The Open Education Manifesto

This is a long piece. I've been interested in radical reforms to the US educational system for some time. This is one of those areas where I'm just a radical radical, nothing moderate about me. So, without further apology or ado:

Introduction

We stand at an educational crossroads. It is apparent that the current public education model is becoming deficient. We must fix it. That fix may involve simple tweaks to make incremental improvements or it may require more revolutionary changes. I believe that the latter will more than likely be needed.

The concept of open education has recently gained some international currency. Many colleges and universities, especially MIT, have committed to providing their courseware online for free public use. Organizations like the Open Educational Resources Commons are acting as clearinghouses for free courseware and software.

However, open education today is a hodgepodge. Even MIT has done little more than place PDFs of course notes, problem sets, and tests online. The current use of open materials is limited to enhancement of the existing classroom regime.

Open education holds much more promise than that. In addition to improving the quality of curricula, open ed can serve as a platform for decomposing the current educational system into its component parts, allowing best-of-breed solutions to arise for teaching, facilities, administrative, and even economic systems.

Open ed can be used to empower teachers, giving them unprecedented mastery of their curricula. With that mastery comes the ability to cultivate a reputation that will lead to higher pay in a truly merit-based system.

Finally, open ed changes the dynamic of who gets schooling and when. While open ed can dramatically improve traditional primary and secondary education, its true promise may lie with the ability to re-train adults, allowing them to gain new, marketable skills at will, for a fraction of the cost that acquiring those skills costs today. Open ed will also enable gifted students to go farther, faster than they ever could with traditional learning models. In short, open ed can be much more than an adjunct to the twentieth century educational system; it can act as the core infrastructure for a complete educational overhaul for the twenty-first century.

Success in the Twenty-First Century

The twentieth century educational system had its roots in a nineteenth-century industrial economy and was amazingly successful at turning out large numbers of factory workers. Factory workers needed to understand complex tasks but they hardly ever had to design those tasks. The reasoning skills honed by learning to read, write, and do arithmetic were adequate to guarantee that most people would be able to work and earn a living for their families.

That manufacturing economy still exists, even in the United States, but it has become so productive that the need for workers actually to work manufacturing things has plummeted. Knowledge workers, who used to be turned out in tiny quantities by the American educational system, are now coming to dominate the workforce.

Knowledge work is highly specialized. It requires vastly different reasoning skills. It requires lots of, well, knowledge. Furthermore, knowledge work is ephemeral. Today’s hot job may be obviated by new technology, either because the technology allows the job to be exported to a place with much cheaper labor or simply because the technology renders the profession irrelevant.

A successful knowledge worker has to learn how to learn. He has to carry a bunch of specialized junk around in his head as well, but ultimately he needs to know how to replace one set of specialized junk with a completely new set. Not only is the American ed system rather poor at transferring such detailed knowledge, it doesn’t equip workers with the flexibility needed to switch fields two or three times in their life without destroying them financially.

The problem certainly has its pedagogical components but an even bigger problem is economic. It’s simply impractical to train for a new job while you’re holding down the remnants of the old one. The American classroom system requires more-or-less full-time concentration and a significant financial commitment to receive specialized training. That has to change.

Who Is the Customer?

In the twenty-first century, there are three major categories of educational consumer.

First, there are still lots of kids. They need to have sophisticated cognitive skills developed in elementary education so that they can begin the process of acquiring advanced knowledge in secondary education and possibly in college. But let’s consider the differences in literacy in the twentieth century and the twenty-first. If a young adult could read, write, do arithmetic, understood American civics well enough to vote intelligently, and had a smattering of other general knowledge, he was considered literate.

Contrast that with today: Can he type? Does he know how to generate intelligent search queries? Can he use office productivity software? Is he resilient enough to deal with overwhelming amounts of information and remain productive, to say nothing of remaining relatively happy? Does he understand how to seek employment in a new field? Is he self-aware enough to know the difference between skills that he can acquire successfully and those that are permanently beyond his competence? A large number of these tasks are essential to economic success and they all require types of cognition that simply weren’t taught until fairly recently.

Even assuming that our children are successfully taught these skills when they’re young, they are still going to need periodic retraining. The number of adults requiring retraining, our second category, will skyrocket. If they’re very lucky, they’ll merely need their skills refreshed. More likely, they’ll need entirely new skills for entirely new jobs. Those new skills need to be imparted efficiently, cheaply, and in a timely fashion. The current educational system can do none of these things.

