Coursera

This isn’t really the main topic of your post, but I’d love to hear more about what you’re going to be doing on the philosophy side of things with the Gopnik book.

the online world arrives, you offer your content online for free, customers come to expect online content to be free, and suddenly you’ve given away the store.

Or the customers become the product.

If you want to provide adult enrichment for a lot of people this could work. One way interactions are not the wave of the future in teaching, however, and doing actual work and getting critiqued on it is an important component of real learning. Neither are served by online approaches, which are to actual education as “say 1 for reservations” is to actual conversation.

Finally, while the idea that providing high quality courses for free to any student in the world with an internet connection has all kinds of obvious appeal, we should also worry about sending the business model of higher education down the path of the newspaper industry: the online world arrives, you offer your content online for free, customers come to expect online content to be free, and suddenly you’ve given away the store.

You might be giving away the store to people who wish only to pursue learning for self-enrichment, but that’s not much. To turn that educational advantage into material advantage you generally need a degree, and the price of certification (as opposed to simple learning) will always be under institutional control. The occasional autodidact who could pass the bar exam without a law school credential is not allowed to practice law. Internet-reliant students who can demonstrate all of the skills and none of the certification of traditionally educated students will likewise have few job seeking advantages over peers who have neither certification nor education.

“This multi-modal, multi-media interaction does not lend itself at all well to current online education technologies, and nor will it ever if those technologies remain focused only on a gradgrindian broadcast-of-information model of education.”

You should try taking some classes at Udacity. After the first few lectures, you will spend more time answering quizzes and programming than watching lectures.

Coursera has less interactive quizzes (it treats the audience more like adults rather than undergraduates with ADD) but the technology for frequent quizzes is certainly there.

Also, while it is not implemented in currently existing online learning, there is about zero technical difficulty in implementing an “adaptive” lecture, where the material changes in difficulty depending on how the student answers an interactive quiz. (The model here is the adaptive testing from various certification exams.)

So, the complaint that online lectures are not or can not be highly interactive is simply detached from reality.

Another area where we can expect a lot of flailing around for a few years until the right model is found, is the organization of student study groups for online learning. There is a social aspect to learning, and for some people this aspect may overshadow others.

But once such a model is found, it can be replicated in any community — including communities that would never have had access to a quality education otherwise. On a shoestring budget you can replicate a working organization of study groups, you can probably even find good TAs. Now it’s also possible to peek into how world class professors organize their thoughts on difficult material, for free.

For online learning to really take off, these courses need high quality certification (meaning, certification that employers take seriously) but as of now only rudimentary certification exists for Coursera and Udacity. No technical difficulty exists for good certification though — it has been done for years in other contexts.

Finally, while the idea that providing high quality courses for free to any student in the world with an internet connection has all kinds of obvious appeal, we should also worry about sending the business model of higher education down the path of the newspaper industry: the online world arrives, you offer your content online for free, customers come to expect online content to be free, and suddenly you’ve given away the store.

Commendable candour!

I’d have thought maths would be the perfect counter example – it’s one of the very easiest things to teach yourself. The way people learn is through problem sets. You try and problem, there’s a correct solution, and you can easily check your own work and see where you went wrong.

Next up, Leederick explains how one learns to be a professional tennis player by batting a ball against wall.

Anyone who does not believe in the value of individual work in math (work on the lines described by Leederick) was probably not good in any math class except remedial math.

As for whether basic math and remedial math can be taught successfully online, millions of views of Khan Academy’s math videos suggest that people find them rather spectacularly useful.

I think its a useful addition. But probably more so for near auto-didacts then Joe Averages. I did try to learn graduate chemical kinetics from some online lecture notes, but I didn’t have the confidence that I wasn’t missing perhaps some key insights to actually apply it professionally. I recommended my employer hire someone with actual domain expertise (which they eventually did). So I think stuff with maths type challenges can be a bit problematic, unless the customer has someone with domain expertise to consult with. Otherwise too many areas of knowledge seem to require the sequential build up of a base of knowledge, get one or more foundational skills wrong, and later on you can find yourself hitting a bottleneck. I think thats where having a mentor comes in. Someone who can tell you the hard news early on “you misundertand foundation skill X, and if you don’t get it your whole project of trying to become an expert at Y will founder later on”, and then come up with some way to overcome the issue. I suspect a lot of complex subjects may work this way, and getting through them without botching some of the foundational understandings may not be particularly likely without expert interaction.

