Online courses and the future of teaching

[toorsdenote]

I had a conversation over lunch with a professor friend the other day about whether online classes are the future of academia -- at least for basic, quantitative classes like Calculus, which is one of the two classes I'm currently taking online.

I'm really loving taking calc online. At first I thought it was an OK substitute for taking it in person, but a lot of things about an online lecture class are better. For one thing, I can pause the lecture to do a problem myself and then unpause to watch him do the problem on the board, which is great practice. Second, I can rewind the lecture if I didn't understand or got lost in my own thoughts. Third, I can watch the parts of the lecture I already understand at 3x speed, which is so much nicer than being bored while he goes over things I already understand. It's also great to be able to watch the lectures whenever I want, which means I can work ahead in one of my classes if I have two assignments due the same day. We always tell students to do that, but often homework relies on a lecture they haven't had yet, in which case they're stuck.

At first I thought the online forums would be no substitute for asking questions in class, but in many ways they're better. I wouldn't get to ask many questions in a 97*-person class, which is what my calc class is. If the class were in real life, I'd get my questions answered by my friends in the class. In the online version, though, I get access to all the conversations classmates have had about the problem sets. Whenever I get stuck, chances are that someone else has gotten stuck in the same place and already asked my question. I can just read the replies they got, instead of making some poor TA go over the same problem over and over. (As an added bonus, the professor and TAs can read all the conversations the students have had, which must help when an academic integrity infraction is suspected.)

A downside occurred to me today, though, and I'm curious what you all think about it. Here's what I wrote to my friend:

It's a big time-saver for the professor that he can reuse a lecture year after year with different students. However, that means that he doesn't get the opportunity to experiment with teaching the material different ways and to figure out which ways are most effective. That seems bad in the long-term for professors' teaching skills.

If we imagine a future in which all college freshmen in the world (or at least the English-speaking world) take the same online calc class with lectures recorded by the best calc professor of all time (MIT's open courseware has calc lectures recorded in 1970!), then that's also a future in which nobody is learning how to teach calculus to other people. That just seems weird. I don't like the thought of having nobody, or a very small number of people, experienced at teaching calculus.

Then again, did bards make the same argument against the evils of recorded music? Did it bother people that we'd lose most of our campfire storytellers when books were invented? I think overall it's probably better to give everyone access to the same amazing recordings of Mozart than to leave us all stuck listening to our mediocre church pianists' renditions. (Church pianists might disagree...) Would it be OK for the same thing to happen to teaching?

* Just rechecked and it's up to around 145. Nice to know I wasn't the last slacker to add the class.

Comments

[roseandsigil.livejournal.com]

I will say that my former advisor is preparing an online class in programming languages ("language zoo", not theory, for the PLers in the audience :-) ). He's taught it 10ish times, and he says he's reworked it a couple times, but really feels that it is perfect at this point. So, this is kind of a final draft which he thinks is really good.

[toorsdenote.livejournal.com]

Yeah, so at this stage that seems like a good way for us to get a compendium of really good lectures. My question is what happens in 20 years. Will people still teach programming languages in person, or will new students just watch the lectures your former advisor is making now? If the latter, are there any negative consequences to the fact that nobody in academia has any experience teaching programming languages?

Or are the consequences the same as the consequences that practically nobody these days has any experience sewing their own clothes or smithing their own horseshoes?

[toorsdenote.livejournal.com]

(And yes, I know we won't be teaching the same languages in 20 years, which is why my original example was calculus.)

[ikeepaleopard]

We've been teaching C for about thirty or forty years now!

[roseandsigil.livejournal.com]

Maybe, maybe not people won't be *lecturing* in 20-50 years, but they will almost certainly be *teaching*. Having someone able to do more detailed explanations, possibly 1-on-1, is valuable.

[http://users.livejournal.com/_tove/]

Yeah, it seems that in the online calc class discussed above, the lectures are basically taking the place of textbooks (video-textbooks), but a lot of the actual teaching is still happening interactively, in the forums. Maybe that's replaceable with a really good search feature on the forums, but probably not any time soon.