Finally, we need a better, more systematic way to allow our high achievers to explore new fields and gallop ahead of the pack. Classroom education is notoriously bad at this—overachievers require lots of support and are therefore disruptive to the normal kids. The social stigma that attaches to this sort of disruption discourages a large proportion of the bright kids and drags them back close to the norm. In an economy that eats innovation for breakfast, we simply can’t allow this to happen.

These three types of students define the task at hand: We need to impart a much wider range of knowledge much faster. We need to target students throughout their entire life. And we need to get as much high-quality knowledge—however obscure—to our brightest students.

How Does the Current System Fail the Customer?

In the twentieth century, the educational system was designed to push a fairly small amount of the same knowledge into a lot of people. We now need a system that pushes a lot of different knowledge into a lot of people while still maintaining the core knowledge of the past.

The old system used vertical integration to accomplish its task. All the elements of education were concentrated at the school.

• Curriculum was embodied in a set of proprietary (closed) textbooks. Because the knowledge to be taught was fixed, textbooks became a big investment and a central way of extracting value (money) from the educational process.

• Courses were taught to a block of students at a fixed time. Scheduling was inflexible because it revolved around the teacher.

• For the same reason, courses had a fixed location where the students and teacher rendezvoused. Schools actually went into the transportation business to get their students to one spot at the right time.

• The scheduling, location, and curriculum were all coordinated by a large, rigid administration. Ultimately, that administration acquired the power to regulate how teaching occurred. It dealt with disruptive students. It provided quality control on teachers. It set teachers’ salaries.

The economics of such a system made sense, the same way that the economics of a “brick and mortar” retailer made sense before the Internet and high-speed package delivery systems.

This model is still valid to some extent, especially for small children, who are unsophisticated and undisciplined. But some of the economies of scale that used to make sense are no longer relevant.

The current system carries with it a huge administrative load that is no longer necessary with the self-organization available over the internet. Instead of providing top-down agenda-setting and quality control with a centralized administration, these tasks can be handled through social networking and reputation-based metrics on the Internet.

Distance learning via the internet can reduce facilities costs, especially for older students. The economies of scale incurred by large, centralized schools no longer make sense. There is still a need for specialized facilities (e.g., labs, gyms, and playing fields) but classroom space can now be managed much more flexibly.

The heavy administrative cost of education also negatively affects the prestige of the teaching profession. In a system where public education is under constant budget pressure, administrative costs severely curtail the salaries of the best teachers. Because there’s inadequate remuneration, the teaching profession attracts fewer talented teachers than it otherwise would.

Finally, there is one additional factor that the Internet has influenced. There is a growing societal consensus that the vast majority of educational knowledge should be free.

How Open Software Changed the World

When I was a lowly computer science undergrad in the late 70’s, we heard rumors of this crazy person named Richard Stallman. While we were busy training to become hotshot, highly paid software engineers, Stallman was advocating that software was ultimately a collaborative, free commodity. We were of course bemused and appalled.

Stallman went on to write the GNU Manifesto and found the Free Software Foundation. He also started writing the GNU operating system, which ultimately became Linux. To say that his vision has been successful would be an understatement.

Why did this work? Something clearly changed between the time that Stallman was sleeping in his office in CSAIL and now. How did open software become such a powerful force?

The first thing that changed is that GNU, and later Linux, were amazingly high-quality products. This occurred because the emerging Internet allowed excellent collaboration between a self-organizing community of experts who cared passionately about what they were building. It was a labor of love. Even more important, it was a labor that could significantly enhance your reputation in the software engineering community. When you get lots of motivated eyes looking at the same content, the content improves rapidly.

But there was a less obvious change pioneered by open software. The content was not the value proposition. It was all the services surrounding the content that allowed communities of entrepreneurs to make a good living from a free commodity. The implications of this are still reverberating through the software industry. Just as in the educational establishment, software was vertically integrated: Production, delivery, support, and follow-sales were all handled by one company. Open software ended that. Production and delivery were free. Support became a viable business. But even more important, it was the uses to which the software content was put that had value, not the content itself.

Recreating Education

The state of the software industry prior to the widespread adoption of open software is almost identical to the state of the educational establishment today. Content is one of the principal drivers of value in education. Vertical integration entails high administrative and facilities costs.