So I think they are a great supplement to a more traditional classroom experience. I hope to get my twins who will be between Sophmore and Junior college level this summer to try a course or two. Maybe they can gain a leg up on the rest of their college careers? Maybe even good students can get 25-50% of the needed graduation credits via this mechanism. That would cut the overall cost of an education somewhat.

The challenge for online courses is similar to the challenge for large (>100 student) classes, but more so. How do you get the students to actually think for themselves as opposed to absorb and regurgitate? Or is there going to be a different standard for online courses, where absorb and regurgitate is the end goal?

The really disruptive education technology, enabling anyone sufficiently motivated and disciplined to learn anything at all, without a teacher, was the book. As the discussion above suggests, though, books are only a partial solution, and online courses are likely to make only an incremental improvement.

Garymar, that’s exactly what I plan to do over the summer.

With all this discussion of mentors, I thought it would be useful to point out that the internet is positively crawling with math and physics students (I suppose that’s what they are, anyway) who seem to legitimately enjoy helping anonymous internet users teach themselves math and physics. One can learn a substantial amount by just lurking around internet communities, and it has been my experience that if you do hit a brick wall, there are plenty of people who would be more than happy to help you get around it.

@Barry
I’m sure “engels” can explain himself further what he meant when he attacked Leederick’s rather sensible statement.

It is intuitively obvious that the $50K/year gold-plated model for undergraduate education is inapplicable to the world population. Technological disruption of this unsustainable model is inevitable. The rear-guard action being fought against the new model is best explained by Marxist theory: vested interests have created educational dogmas that pronounce the new model unsound, unproven, un-etc.

Just as the average income of an American can no longer pay for the average cost of a new automobile, this limited income eventually will not be able to afford steadily rising bricks-and-mortar college tuition. Excruciatingly expensive four-year colleges are like the last grand ocean liners. They are magnificent to behold, and offer every imaginable convenience, but there is something new in the air.

Just as the average income of an American can no longer pay for the average cost of a new automobile

Average new US automobile costs $28,400; median household income is $46,326.
In 1960 it was $2,600 and $5,620. Difficult to see how the one situation is affordable and the other not…

Alex K. 04.19.12 at 1:15 pm

” @Barry
I’m sure “engels” can explain himself further what he meant when he attacked Leederick’s rather sensible statement.”

Leederick is basically talking about learning math on one’s own. Engel’s is (from my interpretation) pointing out that that’s far from sufficient – if one doesn’t have feedback and help from other people, one is going to have an extremely hard time learning math from just problem sets. As somebody pointed out, learning proofs is hard, and proofs are frequently open-ended, so to speak.

IMHO, when somebody talks about learning a serious, complex intellectual skill on one’s own, they are usually wrong.

@32

Difficult to see how the one situation is affordable and the other not…

2012 ratio: 28,400/46,326 = 61%

1960 ratio: 2,600/5,620 = 46%

Do you see now? The delta is about $7,000 (if a car cost 46% of today’s income, it would cost about $21,300) . Of course lots of other middle class expenses, like college tuition and health care have also gone up faster than income, so that $7K is rather hard to come by. You really should read Elizabeth Warren.

@31

Forgive my inelegant prose. The argument is immutable because the numbers don’t change. The costs of labor-intensive university education are already beyond the reach of most of the world’s population and are becoming unaffordable even for the US middle class. That is why student loans are becoming the next giant financial bubble.

