What this covers
John Cochrane from the University of Chicago discusses his experience building and teaching a MOOC with host Russ Roberts, examining what online courses can and cannot do compared to in-person teaching. Cochrane frames the conversation around two core constraints: the economics of scale and the irreducible human elements of instruction. He argues that MOOCs operate under zero marginal cost economics—adding one more student costs nothing—but require enormous upfront effort to produce and refine, creating both unlimited reach and an unsolved monetization problem. The substantive disagreement centers not on whether MOOCs work as content delivery, but on whether they can replicate what actually makes teaching effective.
The conversation covers what Cochrane identifies as teaching's irreplaceable core: knowing the wrong answers to diagnose student misconceptions, and providing genuine feedback at scale. He contrasts this with what current MOOC quizzing actually does—largely machine-gradable multiple choice that destroys learning by allowing guessing. Cochrane positions the MOOC as textbook rather than teacher replacement, suited to flipped classroom use where students consume online material and meet for discussion. The real disruption hinge, he argues, is certification—whether MOOC credentials become accepted as equivalent to degrees. He examines which teaching scales well (codified high school material, Algebra) and which resists (seminars, frontier research thinking), and sketches how superstar economy dynamics might remake academic labor if online teaching takes hold.
Cochrane argues that MOOCs function less as a replacement for great in-person teaching and more as a modern textbook—a zero-marginal-cost foundation of materials whose disruptive potential hinges on solving certification and replicating the social feedback that good teaching provides.
- MOOCs have zero marginal cost but high fixed cost, like other technologies, which shapes both their reach and their monetization problem
- The hard, irreplaceable part of teaching is knowing the wrong answers and giving feedback, which current MOOC quizzing cannot do well
- Disruption of college and high school depends on when MOOC credentials count as certification equivalent to degrees
MOOCs currently lack interactive feedback mechanisms essential to learning, making assessment crude and unable to diagnose misconceptions at scale.
- The most valuable learning—open-ended puzzles, proving a theorem, defending an answer to skeptical peers—cannot be put into machine-gradable form, and converting such questions to multiple choice destroys the learning by letting students guess from the framing rather than reason, which is the core limitation of current MOOC assessment.
“I couldn't figure out how to ask that in machine-gradable form. Or: here's a theorem; prove it. You can't ask that in machine-gradable form.”
- Much of being a teacher is not knowing the right answer but knowing all the wrong answers—being able to instantly recognize a student's specific misconception and correct it—which is the experience hardest to replicate on a MOOC and the core of what distinguishes a good classroom.
“A lot of being a teacher is not know the right answer. It's knowing all the wrong answers.”
- A pragmatic way to grade an open-ended answer at scale is keyword detection—Cochrane gives credit if the word 'beta' appears in the answer to a finance puzzle, and Russ similarly looks for a single key word in answers to his lightbulb-longevity puzzle—revealing both a workable mechanical proxy and the crudeness of current assessment.
“the way I grade it is if the word 'beta' appears in the answer they get credit, and if it doesn't, they don't.”
- The two essential parts of education are the social structure that keeps you going and the feedback that lets you straighten out misconceptions and gauge your standing against a standard of mastery; formalized quizzing, once well-written, can scale feedback to 100,000 people at zero marginal cost, though current quizzing software is one of the weakest links.
“The two parts of education are the social structure that keeps you going and the feedback.”
- The ideal MOOC quizzing would recognize a specific wrong answer or wrong equation, identify the misconception, give a hint, and let the student retry—routing them back through the decision tree—yet the Coursera software Cochrane used was so primitive it couldn't even let a student retry a question after a wrong answer, though this is simple programming likely to arrive in version 2.0.
“I know all the wrong answers... I say, 'Wrong; here's a hint.' And you get to go back into the tree and figure out which branch you should have gone down.”
- Teaching one-on-one means not lecturing but listening and probing—posing puzzles, hearing the student's answer, asking 'is that really how it works?'—to assess competence and surface misconceptions, exactly as a flight instructor does, which is the diagnostic core that is hardest to deliver at MOOC scale.
“A lot of being a teacher is not know the right answer. It's knowing all the wrong answers.”
Certification and employer recognition of MOOC credentials is the foundational barrier whose removal would enable cultural and economic adoption.
- The cultural resistance to MOOC-based education is downstream of economics, not an independent barrier: people will not embrace home-based MOOC education until the certification problem is solved so that the credential reliably signals competence, after which the culture will follow.
“Let's be economists for a minute. The culture follows the economics.”
- A large part of what universities actually do is provide a brand—hiring a Chicago or Harvard graduate works because the institution carefully guards its reputation that its students are good and certified, even though universities may do this job less well than they claim, which is precisely the challenge MOOCs must meet to monetize.