And yet, knowledge wants to be free.

All of the same motivations that made open software successful—a sense of duty, a desire for reputation, steadily improving collaborative tools—apply even better to educational content. A huge community of educators, subject matter experts, and simple enthusiasts is prepared to produce an explosion of educational content. Indeed, the explosion has already begun.

But the open education movement is missing infrastructure. It needs an operating system. It needs the educational equivalent of Linux.

Today, content providers are reinventing the wheel, over and over. They provide their own tool sets. They improvise—or ignore—standards for reporting student progress and teacher quality. They ignore the need for well-defined prerequisites. They have no facilities for consolidating individual courses into well-defined curricula.

As a result, the current model for open education is one where open content supplements the existing classroom infrastructure, rather than changing the underlying educational process itself.

A common infrastructure will change that. With commonality comes reliable standards. Those standards allow not only private and home educators to count on the content. Public institutions can then begin to experiment with models that reduce costs while enhancing flexibility, all the while improving quality

The Economics of Open Education

To see how open education can affect the economics of the educational establishment, let’s look at a cost breakdown for a typical school district. I’ve chosen the Leander Independent School District, a fairly affluent district in suburban Austin, where I live. (I’m still looking for a national average cost breakdown.) Here’s a roll-up of LISD’s expenditures for the 2006-2007 school year:

How would this cost breakdown change under an open education regime? First, note that 12% of LISD’s budget is related to content, curriculum development, or training its teaching staff on that curriculum. Under open education, that figure drops dramatically. However, we’ll need a bigger computer budget, so let’s assume that content costs are still 33% of what they are today. Note also that facilities costs are currently substantial because all students are physically present all the time. While elementary school facilities costs probably don’t move very much, a substantial amount of secondary curriculum can be done remotely, thereby reducing facilities and administrative costs. Let’s ballpark facilities and administrative costs by assuming each drops to by 50%.

When you do the arithmetic on this, the LISD total expenditures drop from $166,416,539, to $142,434,959, or an expense reduction of 14%. I suspect any school district would sell its mother for that kind of cost reduction.

Now let’s look at a re-training scenario for a junior college. Here is a roll-up from the Foothill-de Anza Community College District in Northern California:

Note that community colleges don’t normally provide textbooks, as elementary and secondary schools do. As a result, I’m adding a fudge factor to these numbers to cover approximately $400 per student per year in textbooks. (To its credit, Foothill-De Anza is embarking open an extensive open textbook project, in collaboration with the OER.)

Now, the vast majority of community college courses are classroom courses, but there are also lab and vocational course that require physical plant. However, most of a community college curriculum, especially one designed for worker retraining, can be accomplished remotely. If you were to eliminate the content expenses, reduce admin/ops by 75%, and reduce facilities expenses by 50%, you’d take the Foothill-De Anza expense budge from $335,491,451 to $195,667,598, a reduction of a whopping 42%.

Foothill-De Anza estimates its annual cost for a full-time student living at home to be about $9600. A 42% reduction takes that number down to about $5570 per year. When you add to that the ability to do a big chunk of coursework in a self-paced fashion, which in turn allows the student to hang on to his current job while he retrains, suddenly retraining goes from an long, difficult, risky enterprise to something that’s achievable by almost anybody. (Note also that many retraining curricula are not geared to full-time students, so the annual cost might be even lower.)

Teaching Should Be an Honorable Profession

From the figures above, it should be clear that the largest chunk of education is still teachers’ salaries. Given that we need to drastically reduce the cost of education, that doesn’t bode well for upward pressure on teacher pay.

And yet there is still no substitute for talented teachers. Equally important, there is no excuse to place innocent students at the mercy of untalented teachers. And yet the educational system today is carefully constructed so that the most talented will avoid the profession and the least talented can hang on forever.

Teachers today are at the mercy of an inefficient, vertically integrated educational system. They are constrained by work rules, by salary rules, by administrative dicta that may or may not make sense. They are forced into inflexible facilities, teaching groups of students that may or may not be too large for their capabilities.

There are two ways that the lot of teachers can be improved dramatically.

The first is, as it is with so many fields of modern endeavor, through productivity enhancement. It is always a paradox that increased productivity, which temporarily reduces the demand for workers by allowing fewer workers to produce the same amount of a good or service, ultimately raises the wages of the remaining workers. But when a particular type of labor is in short supply, as it is with teaching, increased productivity merely feeds demand for a higher quality of service.