[toorsdenote.livejournal.com]

"a lot of the actual teaching is still happening interactively, in the forums"

Yeah, my one complaint about the class is that the professors and TA do not, in fact, use the forums. You can have office hours by Skype, but I'd rather see the course staff actually keeping track of the forum and stepping in when someone's struggling. (Some students ask really baffling questions that make me think they didn't understand the lecture at ALL, but I don't really know where to start in answering the question.)

But if the course were done really well, yes, I think the forums would take the place of a lot of the teaching.

[http://users.livejournal.com/_tove/]

Ha, this sounds a lot like the dynamic I see at the intfiction.org forum, which has subfora for development in the popular languages (well, substitute "documentation" for "lecture"). The Inform forum there is more or less considered an integral part of the Inform-writing experience, and is recommended by the Inform website; several people involved with the creation of Inform hang out there.

What's interesting to me is that question-answering seems to self-sort pretty well, in that most people get useful and polite answers to their questions fairly quickly with no one person doing too much question-answering grunt work, because we all answer questions that involve just slightly less understanding than we have. So I'll usually answer things that are kind of subtle but familiar ("ah yeah, you're running into the [such and such] problem"), and whoever asked that will probably answer the questions about basic punctuation. (I get a sort of warm glow of achievement when my questions get answered by the people who work on Inform itself, because that meant it was, in fact, a pretty difficult problem.)

Unfortunately, the search feature for the forum is absolutely terrible, so we do get a lot of repeat questions. But unless it were really amazing, it wouldn't help with the total didn't-understand-the-lecture type questions, because if they knew what they were looking for, they wouldn't have a question at all.

[ikeepaleopard]

Is it google indexed? Everything I have ever wanted to know about cocoa is on stack overflow which I just search by google.

[zahariel.livejournal.com]

I also hang out on intfiction and have seen the same phenomenon. Usually if some newbie shows up with a newbie question, one of the semi-regulars will answer it within a day or so. And then one of the regulars asks some utterly obtuse deep magic question, and zarf and vaporware and frotz have some giant confused discussion about it and everyone else stares in wonder. Which I always find kind of funny especially if it happens by surprise.

[toorsdenote.livejournal.com]

"they will almost certainly be *teaching*"

Yeah, that's one scenario that occurred to me belatedly. As my friend Andrew said, nobody LIKES teaching calculus. It's something that has to be done before you can do the fun stuff with students. So, ideally, professors can spend the time they're NOT lecturing on the basics doing more interesting teaching with smaller groups of students.

[gwillen.livejournal.com]

Which one is your former advisor? How do I take this class?

[roseandsigil.livejournal.com]

Dan Grossman. Here is the course: https://www.coursera.org/course/proglang

I'm not sure this is really for you, though. It's a learn sml-scheme-ruby class, with some focus on interpreter implementation as a tool for understanding. It's probably stuff you already know pretty solidly.

[toorsdenote.livejournal.com]

Hah, I wondered when you said it was a programming languages class whether it was the same one I'm signed up for. Yep!

[roseandsigil.livejournal.com]

However, he is an excellent lecturer.

[ikeepaleopard]

People will also invent new ways to teach things. I was taught Calculus three times, once in a summer program where they tried to teach us about derivates as limits using epsilon-delta, which didn't make sense at the time, once in AP calc, which was basically here's a bunch of rules, which I could do but didn't make sense, and once in AP physics where we learned about what came up in class and all the intuitions finally came out. But now, having developed some mathematical maturity I'd be really interested in really doing the limit based approach.

The point merely being, there is no best calculus course. Maybe there will be ten best calculus courses, but even if they use the same outline, there are so many different pedagogical approaches...

Some of my classes at CMU were pretty good and taught by veterans, but even then there were obvious ways to improve them.

But probably teaching will be left to those who care about it rather than those who need to teach some classes as a requirement.

[toorsdenote.livejournal.com]

Well, that's something I didn't think of. I guess I assumed that most people teach it about the same way. It would be interesting to be able to actually study which approaches work best, and for what groups of people.

One thing that strikes me about my online classes is that the lecture aren't really as polished as I might expect, given that this is in some sense a "final draft" of the course. In the Octave tutorial video for my machine learning class, the professor had a small coughing fit and then actually typed OOPS EDIT THAT PART OUT into the terminal window. Apparently he forgot to take his own advice.)