What bugs me about responses like peterv’s is that (1) it’s not an accurate representation entry-level coursework almost everywhere, but the statement assumes that most students really are receiving that, and (2) it alleges that it’s possible to establish that deep a connection with every student in your classroom, and I don’t see that happening in a massive lecture hall. It’s basically alleging superlative abilities, and while peterv may have those, peterv should also be well aware that a huge chunk of the lecture coursework being presented to students is not expert faculty lecture, it’s lectures presented by graduate students like me who are doing entry-level research and who probably have received little to no training in the craft of teaching (out of three dozen plus grad students I’m one of two with any awareness of pedagogy or prior experience teaching). It really does sound like a description of upper-level courses where you have at most a couple dozen students and where those students have the core skills competence to enable a variety of more sophisticated interactions, but the comment was offered in the context of dismissing and disparaging online lecture material that is clearly intended to predominantly replace 100-level gen ed content coursework, for most of which it’s an unrealistically flattering description.

One very promising feature of recording lectures is the opportunity to recover and preserve teachable moments. Often enough, one student asks a bright and unexpected question that the whole class benefits from receiving a response to. During that time, everyone in the class is basically a spectator, so the types of benefits they receive from that particular exchange should be easy to make available in video. And for that matter, even if the video is only made available to a restricted set of students, the instructor benefits too. Ever get asked an insightful question, for which the wording matters, and forget the details before there’s a chance to write them down? Ever come up with an inspired impromptu response, and want to be able to offer that insight to future semesters, but you can’t quite remember the exact context?

The flip side of this is, of course, that videotaping lectures is devastating, and destroys — students in videotaped courses will lose the quasi-privacy of the classroom and become publicly responsible for things they say in class, which could have a chilling effect on discussion, and could also encourage students to treat their class as a public platform and put on a show. That’s a lot of anxiety to place on people who are there to benefit, but who become effectively conscripted as actors serving in the benefit of others watching the performance. For students who already feel nervous about wasting their peers’ time with a lame question, the “is everyone going to hate me for asking this” factor grows by an order of magnitude. It makes the classroom a less safe place to be wrong.

As someone who just came through what was supposedly a “1st tier” school (Mcgill), I really think Salient is spot on. My math courses basically required one to be an autodidact since the professors mostly just ignored the 50 people (in a 3rd year course, for 1st years its well over 200 at this point) in the class and wrote out proofs from the textbook. On the social sciences side large class sizes made any actual interaction practically impossible. As Salient said, the vast majority of students are just spectating.

What I want to add is that making the quality lecturers available to everyone would have been much appreciated. If the product is basically standardized and homogeneous anyway (all stats courses teach the same thing, ditto for undergrad chemistry) then why wouldn’t we try and make sure the “spectating” component is coming from the best sources available? There are a lot of terrible lecturers out there, I know I’ve had my fair share. Frankly, I expect in many instances I would have learned more watching lectures on Coursera than I did from actually taking the course on campus.

I realize it would be better if we didn’t have large class sizes, but with cost control practically becoming the zeitgeist I don’t see that changing any time soon.

@38: And yet labor costs have not escalated, suggesting that other factors are much more important drivers for tuition. You apparently have an embedded assumption that we need to mechanize teaching because the labor costs are so high. What is this coming from? Because all of the data that I’ve seen points to capital construction costs and administrative overhead as the main factors, with both student to teacher ratios and faculty salaries dead flat over the past 30-40 years. University costs aren’t behaving the same way across the world that they are in the US, and public schools have risen faster than private because we’ve decided not to support them with taxes.

Analogies can be deceptive. Some tasks lend themselves well to automation and others don’t. We can make airplane reservations efficiently with computers, but it doesn’t follow that we can teach as efficiently with them. Some people do indeed learn well, and you’d get more out of these courses than reading a book. But others simply don’t, and there is no substitute for engaging with an actual teacher if you’re interested in learning.

Marc @ 43

Would you have a link for that data? In particular, my impression was that “And yet labor costs have not escalated” is not true when looking at the ratio of “college teachers pay”/”college graduates pay”.

Let’s look at the case of the retail pharmacist in the US.

For foreign students of pharmacy there is a track to get a licence to practice in the US: you need to graduate from a pharmaceutical school that the US recognizes, you need to pass an equivalence exam (FPGEE), pass a TOEFL exam (which can be tough: there are stories of native English speakers failing to get the required speaking score — although, they were from Australia) work as an intern in a real pharmacy and pass the NAPLEX, which is an exam required from the US students as well.

Now some of those foreign pharmaceutical schools can be really poor — online lectures and tests from the best pharmaceutical schools in the US would be a definite improvement.