“Why do you hire somebody with a U. of Chicago or Harvard degree? Well, that's a brand. And it's a brand that is pretty careful about its reputation.”
- The hard question for MOOCs disrupting college and high school is certification: when MOOC certificates (or curated groups of them with proctored exams) start to count for college credit and be accepted by employers as equivalent to degrees, that is the step that enables disruption.
“Let's be economists for a minute. The culture follows the economics.”
- A physical business school's value lies less in instruction than in selectivity (being good at picking smart people), employer connections via a placement office, and an alumni network—so a disembodied online MBA could be viable if it replicated selectivity (e.g., GMAT thresholds) and the alumni network, which already forms organically among MOOC classmates.
“A bricks and mortar business school is selective about who they let in... We have connections to employers. We have a fantastic office that gets them jobs. And we have an alumni network.”
MOOCs work well for codified, well-pedagogy-studied material but struggle with frontier-level research or open-ended learning at the discipline's edge.
- High school education is by and large codified material taught by professional pedagogues rather than researchers, so it has clear economies of scale; the research-university model where professor and student think through fuzzy-edge questions is harder to bring to MOOCs, but in fact little of what must be taught—even in introductory university classes—is at that frontier level.
“High school education is material that is by and large codified... We do not ask high school teachers to be researchers... And there are certainly economies of scale.”
- Material that is cut and dried with a well-studied pedagogy (a pilot's license test, Algebra 101) is easy to put online, whereas knowledge closer to the research frontier where the pedagogy is less settled is much harder to deliver effectively via MOOC.
“It's easy to put online: How do you pass your pilot's license test? Material that's very cut and dried... Putting online things where the actual knowledge is closer to the research frontier and it's less well-studied, that's going to be harder.”
- There is a danger in discussions of education and health care of insisting everyone get the absolute best; it would instead be a huge step forward if the majority of high school students simply came out knowing calculus, basic grammar, how to write sentences, and how to program—even without a deep appreciation of literature.
“There's this danger in discussing things like education, health care, to say, everybody has to have the absolute best. But if the majority of high school students could come out knowing calculus, knowing basic grammar... it would be lovely”
- One great thing MOOCs are doing is rigorously studying pedagogy—what works and what doesn't—whereas universities, despite being staffed by empirical social scientists, do a poor job of empirically examining how well they themselves teach.
“At universities we are pretty bad at actually self--we are supposed to be empirical social scientists but we don't really look at how we do that well. And the MOOCs are looking at what works and what doesn't work”
Online video with graded quizzes outperforms traditional reading assignments in student preparation and engagement with course material.
- Assigning short graded videos with quizzes produces far better-prepared students than assigning readings, because young people today don't read and respond to blank stares when asked to discuss a reading, but show up well-prepared when required to watch 5-7 minute videos and complete graded quizzes by a deadline.
“if you give a reading assignment and you say, Come in on Monday and discuss it, they look at you with blank faces. If you say: Watch a set of 5-7 minute videos, do some multiple choice tests... Lo and behold the students come in Monday morning just very well prepared.”
- Business school formats of three concentrated hours once a week (or condensed multi-day blocks) strain the brain's ability to absorb new material, and online technology improves this by letting students watch lecture snippets daily, quiz, and discuss on forums, then meet to review—producing a much better learning experience.
“it's a big strain on the human brain to get 3 hours of new material once a week... Now, think of that setup with online materials, where every day you watch a snippet of lecture; you do a little bit of a quiz”
- Even without going fully to MOOCs, using online technologies for regular instruction improves classroom teaching: condensed video lectures let students pause, rewind, and review hard material without feeling they are holding up the class, and free the professor to spend live time on deeper discussion and advanced material.
“this is hard stuff, and I loved being able to stop and rewind it, go back, check something, and not feel like I was the dummy who was holding up the class.”
- We had the technology to play videotaped lectures for 20 to 30 years and it never really took off, which proves that a successful MOOC is more than just great lectures—the lectures alone are insufficient.
“We had the technology to play videotaped lectures for 20, 30 years, and that never really took off. So, a successful MOOC is more than just great lectures.”
MOOCs require sophisticated production, technology adapted to pedagogy, and face a monetization puzzle created by zero-marginal-cost delivery.
- One monetization path is that MOOCs make teaching like other superstar-economy fields: aggregating many small payments from a large base lets a few superstar teachers earn a lot of money while everyone else becomes a glorified TA earning little—but this conflicts with the zero-marginal-cost reality that makes charging hard.
“you can put a lot of pennies together and get a superstar economy... this will make teaching like so many other things where a few superstars earn a lot of money; and other people are there at best to be glorified TAs”
- Current MOOCs are essentially Web 1.0—one-directional web pages with rudimentary discussion forums; the challenge and opportunity is to move into Web 2.0 where social and discussion interaction happens remotely at scale, which would deliver a much better experience, leveraging that young people are already adept at remote communication.
