources that students work through until they are able to rejoin the course at the point at which they were first stumped. These subprograms could themselves be linked to a formidable array of computer-based intelligent tutoring systems (ITS) that have been shown to be as effective as human tutors by mimicking the “interaction granularity” of real-life tutors as they “walk students through” a problem and its solution. So if a student got stuck at any point in the course, she could, for example, click on a “hint” button and the ITS would scaffold the student—through prompts, demonstrations, natural language feedback, etc.—toward both the right answer and the logic and reasoning behind it.

Moreover, MIT could have taken all of its data—such as students’ demographics characteristics of gender, ethnicity, English-language skills, prior background knowledge, learning style, number of times they click “hint,” the point at which they actually click ‘hint,” their success rate in each section, the speed at which they submit their answers, which tutorials seemed to help which students best in which sections, etc.—and fed it into some powerful algorithms to create seemingly personalized feedback. Such a “recommender system,” which we’ve all encountered on Amazon and Netflix, would have allowed MITx to discover that perhaps certain lectures or demonstrations or labs, or different types of tutorials given in a specific order, help some students more effectively at different points in a course.

So if “Fred” got stuck on the notion of “sinusoidal voltage,” the system might push him back to review electrical impedance in general or Ohm’s law in particular. If he was still having trouble, a tutorial might kick-in about how to set up the equation and perhaps even relearn how to use sine and cosine in such equations. And if the system noticed Fred had been having difficulty with similar previous “talking head” lecture tutorials, it might now send him to a different kind of tutorial heavy on visuals and graphics. And, to be honest, if Fred was still having difficulty after all of this, the system might softly suggest that he just stick to playing Angry Birds for now and try again next year. I think of this as the “WWN solution”; the amalgamation of Wikipedia, Watson (the Jeopardy-winning computer), and Netflix. A fully-automated, massively-networked, natural language processing, data-driven, feedback-friendly, learning analytics system. This is what MIT should have done.

For if they had done so, MIT would have solved one of the truly difficult issues of higher education—the interlinkage of quality, accessibility, and cost—by giving any student from anywhere the chance to get any set of baseline knowledge and skills they need. Moreover, they would have found the solution for Baumol’s cost disease. Much like an orchestra today still takes the same amount of time to play a Mozart symphony as it did in 1800, there is a seeming productivity cap in higher education due to the labor-intensive process of actually teaching students something. Technology seemingly cannot simply transform or supplant or speed up the deeply dialogical process of student-teacher give-and-take.

Until now. For a well-designed MOOC is relentlessly efficient. It can provide immediate feedback through highly structured and sequenced steps at increasingly fine-grained levels of detail with ever-better responses based on what works for different students learning and clicking in different ways. Well-designed studies have begun to show that not only are such automated models as good as human teaching or tutoring, in many cases they increase certain types of learning by 20-50 percent.

And that makes sense. Educational research has long demonstrated that much of teaching (and tutoring) is highly prescribed and structured, with both students and teachers engaged in a fairly scripted process with minimal “out of the box” interactions or insights. While this may be a sad commentary on our K-16 educational system, I would gladly have a well-designed MOOC do the heavy lifting of answering and demonstrating, and answering and demonstrating again (and again), the basic questions and procedures that all students need to know. For while I deeply respect my faculty colleagues, we are, alas, only human. We have limited office hours; we may teach lots of classes with lots of students; we may have way too many papers to grade; we may even, dare I say it, not care after the 23rd time we have to explain the same dang concept. So I say: go MOOCs!

But such potential success has within it the limit to its own revolution. And that is what is truly transformational about MITx: It shows us the limits of such automated education and thus the possibilities for what it is we do (or should do) in higher education. Namely, such systems are immensely powerful for teaching very specific kinds of content knowledge—knowledge that is stable, solvable, and singular. Ultimately, computer-based systems have to be able to distinguish between “right and wrong solutions” within a closed-loop system. This is what the learning and cybernetics theorist Gregory Bateson termed “learning 1,” the ability to learn specific content through a process of trial and error in ever-increasingly faster ways. It is a closed-ended process of better grasping specific knowledge for a specific purpose.

But what Bateson is really known for is making clear that there is a fundamental leap from “learning 1” to “learning 2,” or what is commonly referred to as learning-to-learn or meta-cognition. Bateson theorized, and much research has confirmed—whether in the language of Schon’s “double-loop learning” or Mezirow’s “transformational learning”—there is a moment of insight, a meta-cognitive leap, whereby the individual is able to see the system within which she had been operating in and modify the pattern of behavior based upon this new-found higher-order perspective.

It is here, in those “aha moments,” where we find the limits of MITx and all such systems. The AI community refers to this as “brittleness,” in that all systems work so long as one stays within the parameters of the system. But as soon as the topic requires us to think about our thinking, to question our assumptions, to be creative in any shape or form outside of the predictable, such a system becomes untenable. The question as such is not whether MITx is better than human faculty. The question is for what content and towards what ends. For in the real world, we humans do open-ended things with closed-ended knowledge. This is what no MOOC can teach.

The real revolution is thus not really about access or cost or quality, as such. It is about the access, cost, and quality of what will be taught. MITx makes all of us in higher education confront the essential questions: What do we teach, why do we teach it, and to what ends? Whether one wants to talk about this through the language of our “value proposition” or the scholarship of teaching and learning, MITx has potentially given us the opportunity to put words and value to why we matter, to why we walk into a classroom (virtual or physical) and push and prod our students to think differently than they did just five minutes ago. For education, in its most profound manifestation, is a journey into the unknown, of learning to think and act in a fundamentally different way about something we thought we knew and had taken for granted.

The philosopher Ludwig Wittgenstein once said true learning was like climbing up a ladder and having the rungs below you fall away; you can never go back to what you had previously thought once you have seen a different perspective. This is the real college revolution we have all been waiting for, and MITx could actually help us get there. So while I do wish those 10,000 students luck on their final exams, my fingers are really crossed for MITx’s ability and resolve to truly usher in a new teaching and learning paradigm in higher education.

For more information, Contact www.eFOURlearning.com

Originally posted by Dan W. Butin in elearn Magzine.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s