Yet, if the foreign student is good enough to pass the certification tests (FPGEE, NAPLEX) then he or she is perfectly qualified, both legally and in actuality, to work as a retail pharmacist.

And what is the difference between a foreign trained pharmacist who passed all exams and a US trained pharmacist? About $200, 000 in student loans.

I don’t understand this debate. On the one hand, there are said to be some students who cannot understand math textbooks written at an appropriate level, or who cannot understand mathematics lectures delivered at an appropriate level, in order to do the problem sets–or who can’t understand them without above-average levels of handholding–yet who appear (I assume) to have the prerequisites. Certainly there are students who have managed to pass math courses with As and Bs (and observably only seem to have cheated enough to get their grade up a half-grade, or a grade at most), who seem able to talk about the material yet don’t seem to be able to do even the easy parts of the problem sets.

But on the other hand, there is the argument that people who can learn from textbooks (very often, math textbooks in particular) are learning something that is really not university-level stuff, and is wrong. There are lots of good arguments about how elementary level math instruction is misleading about upper-level math, or about how what professional mathematicians do isn’t well conveyed to people taking even second-year calculus. However, those good arguments don’t add up to, “if you were an A student in university math classes, learning largely from textbooks rather than lectures, the chances are very high that you know less about math than average.”

In computer science, back when I took courses, even with good textbooks at an appropriate level, the lectures could be very helpful, and there are a lot of textbooks that aren’t at an appropriate level, the way calculus textbooks, say, usually are. But again, this is very far from saying that interaction with a teacher over and above grading problem sets is important for learning math (it’s important if you want a grad school reference, presumably, but that’s not the same thing).

bianca steele 04.19.12 at 7:58 pm

“I don’t understand this debate. ”

Standard netlibertarian: “I learned X by myself, so therefore everybody else can, and teaching is unnecessary”. Not: the person making the statement doesn’t need to actually know X at any serious level to make this, and having actually had massive help doesn’t interfere in making this statement.

Is Alex K a Coursera employee?

There’s another argument that may be feeding into it, as I said, about elementary instruction. I have no idea how fair that feeling is. I’ve overheard mothers of older children complaining, but usually not that their kids aren’t being drilled enough, more because the kids are being asked to think abstractly or discuss their reasoning in words. And there’s the whole debate over how elementary math should be taught. But that’s specific to math, and the more general argument engels and leederick referred to isn’t.

straightwood: that’s the *average* cost of a car. It’s perfectly possible to buy excellent, serviceable cars that will last a good long time for less than $21,000 in 2012.

Even looking just at the averages, there’s a big difference between a 1960 car and a 2012 car. The average car bought today is expected to run relatively trouble free for 10 years and 100k miles doing nothing more than standard maintenance, and plenty of cars in the $20k range will do better than that. I can’t remember the last car I got rid of with less than 150k on it, and even then, they were salable to somebody who could do some repair work.

Very few of the cars built in 1960 could be expected to last that long without exacting maintenance and lots of repair. You may remember that they figured out how to make cars that last a long time around the mid 60s (lincoln, valiant/duster slant-6 engine), and most manufacturers decided they were a bad idea and stopped making them when the market consequences became clear (i.e. many people keeping their cars for 10-20 years). The japanese manufacturers were the only ones that kept making cars like that, needing something to distinguish themselves — once they took over enough of the market, the Big Three had to respond and by the late 1990s long lasting cars had become the norm.

Given how much longer the 2012 car can be expected to last, I suspect it’s actually cheaper than the 1960 car.

@46: A good quick summary is at

http://chronicle.com/article/faculty-salaries-barely-budge-2012/131432

The actual study is at

http://www.aaup.org/AAUP/comm/rep/Z/ecstatereport11-12/

There is more substantive historical work on student to teacher ratios as a function of time that I can dig up if relevant. But the bottom line is that they’ve barely budged. You can also use the mean salaries – even if you add in a 50% fringe and assume no contingent faculty, or an average assistant professor with an average 16:1 ratio, you get ~6K/student/year in direct instruction cost. Comparing this with current US tuition and fees is pretty stark.