If a student has superb curricula and lots of non-teaching support (more on this in a moment), the teacher is freed to teach larger and better-prepared groups of students. If you then set up compensation systems that scale with both outcome and popularity, teaching becomes a truly elastic commodity. As a teacher’s reputation improves, he can charge more for his services. As his productivity improves, he can increase his output. Both these factors mean that the best teachers can be substantially better remunerated.

Note that we’re assuming a radically different pricing and attendance model for teaching services. Far from the salaried, one-size-fits-all model that’s used in most schools today, teaching can be charged for per course, per student, even per individual teaching session. This gives enormous flexibility to the teacher and the student. One can even imagine a pool of nothing but the equivalent of substitute teachers, who are assigned to pickup distance learning classes as enough students become available for that class to be convened. Teachers can work as much or as little as they like—as can students.

To have the market reward the most productive teachers, metrics for measuring that productivity are essential. This should be a major factor in an open education architecture. If the outcomes for teachers of specific courses are available to parents or students wishing to take that course, they should be better able to select the most cost-effective teachers they can find.

The second factor that can improve teachers’ lives is specialization. Today, a teacher (at least an elementary or secondary school teacher) does lesson prep, runs his class, grades tests, and helps individual students with the class. In an open educational system, all of those functions can be distributed to different categories of “teacher.” Teachers can teach. Tutors can supplement teaching by helping students one-on-one (or virtually one-on-one). The tasks of grading exercises, papers, and tests can be jobbed out to individuals certified to do those tasks. (Of course, exercises and tests that have unique right or wrong answers can be graded automatically.)

Some classes will require facilities, and a business can be made from providing teaching facilities. Dedicated lab providers can make a business out of their expertise with specific lab equipment. Where classrooms are appropriate, classroom managers can procure, maintain, coordinate, and schedule class time. Athletic coaches can run their own athletic facilities.

Finally, administration, especially student administration, doesn’t go away completely. If education is going to have value, there still has to be someone to certify that students have completed a particular course and have performed to certain standards. Student record-keeping and certification will therefore be an important business, depended on by students, teachers, and prospective employers alike.

All of this specialization can allow the teacher to concentrate on what he does best, be it class teaching, one-on-one tutoring, homework correction, or any combination. That can lead teachers to specialize in areas where their skills are best suited to making a living, further maximizing productivity.

An obvious question needs to be answered: If all of this teacher productivity is supposed to generate revenue, how are content providers to survive in this ecosystem? That question has a few answers, none of them totally satisfying.

First, it should be pointed out that content providers are teachers, too. To the extent that somebody is teaching their own content, they charge for the teaching, while the source material remains free.

Another answer is that there is still another role in the ecosystem. Somebody has to certify that teachers are qualified to teach or perform other roles with a given set of course content. It seems natural that a content provider should be able to charge for such certification. That isn’t the most robust revenue stream ever, but it can certainly provide a living.

Perhaps the best answer, though, is that free content is likely to be collaborative content. Many experts will contribute to and refine it, for no other reason than pride, reputation, and belief in the subject matter. As with open software, this answer is a bit counter-intuitive for us hard-bitten, money-grubbing professionals. And yet it’s clearly true.

An Open Educational Architecture

Having spent a lot of time on what open education is good for, it’s finally time to describe what it is today, and what it ought to be going forward.

Today, open education is a grab-bag. There is courseware. There are various tools for enhancing various educational experiences. There are exercises, some with answers, some without. However, in almost all cases, there are several things that open education is not, and needs to be:

• There is no standard for how to author course content.

• There is no way to define course prerequisites or links to further information.

• There is no way to package sets of courses into curricula.

• Tools are by-and-large not reusable.

• There are no standards for how to package exercises, projects, and tests.

• Facilities for gathering statistics on exercises and tests are sparse to nonexistent.

• There is no common record-keeping mechanism, nor are there tools to identify students who are struggling and help correct their problems.

• Teacher certification is nonexistent.

• Open educational materials, with only a few exceptions, are not standalone courseware. Rather, they are designed to supplement existing classes.

• Distance learning tools—audio and video conferencing, screen sharing, and lecture moderation tools—are hit-or-miss.

• Hardware and network requirements are indeterminate.

So the first thing that’s required for open ed to take off is an architectural framework. Ultimately, this means that a “course” becomes a well-defined, complex media type, with predictable characteristics.

