Continuing Conversation...Cochrane on Education and MOOCs

EconTalk Extra
by Amy Willis
Cochrane on Education and MOOC... Postmortem on Cochrane on Educ...

In this week's episode, Roberts talks with the University of Chicago's John Cochrane about his experience teaching a massive open online course, or MOOC.

Let's continue the conversation...Share your responses and/or share them with others. We love to hear from you!

Questions are below the fold:

Check Your Knowledge:

1. Cochrane describes MOOCs as characterized by zero marginal cost and very large fixed costs. What does this mean? How does this distinguish MOOCs from traditional, face-to-face courses?

2. Roberts equates MOOCs with textbooks. What does he mean by this? To what extent is this an appropriate analogy?

3. Roberts asks Cochrane what he thinks the next step is for MOOCs. Cochrane argues MOOCs today are still "very much Web 1.0." What does he think is necessary for MOOCs to get to the next level?

Going Deeper:

4. Toward the end of the interview, Cochrane says, "There's a danger in discussing things like education, health care, to say everybody has to have the absolute best." Why does he say this? To what extent do you agree?

5. Do you think MOOCs have greater potential to "disrupt" high school or college level education? Why?

6. Arnold Kling is skeptical of the potential of MOOCs. Where do he and Cochrane disagree?

7. Kling argues that MOOCs should try to re-create the experience of a single student learning one-on-one with a teacher. Cochrane suggests the flipped classroom (where students watch instructional videos before coming to class) tries to re-create that. Is Kling right? Is this the best way to think about improving on-line education?

Extra Credit:

8. As they are discussing how to assess students in a MOOC, Roberts poses a challenge: "The puzzle is: Why do lightbulbs last longer than they used to? Because a lot of people think, lightbulbs wear out because they want to sell you more lightbulbs. It's a really bad answer." Roberts suggests the right answer is easy to identify and that a single word would help him grade your answer. So what IS the right answer? What is that single word?

Comments and Sharing

TWITTER: Follow Russ Roberts @EconTalker

COMMENTS (27 to date)
Jeff King writes:

8. The light bulb question:
Not sure what the single word is, technology? knowledge? The march of technology means we have better materials, better manufacturing processes, better design and analysis tools. The result is longer lasting lightbulbs that use different light emitting phenomena and end up lasting longer...until you drop one on the floor. We haven't solved that one yet!

Dave writes:

8. Competition: unless you assume collusion among all the manufacturers, providing longer lasting light bulbs should provide the innovator with a larger share of the market, and threaten the profitability/existence of its competitors.

Jonas Kölker writes:

1. [Explain: MOOCs are zero marginal cost and very large fixed costs.]

Marginal costs are the costs of one extra unit, in this case student. In a traditional classroom setting, adding more students means you sooner or later have to move into a larger lecture hall, which gives you a somewhat kinked marginal cost curve. The more significant exclusive resource (private good) in education is human attention, in the form of grading papers and giving individualized feedback. In a MOOC you don't have a lecture hall and individualized feedback is limited and/or automated. (On the demand side, there's a huge difference between free and cheap, so there's great pressure to keep it at zero.) There are fixed costs in videotaping lectures (camera, location) and especially software to give automated quizzes. If the MOOCs increase demand for some related service (e.g. exam certification), the charges from that can be used to cross-subsidize any non-zero marginal bandwidth costs.

2. [MOOCs ~= textbooks; how so?]

(I forgot what was said, but here's my analogy:) They are a resource which the receiver (student) can consume at their own pace, with them in full control of the pause and rewind buttons. They are also both identical for all consumers (unlike the individual feedback), and have a reasonable chance of becoming The Standard.

3. [MOOCs are Web 1.0]

You get the basics—play, pause and rewind—but no smart adaptation depending e.g. on right or wrong answers to interspersed quizzes.

4. ["everybody has to have the best" is bad, why?]

(I forgot again, but my take:) For one, it's not possible, since good service providers only have limited (and exclusive) attention, at most 24 hours per day. Secondarily, specifically for education, depending on how fast each student can drink from the fire hose, what is suitable for one may not be suitable for the other. But, a counterargument: everybody wants their instruction to use the best didactic tools, independent of intensity. Thirdly: unless we have airtight theoretical models (and how did we get them in the first place?), it is only by trying different things and seeing based on experience which one works best that we know what is best. Imposed uniformity (even of the present best, which put mildly isn't guaranteed) stifles improvement.

5. [will MOOCs disrupt college or high school the most? Why?]