“MOOCs are now very much Web 1.0. It's one-directional. It's web pages... the challenge and the opportunity for it is to move into Web 2.0, where that kind of social and discussion thing can happen remotely on a fairly large scale.”
- Rather than displacing professors, a MOOC functions like a textbook: a set of tools and materials that form the foundation for another faculty member's flipped classroom, where students take the MOOC for the basic material and then meet for discussion and exams—just as the printing press did not eliminate lectures.
“the MOOC experience is not just a complete substitute for taking a class. It is also a set of tools and materials that are the foundation for somebody else teaching class. Much the same way a textbook is.”
- Producing an effective MOOC is not a matter of turning on a webcam; it requires a support team that understands both the technology and the pedagogy—preparing short videos with embedded quizzes—because the technology must be adapted to teaching, not just operated.
“to put together a good MOOC, you really don't just turn on your laptop webcam and start talking. It really helps to have a team of people who know what they are doing, who understand the pedagogy of it”
- With a hundred competing calculus MOOCs, the economic rent may flow to whoever curates and certifies the good ones—issuing a 'degree' for completing a vetted list—and the curator could split that rent with the professors, while the alternative model is universities giving content away free to win global brand recognition.
“Maybe the economic rent will come to the person who says, Look, I've studied all of these and here's the 5 good ones. I give you the Russ Roberts Degree because you've taken this list of MOOCs”
- Like all technology, MOOCs have zero marginal cost—adding an extra student costs nothing—but a very large fixed cost: it is far more expensive in time and effort to produce and revise a MOOC than to assemble notes and lecture, which both enables unlimited reach and creates a monetization problem.
“MOOCs are, like all technology, an innovation that gives you 0 marginal cost and a very large fixed cost.”
MOOCs threaten second-tier universities offering large lectures with weak feedback, not elite colleges or specialized in-person instruction.
- Despite fears that MOOCs threaten college, their current market is primarily adult education—people not going to or interested in going to college, such as industry professionals, alumni, and faculty wanting to deepen knowledge—so they are not yet a threat to universities' core business.
“mostly MOOCs right now are adult education. They are people who are not going to college or interested in going to college.”
- For students who lack access to good faculty—out in the middle of nowhere with weak teachers—a MOOC from a great professor can provide a much better experience than a third-rate college education, even if it never matches a top in-person professor.
“If you are out in the middle of nowhere, and your faculty members aren't that good, well, I think that MOOC provides the great experience--it can provide a much better experience than a third-rate college education.”
- What MOOCs really endanger is the 150-person lecture class at a second-rate university with multiple-choice quizzing and a barely-English-speaking TA—because a MOOC can do that better—while the 20-person class with great feedback from a great professor will survive; criticizing MOOCs for not matching the latter is like faulting Southwest Airlines for not being a private jet.
“what the MOOC really endangers is the 150-person lecture class at a second rate university... The 20-person class where you get great feedback from a great professor, that will still be there.”
Large initial dropout rates reflect rational free-sampling behavior, not MOOC failure, and social structures must replace classroom incentives.
- There is far larger demand to passively watch MOOC videos and take some quizzes than to do the hard work of a rigorous course, evidenced by 37,000 sign-ups dropping to 4,000 video-completers and only 300 who did the full problem sets and exam.
“I got 37,000 who signed up. But in the end 4000 were watching the videos. And about 300 took it all the way through and took the exam.”
- One key function conventional education provides that MOOCs lack is a social and precommitment mechanism: being on a schedule with the threat of a bad grade keeps students going even when motivation flags, so MOOCs need to replicate this to substitute for traditional schooling.
“one of the functions that conventional education does is it's a social and precommitment mechanism... The threat of the bad grade keeps you to keep going.”
- The large initial dropoff in MOOCs is not a flaw but appropriate shopping behavior—because signup is free, tens of thousands sample the first lecture and correctly self-select out when the material isn't for them, so completion rates should not be treated as a failure metric.
“There is a quick falloff as the shoppers leave... The whole thing about MOOC is it's free to sign up.”
MOOCs enable unprecedented customization and access to specialized topics unavailable through brick-and-mortar institutions at any price.
- MOOCs may permit far more customization than physical universities: because a brick-and-mortar school can't run a class for the handful of people worldwide interested in 13th-century Iranian poetry, MOOCs can offer courses both more specialized in topic and more tailored to the individual's interest and level, so the right introductory finance class exists for each learner.
“if you want to teach a class in early 13th century Iranian poetry or something of that sort, there are 150 people around the world who might be interested in that. So one thing this might do is to give us classes that are both more specialized in a topic and more specialized to the person.”