[jcreed.livejournal.com]

Yeah my feeling is that the right comparison to think about is less

1M people thinking about teaching calculus
vs.
1000 people thinking about teaching calculus

and more

1M people thinking about teaching just calculus independently
vs.
1000 people thinking about teaching calculus with the benefit of being trivially able to see everything everyone else has tried for teaching calculus
1000 people thinking about teaching physics
1000 people thinking about teaching basket-weaving
etc., times 1000 different topics

[toorsdenote.livejournal.com]

with the benefit of being trivially able to see everything everyone else has tried for teaching calculus

That's a very significant point that I hadn't thought about. When I say it's good that professors can teach a class over and over until the lectures really gel, I'm really extolling the virtues of each grad student/new professor trying to reinvent the wheel.

[shmuelisms.livejournal.com]

toorsdenote, you already have some amazing debate going on here, and I have only a minor point to add:

"It's a big time-saver for the professor that he can reuse a lecture year after year with different students... That seems bad in the long-term for professors' teaching skills."

What makes you think this is NOT already the case?? The typical academics or teachers that I have known, considering lecturing to be the least desirable/worthy part of their jobs. Few, if any, have any formal pedagogic training. So they look for a formula that gets the material across. IF they care or are efficient, they'll tweak it over time, trying for better results.

While yes, they certainly ARE excellent lecturers, who do amazing jobs, most of them come by their skills naturally (and not through training), I think the vast majority do a middling job. So if, as jcreed points out, we reduce it to 1000, even 100 lecturers per topic, but they are the most passionate and specifically trained for it. Then we are in for a vast improvement. Who knows, I might even become less cynical/bitter about Academia.

Of course the drawbacks of this are twofold. Before you know it, the Admins will be saying "Why do you need to rework that? We already have a perfectly good set" and it will make the brilliant but ornery recluse researchers even more isolated.

[toorsdenote.livejournal.com]

You make a good point. I've been lucky in that (a) I went to a liberal-arts college where teaching was everyone's #1 priority, so while I had my share of mediocre instructors, I never had a professor who thought students were a waste of time; and (b) the professors who go to teach at CMU-Q tend to be pretty teaching-focused. If they didn't like students, they wouldn't be willing to move across the world to teach different ones.

So yeah, having the teaching of 101-level classes done by a few people with great pedagogy instead of by thousands and thousands of untrained and grouchy grad students/postdocs/unexperienced professors may well be a win for everyone.

[(Anonymous)]

You are drawing conclusions about lectures not being created that are quite absurd.

First of all, with online lectures you actually get shorter lectures pieced together. This means you can create them in little chunks, and you can redo parts of the lectures you find suboptimal. This means, actually, that you keep redoing lectures -- but only the parts you think need the most work.

Second of all, in general, with online lectures you get shorter lectures in general. Vaguely speaking, in education as of the last decade, the trend is towards more small chunks of time on specific topics, rather than 50 minutes working through a textbook. Regardless of how things proceed, it is likely that the delivery format will change, which means the lectures will, too.

Thirdly, obviously textbooks are modified, and as they are modified, so are lectures or explanation videos. Seeing as textbooks are technology-specific, they will most certainly see continued modification in the future, and so will the texts.

So your worries about a single textbook being used to the exclusion of others which in turn would lead to a dying textbook-writing field are rather silly.

One could go off on a tangent about how free textbooks are going to dominate the market soon enough, but it would be a side show, and would lead us to the same conclusion as above, anyway.

-jedgar

[toorsdenote.livejournal.com]

You bring up a good point, Mr. Hoover. My calc class actually just has one video per topic, around 60 minutes or 90 minutes long. But there's no reason online classes would have to keep the structure of a real-life class. My machine learning has a series of 10ish-minutes lectuers per topic. That was particularly nice for the linear algebra review, since I could just skip over (or watch on high speed) the parts I already remembered.

I didn't think about the open-source part of things. I debated between two calculus classes this fall. MIT's Open CourseWare is freely available to anyone, but I picked SFSU's online class, which is for credit. Actually, it turns out the SFSU lectures are freely available too (http://mangoroot.com/CalculusI/all), so really I'm paying just over a grand to get my homework and final graded, because that's what it takes to get actual course credit. I've certainly read things predicting that colleges in the future will play more a role of certifying that someone has learned what they say they have, rather than delivering the lectures themselves. That is also problematic, of course.