Beat the book, not the teacher

In the early days of using computing for education, the challenge was “to do better than a teacher.” The challenge was to produce instruction as effective as one-to-one tutoring, what Bloom called “the two sigma problem.”  In response to that challenge, researchers created Computer Assisted Instruction (CAI) and later Intelligent Tutoring Systems (ITS), based on studies of how effective human tutors worked.  For example, Davide Fossati used human tutors as his comparison for his iList ITS for teaching data structures, and he was able to do as well or better with iList.

I’ve been perusing various offerings of distance courses for Computer Science the last few months.  For example, Andy Begel called me on my comment in a recent blog post, and pointed out that Berkeley offers self-paced CS courses based on the Kellar method.  I’ve also been checking out other on-line offerings to help figure out how we can do on-line CS teacher training for Georgia Computes.

When I look at these classes, I don’t see anything that looks like an intelligent tutoring system, or even looks like a teacher.  There is usually a teacher (an over-worked teacher — it’s hard to run these distance courses) playing the role of assessor and tutor (in the sense of responding to questions).  In general, computation is used as a transfer medium, a cheaper form of paper.  The source of information to be learned is almost always a textbook (often delivered as a PDF), or perhaps a set of HTML pages, or at worst, a set of Powerpoint slides.  Mike Clancy kindly shared with me a document about the evolution of courses at Berkeley, in response to my questions about their self-paced classes. Much of that evolution has to do with the creation of textbooks, which are considered particularly critical for the self-paced courses.

Why is that?  We seem to have given up on trying to do better than the teacher, perhaps because of the expense of developing ITS or because it’s so difficult to get a computer to teach better than a good teacher.  Perhaps it’s because replacing the teacher breaks our model of what makes courses effective.  We know how to make courses around books.

I propose a new goal: Beat the book, not the teacher.  Can we build computation that works better than a book for providing learning opportunities for students at a distance, and can we do it for less expense than building an ITS? We tend to design courses around textbooks.

Even the Open University’s new course My Digital Life is built around four books.  Books?  For the most cutting edge distance CS1 course?  John Daniel’s book on the Open University UK explains why they rely so much on books.  Books are manageable and cost-effective.  We know how the editing cycle for books works, and we can control the costs carefully.  Imagine that you have a bug in your Flash presentation.  Can you control the costs for finding and fixing that bug, at the same level that you can control the cost for proofreading, fixing the Word file, and reprinting?

Books are great, but most distance ed offerings don’t try to do better than books.

• There is no way that a set of Powerpoint slides is better than a book for learning.  What does Powerpoint add?  Less text, so less explanation, and now we can add flashy transitions and animations that make the meaning even more inscrutable.
• I’ve been surprised to see less classroom lecture videotapes than I expected.  I figured that taping the lectures and putting them up would be the cheapest way of putting any class on the Web.  I suspect that even that is more expensive than many schools are willing to bear.  Why offer the lectures at all to tape them if you can just post the notes and slides?

Can’t we do better than books?  Can’t we provide information that is more accessible, more engaging, and offers opportunities for practice — and at a low cost for producing the material?  The economics do matter.  ITS really do work, but we don’t use them more, most probably because of the cost of producing them.

That seems to me to be the next challenge for educational technologists that could improve the state of distance education as it is currently practiced. Not as it could be theoretically — actual practice is inhibited by costs. We need a a system (a cross between PowerPoint, Word, and LaTeX, with HyperCard-like scripting for interactivity?) that gives teachers a medium for producing instructional texts that are better than a book.  Something that would actually be used could make a big difference in making distance curses more effective.