“In effect, a common argument is that the skills of a few super-teachers distributed online will supplant inferior teachers in traditional models, and that we can’t afford to pay all of those people anyhow. This is, I thing, very wrong on both counts.”

If you’re right, then there is absolutely no reason not to encourage the development of such online learning initiatives. It may help some people and it poses no threat to the current system. It also gives an option to those that do not have access at all to quality education or those for whom such education is prohibitively expensive. Let a thousand flowers bloom.

It’s been possible both to broadcast lectures live and to distribute video recordings of lectures for decades now and yet the promised eclipse of classroom instruction hasn’t occurred.

Are there arguments in favor of recorded online lectures that are substantively different from arguments in favor of selling taped recordings and doing grading via mail? That is, arguments that don’t boil down to “But…it’s on the internet!”

“The point of the new crop of online offerings is that they add certification to the free information. This would be the most substantive difference, if implemented well.”

Agreed. (of course, ‘if implemented well’ is a rather large caveat)

“Whatever certification Harvard decides to grant to folks streaming their online lectures, I think we can be sure it will be of negligible value compared to an actual Harvard degree.”

I don’t think online degrees are any threat to Harvard and equivalent. Ivy league universities sell a luxury good (among other things, direct access to say, Nobel laureates) and they will do just fine.

There are plenty of students (perhaps the majority) who go to college in order to make a living. Earlier I gave the example of pharmacy schools: graduates of rather poor non-US schools are able to practice as pharmacists if they pass some standard tests. Conditional on passing those tests, these people are still good at their profession.

If students would be able to use online courses and certifications to qualify for taking those standards tests, it would mean less financial pain and more opportunities for a good life for those would not have had a chance to it.

Computer science is another area where the difference between an online degree and a brick and mortar degree is potentially minimal — yet more opportunity and less financial pain.

Computer science is another area where the difference between an online degree and a brick and mortar degree is potentially minimal—yet more opportunity and less financial pain.

Just for the record, in my opinion the importance of online courses for computer science is that this is an area where continuing education and self-education are often expected, where people who are working in a field without having important coursework (like an Algorithms course) can often benefit from picking it up later, and so forth, where postgraduate “certification” is nonexistent in some areas (and arguably should be nonexistent, though there’s a wish in some quarters for a way of telling whether someone really knows what they say they know), where there is a strong interchange between industry and academia in both directions, and where the places that offer advanced instruction in the more cutting-edge areas are limited. Breaking this down to a choice between “traditional” teaching (which is what? Jedi Master stuff?) and “we don’t need no stinkin’ school bond issues” makes no sense, unless one or the other of those two things is all one really wants to talk about.

“The really disruptive education technology, enabling anyone sufficiently motivated and disciplined to learn anything at all, without a teacher, was the book. As the discussion above suggests, though, books are only a partial solution, and online courses are likely to make only an incremental improvement.”

Books are good, but the problem with books is that there are an awful damn lot of them, and reading them all is impossible. What one really needs is a syllabus – something from someone with a reasonable amount of respected knowledge in the field who can say “read this book and this book, a few chapters of that book, and these critical articles.” A few problem sets or questions can help as well.
The prime advantage of some of these online courses is that they offer some structure and guidance for what to learn. Of course, one can also find myriad syllabi that professors post online, and hang around the right blogs to get much the same thing.

@71: Most of the faculty increase was a recovery from a severe drop in the mid 70s – mid 80s. The level is almost identical to that in the early 70s.

Follow up to Marc @74: do you have a link for that? I want to expand my list of go-to studies on these topics.

“The prime advantage of some of these online courses is that they offer some structure and guidance for what to learn. ”

Except that there are ‘an awful damn lot of them’. :)

@ 77 and 78,

Well, true – one can start with book A and proceed onward with footnotes and bibliography and “recommended readings”. And it does take some filtering and knowledge to select the right sylabus as well. But I find it useful to be able to look at what is assigned for an MIT class in Real Option Analysis, or a graduate level class in economic history at U Mass Amhurst. For a variety of reasons I will probably never be able to go to grad school at a hub of economic heterodoxy, but I can at least get and idea of what they are reading, and a few lecture notes wouldn’t hurt either.