College, because money. In an institutional context where high school is paid through taxation and college is paid via tuition fees, to the extent MOOCs really offer a genuine competitor to college (certification issues spring to mind), the lower price of MOOCs will make them highly competitive.

On the other hand, to the extent that high school curricula change more slowly than college curricula, the fixed costs of producing a course will have to be incurred less often in the high school version, so in a different institutional context (fully public or fully private, but not half/half) MOOCs may be more attractive for the high school setting.

6. [Arnold Kling disagrees]

I forgot his points.

7. [MOOCs: flipped classroom vs. similar to one-on-one]

Cochrane's argument for a more Web 2.0-like MOOC experience, which I took to mean more adaptive to the students' individual needs and skills, suggests that he would like to see some of the benefits of one-on-one teaching in a MOOC setting. There are limits to how adaptable you can make things; perhaps an individualized process aiming at the absorption of the textbook material is the happy middle ground. Certainly video lectures will benefit from being complemented with one-on-one individualized feedback.

8. [Light bulbs last longer; why?]

Fixed vs. variable costs. If you can make a light bulb last longer by making the filament a little thicker without changing the screw cap, electrical contact or glass envelope, you get more illumination per dollar, i.e. you're more cost competitive—up to some point where you hit diminishing returns to stronger filament.

I imagine the magic word is competition: if "they want to sell you more light bulbs" is said with negative tones of voice, this demonstrates that the speaker would rather buy fewer, longer lasting bulbs; and where there's competition the consumers get what they're most eager to pay for.

Michael Byrnes writes:

Here is a wiseguy answer for #8:

"Regulation". CFL and LED bulbs last longer than incandescents, which are no longer sold.

Competition is a good answer for why light bulbs last as long as they do (or don't). However, it is not clear that competition will necessarily lead to longer-lasting bulbs. There are other dimensions (besides how long the bulbs last) on which light bulb manufacturers might compete: price is an obvious one, perhaps also brightness, quality of light, etc. Light bulbs may have a shorter lifetime than is possible because most people would buy a typical light bulb for $0.50 rather than an ultra long-lasting bulb for $50.

John Sallay writes:

7. (And to a lesser extent 6)
I disagree with Arnold's premise. When I listened to the referenced episode I had a lot of questions come to mind. Why is one-on-one better than any other method? What about the things that the teacher doesn't know? How are they going to teach differing opinions effectively when the teacher has their own viewpoints? (If I wanted to better understand the Keynesian way of thinking, this website may not be the best place for it.)

When I'm in a group of my peers, there is a lot of back and forth and discovery that occurs as we work through problems. We throw out ideas, some of which are awful, but we aren't smart enough to know that yet. This back and forth is great in a group or peers and I don't think as easily facilitated in a teacher-student environment.

There is a lot of value in one-on-one education, but the other methods and groupings have value too. I don't think the ideal should be to pick the best and stick with it, but try to draw out the best from each method. That's what I really liked about Dr. Cochrane's course. He allowed the students online time to learn the material, had a group setting for extra discussion and learning, and presumably had office hours for one-on-one help.

I think the great potential of MOOC's is to allow people to learn when they want to,how they want to. I agree with Dr. Cochrane that they still have a long way to go to get there (web 1.0). There is a level of interaction both with the class (via feedback) and with other students that is lacking.

Peter writes:

If moocs give you valuable credentials, how do you prevent cheating? How do you know the student is answering the questions on the test and not his PHD friend?

Alim writes:

[8] The lightbulbs could last longer if we are keeping the lights on longer without switching them on and off all the time. It is not the 'on' state (equilibrium) that wears a bulb out as much as rapid heating and cooling when you turn it on and off. Same with car engines, by the way.

Amy Willis writes:

@John Sallay I love your answer in response to Kling...Although, I wonder if it's the premise of one-on-one that you disagree with, or a different sort of methodological/pedagogical point. That is, what you describe (especially in relation to the instructor not knowing some things) sounds like education as conversation, which appeals to me as a Liberty Fund rep. Like in your peer group, the back and forth may lend itself to deeper discussion. But what's the key variable here- is it conversation or group size? That is, can you have the same sort of deep learning conversation one-on-one that group can? And even if it's only possible with the latter, can the same sort of conversation online that you can face-to-face? Just thinking out loud...

Fred Kavanaugh writes:


The best analogy is a Cadillac - many folks will do just as well with a Buick if they have better things to spend their money on. Students with the greatest, strongest ambition to learn and engage in intellectual adventures may still decide to take the cheap and mediocre coursework that is secondary to their efforts. A course on French history that can fulfill a general curriculum requirement that can be taken online, instead of disrupting the schedule of classes in his or her main schedule may be infinitely more valuable.