The Infrastructure Surrounding Courses

A course can’t stand alone. Teachers or other “course conductors” guide students through the course and monitor their progress. The course content has to be rendered to students and teachers using a well-defined set of tools that are appropriate to the kind of content being taught. Students and the various types of course conductor (teachers, classrooms, labs, tutors, etc.) have to be able to rendezvous at specific times and/or places via a common schedule or “registrar,” which can also broker the student’s selection of course conductors, if needed. And, finally, a record-keeping system that someone will vouch for—a “certifier”—needs to back the whole thing up or the educational content can’t be relied upon to offer value to students, employers, and educators.

But the course is clearly where most of the intellectual value lies in open education. As such, we want the courseware to be as independent of the nasty details of enrollment and record-keeping as possible. However, because registrars and certifiers and teachers and students need to interact with the entire open educational architecture, it’s essential that the course have a well-defined structure. On the Web, we often use an Extensible Markup Language (XML) schema to define such a structure. A course content schema—let’s call it CourseML—would represent a major leap forward for open education. By standardizing how the lessons, exercises, projects, and tests that make up a course are represented on the Web, the rest of the architecture can be engineered to support students and teachers that depend on the course, no matter what its content.

Courses and Tools

The content contained by a course needs to be presented to and interacted with by the student. Course conductors also need to be able to control and present content to the student while they deliver a lecture or other learning session. This means that content must have a set of tools associated with it. Since we can’t judge what sorts of content will be presented in a given course, there will be many different types of tools associated with content.

However, we should be able to rely on a standardized set of basic tools that many courses will need. Like the courseware itself, tools need to have a regular set of properties, so that content can count on tools behaving in a predictable fashion when they mediate the interactions between teachers, students, and content.

On the Web, the tools are attached to the content as part of the description of the content. Tools are then executed, partially on the web server, partially in the user’s web browser, using Web 2.0 technologies similar to “widgets.”

The most important tool is the course player. It is responsible for managing the flow of lessons, exercises, projects, and tests associated with the course, interacting with students and course providers, and interacting with the record systems. Because the tool is a standard tool, content providers merely need to know how it behaves and don’t need to know the details of how it keeps records or manages students and teachers.

The course player is also responsible for launching many of the other types of tools. If the class is a distance learning class, it is responsible for activating and interacting with class conferencing tools. If the current flow of the course requires that an exercise be performed, it can activate an exerciser tool that will ask the student questions and gather the answers. If an exercise needs human-mediated grading, the course player can activate grading tools for certified course conductors. Some exercises need time constraints, for example in the case of tests or academic competitions. Specialized tools can provide these services as well.

There are types of content that occur over and over in an academic situation. Students need to read text. They need to write text. They need to work math problems, all the way from simple arithmetic up through complex mathematics, using mathematical notation that’s difficult to enter on a standard keyboard. By having a common, open set of tools for these activities, students, teachers, and content providers can rely on a standard way of performing these basic tasks. This dramatically simplifies exercises, testing, and grading.

Finally, some courses will require content which doesn’t fit neatly into the standard tool set. The open education architecture must therefore accommodate unanticipated custom tools. These tools may be specific to a particular course, or they may be re-used by many courses sharing the same need. Ultimately, the only difference between a standard tool and a custom tool is its generality. In an open environment, all tools should be available to all content providers. Only at the highest level should specific tools (like the course player) be mandated.

Students and Course Conductors

A student in the open educational architecture is any person who’s learning the course materials. Because learning is a lifelong endeavor, it’s important to maintain a long-term, authentic, verifiable persona for each student. The student should have permanent access to all of his coursework, all of his grades, and all of his learned skills. With the student’s authorization, a third party, such as an employer or an educator considering the student for admission into a course, should be able to access the student’s records.

While a student is taking a course, note that the course itself doesn’t maintain the student’s current progress. The student and the student’s records do. When the student accesses the course, this information needs to be passed to the course to so that the student winds up in the right spot in the lesson plan. (This is obviously more important for self-paced courses than it is for teacher-led courses.)

Course conductors are individuals (or robots) who lead one or more students through course materials. (Note that self-paced courses may have no course conductors.) Because the teaching of a course may be quite complex, many subtypes of course conductors may be associated with a given course.

