Copyright © 2013 jsd

Teaching – Some Ideas on How It Should Be Done
John Denker

1  Introduction

Here are some ideas about the future of education. Some are obvious, and some not. Some are easy to implement, and some not.

The central idea is to imagine a world-wide educational system, and to make the system work well.

The motto should be: You have to make it good and you have to make it scalable.

Tradeoffs will be necessary. For example, if your name was Philip of Macedon and you had unlimited resources, you would hire Aristotle to tutor your kid. However, most people don’t have unlimited resources, and there aren’t enough tutors to go around. So we will have to cut come corners.

On the other hand, as Knuth put it, premature optimization is the root of all evil. We don’t want to cut corners more than necessary. Consider the contrast:

Doing things at scale imposes some limitations, because some things you would like to do simply can’t be done at scale.   If you are successful at scale, it releases some limitations, because you have more resources. You can do a lot more preparation. Any fixed costs can be amortized over a huge user base.

We should recognize multiple market segments:

We need better textbooks. On my desk I have a physics text that is 1200 pages long, with roughly 1200 significant deficiencies in it. And this is one of the better ones. I’m not talking about missing commas and other typos.

Remember, this is one of the better books!

Writing books is hard, far harder than most people imagine. This does not, however, mean that we should tolerate lousy books. Instead it means we need to apply sufficient resources to the problem. By way of analogy, suppose you want to build a fire:

The point is, if you apply insufficient resources, all the resources go to waste. On the other hand, if you arrange for sufficient resources, you achieve ignition, and the result is tremendously useful.

Rather than a large selection of lousy books and videos, we would be much better off with small number of really good books and videos. The task of organizing and focusing resources is a management task. The educational system and the publishing industry must be considered prime examples of managerial malpractice.

If you’re going to write a modern book, you need to write two books: one for the students, and one for the teachers.

Let’s be clear: The teachers are a super-important part of the team. You neglect their needs at your peril.

Consider the example of PSSC. The book was clear, correct, and up-to-date when it was published in 1960. From a student’s point of view, it is probably the best physics text ever written. The course was supported by videos and by numerous “canned” experiments for students to carry out. However, it was failure in the marketplace. Anybody who wants to write a book had better understand PSSC failed.

My theory is that most teachers didn’t like it, because it was too radical. It was too unlike how they had been taught, and too unlike how they had been teaching.

It does not suffice to create a “teachers’ edition” that is basically just the students’ edition plus answers to the exercises. The teachers’ needs are different from the students’ needs. The teachers start out in a different place, and need to end up in a different place, different from the students.

Figure 1: Teachers’ Needs ≠ Students’ Needs

We need an integrated science curriculum. The math and the computing need to be integrated with the biophysics.

Caveat: This is hard to set up. It requires exceptionally wise faculty and wise administration.


I’m not convinced it is good to think in terms of MOOCs, but if we do that, we can discern a three-step progression of ideas:
  1. There is the classical on-campus class, with a few dozen students.
  2. There is the ordinary MOOC, with a few thousand students from around the world, all taking the course at the same time, phase-locked to the on-campus course.
  3. There is the set of online resources which can be used and re-used in many ways, including:
    • This includes organized group courses. The groups do not need to be huge; one could imagine a new group of 100 starting every week.
    • This includes fully-individualized courses, such that a student could start any time, progress at any rate, and finish the course at any time.
    • This includes using arbitrary subsets of the materials for independent study, without any notion of starting or finishing “the” course that gave rise to the materials.
    • This includes incorporating arbitrary subsets of the materials into other courses, possibly similar, possibly very dissimilar.

It seems to me that item (3) – re-using the materials – represents a great deal of the value. That’s where most of the users are, if you integrate over time. Let’s be clear: We should think of the online materials as a valuable thing unto themselves. They exist partly as an offshoot of the original course, and partly as a resource in support of the future courses, but they also exist as a thing unto themselves. We cannot fully foresee – much less control – how they will be used.

We need to devote serious effort to optimizing the online materials. We should not think of them as a passive offshoot or by-product of the original course. Similarly, as noted in item 1, we don’t want to cut the wrong corners. There are some optimizations that are appropriate for the online materials that would be unnecessary in an on-campus course ... and vice versa. If the goal is to optimize the online resources, don’t be constrained by archaic classroom-scale thinking.

We should keep the videos short. It is always easy to make a long presentation by stringing together short segments. In contrast, picking apart a long presentation is not so easy.

Rationale: Having lots of short segments gives the users more flexibility. I refuse to argue about the order of presentation of the various topics and sub-topics.

Many of the Khan Academy videos are rather poor by any absolute standard ... yet students find them useful. Anybody who wants to play in this space needs to understand this. Factors to consider include:

The online materials should include tables of contents, lots of hyperlinks, and other aids to navigation.

I recently saw an hour-long video where one of the students asked a question, and the professor said, “That’s a good question. We will cover that in detail in about six weeks”. It would have been much nicer if the video contained a hyperlink from the question to the answer.

The whole course has been on the web for years, so there has been plenty of opportunity to go over it and do some nonlinear editing. It strikes me as penny-wise and pound-foolish to skip the editing step. It’s disrespectful to the viewers.

This is related to the idea that the online materials are a thing of value unto themselves, not merely a transcript of the classroom lecture, as discussed in item 6.

We should not design the course to suit MIT students. They have been selected to have strong technical background, strong study skills, and high motivation. If you can successfully teach them, it proves almost nothing, because they would have done fine without you.

The real trick is to teach the other 99.9% of the population, the mass market, the ones who do not have a priori high motivation, strong study skills, and strong technical background.

We must treat the documents as living documents. That means there needs to be a well-oiled mechanism for making continual improvements.

We can use the Linux kernel as a reasonable standard of open-source professionalism unless/until we come up with something better. Salient features include:

  1. There is a formal document control system.
  2. There is an elaborate bug-tracker.
  3. Proposed changes are discussed amongst the stake-holders.
  4. Approved changes undergo an elaborate sign-off process before being committed to the repository. Here’s an an example of a patch committed to the Linux kernel.

The approach suggested in item 11 is emphatically not the same as the Wikipedia approach.

Allowing free-for-all editing is not a good model ... especially if reform is one of the goals. Wikipedia articles on introductory physics concepts tend to have problems. Even if you fix them, they don’t stay fixed. Everybody who has ever taken the intro physics course thinks he knows the right answer, even if the course was 20 years ago and was 100 years out-of-date at the time. People who don’t know what they’re talking about vastly outnumber those who do, and tend to have a lot more time on their hands. Reform – more or less by definition – goes against the conventional wisdom. In accordance with Wikipedia rules, anybody who can find an “authoritative reference” for the unreformed approach is allowed to undo any reforms ... and there’s not much the reformers can do about it.


Copyright © 2013 jsd