On a related point, I think Dr. Cochrane has missed the point of MOOC's. Each student that completes a course of study has to admit that the bottom 10%/25% or 30% of their courses could have been replaced by online courses that would endow them with "awareness" of the material (not mastery, just a decent respect for the depth and capacity of the discipline). A cheap/quick/dirty/superficial alternative to the secondary coursework could liberate both the faculty and the students to do more with the traditional format of courses and slough off the "Rocks for Jocks" or "Stats for Advertising majors" efforts to focus on solid coursework.

The point of technology is not so much to do great things, but to do mediocre necessary tasks quickly and cheaply. 100,000 people can download and read Hayek and Bohm-Bawerk for nothing, but it would take the same amount of time today to write books as insightful as theirs.

John Sallay writes:

@Amy Willis If I were to pick a key, I'd actually go with variety. I think of education as communicating understanding. A one-on-one environment communicates understanding in a different way than discussion among peers. Although some may prefer one environment over another, I don't think that either in inherently superior.

With complex concepts in particular, I find that understanding frequently comes in an "aha" moment. Sometimes it's from the explanation of a teacher, sometimes it's from experimentation on my part, and sometimes it comes from trying to explain the concept to someone else. Each kind of interaction requires my brain to process the information differently.

When I was in high school, a lot of the work was listen to lectures and then regurgitate what I had been told. I didn't really have to think much about what I was doing.

When I was working on my masters degree, I listened to a lecture, did some of the complex homework problems on my own, and then went to a study group. In the study group, I'd need to reconcile what had been taught in the lecture and my own experimentation. I would have to explain my approach to a problem and I was able to hear how other people did it.

I think that if I removed any of those parts, my level of understanding would have diminished greatly. Hearing different explanations, solving problems on my own, and explaining myself to a group were all invaluable.

In terms of doing this in a MOOC, I think it is possible, but technologically we are a ways off. I had a remote student in my study group for one of the classes and it was really difficult to speak equations to each other over skype. I'm sure that hurdle will be solved soon enough.

MOOC's have the potential to bring people together. In the udacity courses that I've taken, you have a wide variety of students from subject matter experts to people who have never programmed before. There has to be some way of matching these people up so that the novices can get help from the experts, and those at similar levels can meet together. Some ways do exist and I'm sure they will improve over time.

There are also interactions available online that would be difficult to reproduce effectively in a traditional course, such as forums.

In response to your final question, everyone has there own preferences for learning. For some people, nothing will ever replace human interaction and being in the same room as those you work with. For many people, the online courses already produce a preferable interaction and for them chat rooms, forums and videos are better than study groups, one-on-one, and lectures.

Yavor writes:

Q8 - The word Russ is most likely looking for is "competition" I think ...

However, I would contend that it should be "preference." Most customers are likely to view light bulbs as commodity - they focus on price and on perceived quality (a major factor of which is longevity), hence (most) vendors compete on offering cheap and long-lasting bulbs.

If customers, for whatever reason, did not see light bulbs as a simple commodity, the bulbs would not necessarily be longer-lasting than in the past. Look at Zara - the Spanish clothing store. Their clothes are known to be very trendy (their designs change bi-weekly) but everyone knows they only last a couple of washes. How is it possible Zara is so successful when their competition offers clothes that last years? In a word - preference.

Drew Yallop writes:

The single word: Innovation.

For the rest I have no comment because I am a life long learner not a student in the sense used here. For me MOOC's provide the background I need to intelligently assess the reasonableness of the many arguments on the Web.

For example, I have taken a money and banking course by Perry Mehrling to better understand the financial crisis, a course on argument and logic from Duke to better distinguish good argument from bad, a course on Randomness and Chance from U of Singapore to better understand statistics.

And of course Econtalk which is a kind of MOOC, I think.

David Zetland writes:

On (2), I'd agree. We're passing knowledge in words instead of writing. The discussion of the material is STILL where the learning occurs.

On (3) & (7), I'd reply that MOOC 2.0 will involve peer grading. I use it for my classes when students writing briefings. Peer grading works BETTER with large classes (=diversity) b/c: (a) students see how others think about the same question/answer and (b) students give feedback on the quality of their grading. (This is because the author is graded by 3 others. Each author then gives feedback to 3 graders and each grader gets feedback from 3 authors. VERY interesting results that, in the wisdom of crowds tradition, beats teachers/TAs...)