The most obvious course conductor is a teacher. Teachers interact with a set of students in a class. The interaction may be in a physical classroom or it may occur through video, audio, and data conferencing tools provided by the course player. Typically teachers teach students in a one-to-many relationship, lecturing and answering questions in a structured way.

In contrast, a tutor typically teaches an individual student one-on-one. Furthermore, tutors may be used on an as-needed basis. For example, suppose a student is taking a self-paced course and runs into something that he doesn’t understand. The student can enlist and conference-in a tutor to help him. Once the question has been answered, the tutor can be dismissed. Alternatively, a student can make regular use of a tutor.

Notice that the open educational architecture provides a new, more flexible market for tutors. Tutors can be certified as experts in various course materials and registered as available resources, to be used—and compensated—on demand. This can make tutors a very cost-effective means for students to get questions answered, while providing the tutor as much work as they wish to handle.

Note that a teacher and a tutor may have nearly identical skills, although a teacher may be more skilled in dealing with a class, while the tutor deals with students one-on-one. Another role that can relieve a teacher of much of the drudgery of teaching is that of a grader. Graders examine, correct, and grade exercises or tests. They may not have quite the same grasp of the material as teaches or tutors, but they must be able to examine and correct the material.

Note that, in many cases, courseware may be able to automatically correct exercises and tests. For example, math exercises often have a single correct answer. However, even in math, complicated problems often have more than one solution, and the examination of the solution may indicate whether the student has as firm a grasp of the material as it appears from his final answer.

Again, grading of exercises becomes a nice market for individuals who are expert in a course but who don’t wish to interact directly with students. Typically, graders will be retained by teachers, but there’s nothing that prevents students from directly obtaining graders for self-paced material.

Since some courses require physical premises to be completed, a final type of course conductors is a facility. When classrooms, labs, playing fields, etc. are required for a course, the facility object represents the individual or organization that can operate the physical premises.

All course conductors share a common property: somebody needs to certify that the course conductor is competent to conduct the course. Each type of course conductor may have a different certification process. We’ll cover some of the issues surrounding certification later, but for now note that a final type of course conductor is indeed a certifier for the course. Perhaps a better way to think of the course certification process is that course conductors need to be students of a separate set of courses that teach them how to teach the course. However, since the course content provider is likely to be the expert in how to teach the content, it makes sense to package the course materials for certifying course conductors along with the course itself.

To become certified, the course conductor goes through the certification content associated with the course. This content has exercises, projects, and tests, just like any other course. The material may be self paced, or it may be taught by a teaching certifier, who is, in turn, the final type of course conductor.

All course conductors also have a need to demonstrate their relative proficiency in teaching the course. To this end, course conductors have a reputation. The reputation of a course conductor is key. If reputation is authentic and transparent, it will be an excellent indicator of how successful students will be if they take the class from that individual.

Registrars

Because we’ve taken care to ensure that course developers are unencumbered by (nonetheless important) administrivia, the course itself is stateless. This means that all the mechanics of advertising, scheduling, record-keeping, and even payment must be located somewhere other than the course. In this open education architecture, the focal object for this is the registrar. When a student enrolls for a course, he does so with the registrar. When a teacher or other course conductor offers to teach a course, he does so with the registrar. When new courses or new versions of courses become available, they can be advertised through a registrar.

The registrar is a focal point where students can find out about offered courses and the teachers that teach them. It may or may not act as a repository for student records. Perhaps the easiest way to think of the registrar is to think of it as a current school administration.

A final, extremely important role for registrars is that they can act as a clearinghouse for financial transactions between students and course conductors, or between course conductors and other course conductors (e.g., between teachers and graders). The open education architecture specifies nothing in this area, but the registrar is the place to add such functionality, should it be needed.

Record Keepers and Certifiers

Because student records have value, both to the student and to organizations that need to know the student’s proficiency in certain subject matter, the authenticity and validity of records is vital. Course conductor records and reputations are equally important.

Many of the objects in open education have records associated with them. Students have records associated with each course they’re taking or have taken, and also have a skills record that provides a view of the student’s accumulated proficiencies, independent of the courses he’s taken. Registrars need to maintain enrollment records, which also include course scheduling information. Teachers and other course conductors need to have certifications for the courses they’re qualified to teach, as well as reputations.