Charles writes:

On 8. I like the word 'learning', though I suspect that Russ would contend that 'competition' is what drives the learning to improve on the quality of consumer products.

I don't know why, but I have some prior bias towards disagreeing with Russ' - as imputed by me - sentiment. I think that learning is primary. Competition just seems to be presently efficient at facilitating it.

Michael Morton writes:

1. It takes very large fixed costs to develop the course and once it’s developed, those costs are sunk. Whether one person takes the class or ten million, there are no variable costs to count against any possible revenue the course could generate. It is open for all to see or pay to view. The traditional classroom would demand more classroom space and more instructor time to accommodate a larger audience and so the costs to provide this would go up also.
4. I don’t recall why he said it but I would argue that what is best for one person may not be the best for another. It is trying to assign an absolute solution to a problem that requires a subjective solution. Also, most people act economically whether they realize it or not by shopping for the best product at the cheapest price. Not everyone needs the best trained neurosurgeon to help them with their headaches and Johnny doesn’t need Stephen Hawking to show him how to find the Big Dipper. People can learn at different levels for their needs and MOOC’s can be created/tailored at those levels. One size does not fit all!
5. I think MOOC’s can disrupt those levels but not replace them. If anything, they definitely have the potential to enhance the education experience at a cheaper cost than hiring more teachers.
7. Kling has a good point but the fixed costs of creating one MOOC tailored to the level of understanding for one student is prohibitive. That is why the flipped classroom idea is better in my opinion.
8. Technology is one possible answer. Consumers want their products to last longer and at a cheaper price if possible but are willing to pay for one that claims to last longer. Advances in metals technology for filaments, different gases, and now CFL and LED technology have produced light bulbs that operate with lower wattage, less heat and yet produce excellent light. Competition is also another answer. The competition to find the technology and use it to produce better bulbs for consumers demanding a more lasting product and likewise providing a company a way to differentiate their products is feasible. If I had to choose, I would pick competition as my answer and that competition led to the discovery of better technology.

Amy Willis writes:

@Peter, you're right that cheating is a concern in MOOCs. Cochrane struggled with this in his course. Does the technology exist today that can manage that? Is there a solution?

Amy Willis writes:

@Charles, interesting how you draw out the complex relationship between learning and competition...Which is the primary driver, or is one (only?) a means to an end?

@John Sallay, thanks for your response...I agree re: the importance of variety, and individuality in learning. It will be interesting to watch MOOCs progress!

Peter writes:

@Amy, I can't think of any new virtual technology to prevent this. But maybe a combination with old fashioned supervised tests would work if MOOCs become common. You could have separate companies that provide supervised testing for different MOOCs and different courses. They would have locations spread around the world where MOOCs could send tests and lists of students. They would check ids and make sure no one cheats and then return the tests to the MOOCs. The tests would probably be taken on computers.

You have that today with certifications in the computer industry. Maybe those companies could expand to help MOOCs.

Christian Mayr writes:

At Coursera you can get a "Verified Certificate" for some courses. As far as I know it involves keystroke pattern recognition and doing the quizzes in front of a webcam. Of course this still leaves room for cheating, but it makes it at least a bit more difficult.
However, I think the only way to go for any "serious" certificate is a final exam or something along those lines at some location with a controlled enviroment.

Christian Mayr writes:

@Russ Roberts
In regards to improving the educational value of the podcast, a real forum would be the way to go in my opinion. I don't know if that's possible because of legal or practical reasons, but it would clearly be the best way to discuss and spread knowledge, if you want user involvement. A very easy although external way would be a EconTalk subreddit at

Charlie Weesner writes:

Your podcasts have been an enormous education resource for me, and I thank you deeply for them. I'm not in agreement with all your opinions / perspectives, but I have learned more Economics from listening than I knew existed (in college in the 70's). I struggle to imagine a more-effective way to teach massive numbers of people with moderate levels of interest about self-organizing market principles.

I disagree that "competition" is sufficient to make a longer-lasting lightbulb. Specifically, the consumer has to know and trust that the lightbulb that costs more lasts longer.

When I'm in the store, there are now labels (required by regulation) which I don't trust much. A self-promoted label is even more suspect. Absent those labels, I have no source of information to make the "informed decision" on which lightbulb will last longer. In the presence of the labels, I don't trust the information.

Andrew McDowell writes:

I'm going to have a go at the lightbulb question, from the point of view of somebody who always screws up when analysing motivations - so I'll just list some pros and cons for the manufacturer.