Course conductor reputation deserves a bit of scrutiny. To run an efficient educational market, students need to know as much as possible about the teachers, tutors, and facilities they may decide to employ. The reputation record provides the student with the information he needs to make these decisions. Reputation will contain student comments, survey information, and proficiency information about the students that have been previously taught (i.e., how well the students who took the course with this teacher did).

Since records have distinct economic value, the temptation to forge them or otherwise fraudulently represent them is real. For this reason, the open education architecture contains the idea of a certifier. Certifiers have a wide range of functions, but the common task they all perform is that they guarantee that a record is authentic because they monitored the process by which the record was generated.

Because certifiers have to monitor a wide variety of open educational processes, auditing interfaces and tools must be built into many of the objects in the open ed architecture. This allows a certifier the ability to ensure that the educational processes for which it’s responsible are being conducted in accordance with the certifier’s standards.

A certifier certifies a record using standard certificate technology. If you’re a computer geek, you know what I’m talking about. If not, well, count yourself lucky and trust that this is a solved problem—sort of.

Here are some of the roles that a certifier can take on:

• Courseware can be certified to be in conformance with the certifier’s standards. This is a big deal for educational organizations that have to comply with external agencies’ standards.

• A course conductor can be certified as proficient in the teaching of a course or set of courses.

• Grades on exercises, projects, and tests can be certified by graders.

• A student can be certified as having taken a certain set of courses, received a certain set of grades, and acquired a certain set of skills. Note that a registrar will often act as the certifier for student records.

Hardware and Networks

Open education is inherently computer- and network-based. The hardware for pervasive open education is still a bit too expensive, but it won’t be for long. One Laptop Per Child (OLPC) hardware is close to what’s needed, but open education needs to be able to do good video and audio conferencing. That’s compute- and memory-intensive.

But it’s not necessarily disk-intensive. I suspect the ideal PC for open education has the following characteristics:

• It’s Flash-based. No hard disk needed.

• It has a video camera, audio microphone, and a speaker.

• It needs at least a half-VGA display, maybe full-VGA.

• It needs 802.11g and/or some sort of 3G mobile access, capable of handling 15 fps CIF or maybe VGA-quality video.

• It needs an excellent pointing device. (If children are going to learn virtually everything on a computer, the input techniques they use will have to be good enough to promote healthy hand-eye coordination. A mouse pad isn’t going to cut it.) I suspect that that means the PC needs Bluetooth and has the ability to accommodate a wireless mouse.

• It has to be incredibly rugged. Six-year-olds drop stuff. Since there’s no hard disk, the ruggedness goes way up. But hardening the PC against little kids is a major cost. Careful engineering will be needed.

• I suspect that more exotic input devices will sometimes be needed. Hopefully, these devices can be shared in classroom or lab settings. One that I can think of immediately is a graphics tablet that displayed as you drew. Maybe the time has come to stop teaching kids to handwrite, but I suspect not just yet.

In terms of network bandwidth, the hardest criterion to meet is to have good videoconferencing. This implies at least a megabit per second both upstream and downstream with low jitter. That implies a broadband network with Quality of Service (QoS) enhancements.

It may easily be that the lack of pervasive broadband access is the single biggest impediment to the adoption of open education. Hopefully we’re starting to wake up to the fact that antiquated network access is becoming a significant drain on American productivity.

Some Examples

Putting it all together, we have a relationship between courses, teachers, students, registrars, certifiers, and record-keepers that looks something like this:

Let’s examine two scenarios to see how this works in practice.

Self-Paced Education

Suppose I want to take a course to learn about, oh, say partial differential equations. As I browse through the web searching for open courses, I come upon a site called “Math ‘R Us.” This site is actually a registrar, specializing in self-paced, open mathematics courseware.

I find that Math ‘R Us offers a self-paced PDE course and I decide to sign up for it. During the sign-up process, I provide my current student record, which contains previous courses I’ve taken, certified by various certifiers.

Math ‘R Us first checks my records to make sure that I have the proper prerequisites. The PDE course requires that I have skills in Calculus, Linear Algebra, and Ordinary Differential Equations. It turns out in this fantasy example that I have those skills, so Math ‘R us begins the enrollment process.

First off, as a self-paced provider, Math ‘R Us charges for grading and testing services, not for actual classes. They charge a modest flat fee for the class, but they also maintain a pool of tutors, for which they charge an hourly rate. I pay my enrollment money and begin the class.