1) I looked at a pack of cheap incandescent candle bulbs and they listed a life of 1000 hours. So if I list 100 hours I'm at a disadvantage, but if I list 1000 hours and lie I'll probably get caught sooner or later.

2) With modern manufacturing I probably can't save much money by cutting quality - at least that's what my six sigma people tell me.

3) I could probably sell really bad lightbulbs via fly-by-night traders and scammy web sites, but there's a big corporate market out there. What's more, they have to pay workers to change bulbs when they fail, so they really care about bulb life - even really cheap poor quality bulbs won't sell there.

Thomas A. Coss writes:

This "certification" issue strikes me as one ripe for innovation. I too have greatly benefited from EconTalk, it has directly impacted my business, enabling me to use them in communicating concepts and directions I could not have done on my own.

Though I've not missed an episode in over three years,logging in nearly 200 hours listening, reading comments and responding. Today, on the Friends of EconTalk on LinkedIn, we have over 3,000 subscribers from all over the globe discussing all matter of economic topics affecting real people, many students of Economics, many asking probing questions affecting their local communities.

Perhaps there is space for Fellow's of EconTalk, or some other means by which those less familiar with what goes on here to attach some value to the learning and it's meaning to their respective enterprises.

Jay Wysocki writes:

You mention the difficulty of machine grading typical exams. How about crowd sourcing the grading of essay questions by former (or current?) students? Maybe you give the graders credits for future classes or even pay them.

Katie L writes:

4. To argue that "everyone has to have the best" in education is to allow the perfect to be the enemy of the good. Education solutions don't have to be better than the very best option, they just have to be better than the option they replace. Not everyone has access to "the best" right now.

5. I think MOOCs have greater power to disrupt college education, for two reasons. College costs more than high school, and the college learning environment is set up for students to bring their own motivation. MOOCs can augment the high school classroom or they can provide high quality resources for self-motivated high school students who are already outside traditional brick-and-mortar school settings (homeschoolers, adult students, and students in online charter school programs).

8. Competition

Calin G writes:

2) One aspect not discussed in John Cochrane interview, and I think makes MOOCs better than textbooks, is peer grading. Many people hate grading (I was a TA myself) and consider it a necessary evil of MOOCs, but I found it engaging and interesting in the two MOOC courses that I have completed.

Coursera already has in place a system for peer grading. For example, Scott Klemer’s course “Human-Computer Interaction” has weekly assignments in which the students need to answer open-ended questions. The way the course is organized, each student has to grade 2 assignments which are compared with a TA grades (this is a calibration step) and then grade another 5 assignments. So, by the end of the week I would read the answers of seven people. This was a great chance to understand of the material at a level way above what I generally got by reading the text or watching videos. It made me realized that I sometimes misunderstood the material or missed nuances.

Watching lectures on video is often boring and seems stalled. Reading a different point of view or getting feedback makes it more fresh and exciting. It is also hard work and the quality of the grading is not always great. To some extent the forums accomplish the same thing, but only a minority contributes to the forum while grading requires everybody to participate.

I can easily see how EconTalk can move from a great podcast to a great MOOC. Almost everything is in place: class on Monday morning (with invited guests), extra readings and assignments published on the web-site. Adding a peer grading system and a symbolic certificate will actively engage many more listeners and get the entire experience to a new level.

Harry Chiam writes:

I have an observation about how much learning you get from a MOOC. Learning occurs when you acquire new knowledge, apply that to reinforce the learning and get feedback.

MOOCs are an ideal way to achieve the above because of the machine learning nature of many courses and the ability for students to acquire and feedback their knowledge.

Having said that, many MOOCs are so poorly written that the feedback loop is not challenging, so learning is not great, or there is no feedback loop at all.

I can't speak for Udacity because I have not done their courses. However, when you look at some of the early MOOCs from MITx, they are very, very good at this learning process, feedback and assessment.

Given this, and the rigourous nature of courses from MIT generally, I think it won't be long before rigorous MOOCs from good universities are considered for credit; certificates and credits towards a degree will surely follow.

What does this mean for the other providers?

To other MOOC providers, shape up your courses so that they are at least as good as some of the MOOCs from MITx. A poorly written MOOC teaches nothing.

To traditional universities, I think some of the better ones will gear up for enrolments that approach 100,000 maybe 500,000 and they will be fee paying students.

Sadly, there will be more than just a few universities left behind. Universities are full of smart people. They should be figuring out the next big thing they can do in higher education.

Comments for this podcast episode have been closed
Return to top