Note that, so far, I could have found the open ed PDE courseware and just plowed through it on my own. The reason I decided to use the registrar is because they have a relationship with a certified pool of graders and tutors and I want a skill certification for the course when I’m all done. So the courseware itself is free, but the value-add for certification, grading, and testing isn’t. Nor is the value-add for tutoring.

As I go through the lesson plan for the course, I am presented with various nasty math exercises. I use the standard math tool to work these problems. In many cases, the answers can be graded automatically by the courseware (which would be free). In other cases, I have to prove things that are beyond the capability of an automated grader. I submit those exercises to Math ‘R Us and they assign a grader to correct my problems.

When I come across something I don’t understand (a nasty Jacobean, maybe), I can’t request tutoring help. An available tutor is then brought into the courseware using conferencing tools and the clock starts running. I can ask him questions. Using his version of the math tools, he can demonstrate to me what I didn’t understand, walk my through problems, and generally set me straight. When I understand, I thank him and he drops out of the conference. Math ‘R Us will bill me for the tutor’s time, makes sure he gets paid, and no doubt take a little off the top for providing this invaluable service.

Math ‘R Us monitors my progress through the course. As I do exercises and tests, it accumulates those grades in my student record. When I’m finished with a lesson at the end of the day, I simply close off my web browser. When I return to the Math ‘R Us site, it offers me the ability to resume lessons where I left off.

When I’ve finally complete the course and passed all the tests, Math ‘R Us completes my records by adding in all grading information, updating my accumulated skills, and certifying that I completed the course. Note that Math ‘R Us is responsible for ensuring that I didn’t cheat. How they do that is up to them. Most likely they monitor my actual exercise and test work looking for indications that I’m not really working my own problems. But since I’m honest, their automated cheat detection software catches nothing and they’re willing to certify me. Note that Math ‘R Us has some skin in the game: If I turn out not to be able to do PDE’s in my next course, or in the course of my next job, their reputation suffers.

Public Education Using the Open Educational Architecture

Let’s now look at a traditional public elementary school that has adopted open education. For the most part, this is going to look a lot like a regular elementary school, but there are some notable differences. The most obvious is that there are no textbooks, but every child has a laptop (see above for what that laptop looks like). Laptops can be interchangeable, so they can be assigned to the child or kept in the classroom. A good compromise might be to keep them in class for very small children, but they need to go home by the time that kids are doing homework.

For children under the age of 8, almost nothing will change. They’ll go to school all day and will do almost all work from inside the classroom. Little kids simply need too much supervision to allow them to work remotely. However, older children probably don’t need to be in school all day. It should be possible to run two shifts of children per day through a single classroom, then teach remotely for the rest of the day. This should result in a significant facilities savings.

Teachers can make more profitable use of volunteers in an open ed elementary school. Most adults should be capable of acting in the “tutor” and “grader” roles. Since they don’t have to physically travel to the school, you’ll get more volunteers for more hours. As a result, teacher productivity should improve substantially.

Note that the principal’s office essentially becomes the open ed registrar. The office owns the teaching force, schedules the classes, coordinates the volunteer tutors and graders, certifies the curriculum and the teachers teaching it, and manages the student records. However, because the architecture is open, the school has the ability to allow students to supplement their education with outside tutors and teachers or even with additional courses. The teaching workforce becomes much more flexible, much more responsive to the needs of its students.

Migrating to an Open Educational System

The architecture I’ve laid out here is admittedly a long shot. A full implementation of the open educational architecture is nothing short of a revolution in education and societies aren’t very happy about revolutions. But revolutions do occur. When a situation is desperate enough, or dysfunctional enough, or threatening enough, change occurs. The current state of American education seems to fill the bill pretty well.

Some will argue that a computer-based, network-based, remote education will irretrievably damage human society. This argument has some merit. If kids aren’t socialized in school, where will they learn their interpersonal skills? However, as we have seen, open education doesn’t necessarily mean remote education, despite the obvious cost saving of remote education. Care will be needed to balance each child’s need for personal interaction with the demands of improving teacher productivity.

The textbook industry obviously won’t be wild about this. But the advantages of the open software experience are definitive. General knowledge simply isn’t going to be a commodity for which money is charged directly. Textbook publishers will have to find other avenues to add value. As we have shown above, those avenues exist.

The trend to open education has already begun. What is needed now is a way to scale it up so it can be used universally. That’s a big task but we have the tools to tackle that task.