The perils of online college learning

This article and a lot of the articles on the San Jose State are all of over the map when it comes to improving MOOCs: more TAs, more focus on the content, etc. They do, however, share one major thing: that half of the kids failing the SJSU class is, in fact, a problem that we need to fix.

What?

What about the other half of the kids? Think about it. This was an online class, with presumably very little (if any!) vetting of the students, with little or no supervision (they didn't even know a lot of the kids didn't have computers until three weeks in!), and a whole 50% didn't fail? I mean, this isn't a failure, this is a revolution.

Let's face it, these courses aren't designed to signal anything more than competency and --arguably-- the only way to actually display your competency to the world is to have not everyone pass. Because imagine if everyone did pass? You would go to your future employer and say, "Yeah, so, I took this course, online, where not a single person failed, and it's obvious that I'm competent, because they have this, ya know, innovative learning algorithm." If everyone passed than you might as well do MIT OCW. Which, by the way, is the exact same thing as a library, which we've had in every modernized municipality on the planet for 200 years.

No, online education is to regular education as the CFA is to an MBA. What is it, like 30% of people pass the CFA? Yeah, 30%! But no one is clamoring about how the CFA needs to be fixed. Quite the opposite--it's the gold standard. And not only that, it's downright democratic: anyone can take it in the world, and if you are in fact one of the ones that pass, the certification really does mean something. I mean, people put CFA after their name!

That's the future in online education. Not prestige, not A-'s and B+'s, and not feel-good relationships with professors: no, it's cold-hard certifications and frankly it's a godsend.

It isn't about replacing in person instruction with online. It is about making college courses available to anyone.

At the turn of the 20th century, my grandfather certainly could not afford college. He did what people at the time did who wanted to get a college degree - he took it by correspondence (by mail).

He liked it enough that he started a business selling correspondence courses. He met my grandmother through selling her engineering courses (she was a mechanic for the phone company).

Online courses are just a modern version of that.

I think the real failure here is the belief that the majority of people need to pass for something to be worthwhile. It doesn't seem likely that online learning will work better for everyone, but I strongly suspect that it will work better for some subset of people. In the long run, personalization will be extremely important in education.

Have you ever had a great teacher you really engaged with, but also heard from other people that they didn't really like the teacher? For myself, I'd say that I've found that only about 5% of teachers are actually engaging, with depressingly few of those before college. Imagine if you had a personal learning profile built up over your education, so that over time you were matched with the best possible (world-class) instructor for your learning style for every class. I'd bet most of the people here were near the top of their class; imagine never having to slow down for anyone, and automatically skipping reinforcement you don't need, so you can learn and advance as quickly as you are able. I think some people respond best to coursera style lectures, others respond better to Udacity-style continuous socratic method lessons, while others respond better to group work and discussion. Why bother trying one-size-fits-all solutions, when technology can allow for direct personalization. My Netflix experience isn't like anyone else's, why should my education.

I think there is actually a great opportunity to build a platform business in this space, where different instructors can design and build different types of courses. Imagine 20 different calculus courses, but the student only sees one. The best lessons are cherry-picked for them, and the curriculum and delivery is tailored to the rate that they are learning. I'd really like to build something like that, if anybody else wants to give it a shot, shoot me an email.

> Anyone who has come into personal contact with an inspiring professor at the top of his or her field knows that the experience can't be reproduced by remote control.

I was in Andrew Ng's first class of Machine Learning. He is one of the best professors I have ever had the pleasure to learn from, online or offline. He can transmit his passion for machine learning plenty well through video.

It's not the medium. It's some of the professors that teach online courses that haven't risen up to the standard imposed in the first classes.

If it's going to be an online class for millions, then at least find the most inspirational speaker you can before starting to record. Coursera has grown up too fast for its own good, probably many classes are just so-and-so.

Then the author goes on to say:

> My own entire graduate school experience was made worthwhile by a single extraordinary teacher, without whom it might have been an utter waste of time.

Exactly what I thought. Many people pass through brick-and-mortar universities wasting most of the time and money. It's not like real life education has magic. All it has is a real life person monitoring your presence in the lecture hall. It is just a little bit of motivation that is added on top.

We could simulate that with human monitoring - we don't need professors to monitor people taking online classes. We could do with "coaches"/people with psychological skill that would also have access to the students' online stats and have regular contact with the students.

Everybody works better when their efforts are being acknowledged. We are social animals and rely a lot on the social signals to build our motivation. So, why don't we just think of a way to have this aspect of offline learning in MOOCs?

I think the author buried the lead... Let's say 40% passed. At 7.5% of the cost of normal tuition, I would say this is a net positive. The students who passed have course credits and saved over 93% what they normally would have spent and the students who did not pass were able to see if college is right for them, without being destroyed by debt.

Not that the author's views are without merit, but this feels like it was warped into a story about greedy university administrators and profit-blind venture capitalists.

What kind of complete and utter idiocy is this?

50% of these brand new college students passed an entirely online course at a tiny fraction of the cost for the student, a tiny fraction of the cost for the university, at an infinitely increased level of convenience.

Does no one else see this as downright incredible?

> Anyone who has come into personal contact with an inspiring professor at the top of his or her field knows that the experience can't be reproduced by remote control; my own entire graduate school experience was made worthwhile by a single extraordinary teacher, without whom it might have been an utter waste of time.

As a CS undergrad, I can tell you that I've studied under a few professors that are famous academically, but the lectures themselves were nothing to write home about. In fact, when video was available for a course, I'd save myself an hour of commuting and just watch the lecture at home.

There might be certain disciplines where the style of the lecture is more dynamic, and where the lecturer needs to interact with the students, but I don't believe this to be the case with STEM subjects. Graduate school might be different, but I don't know much about that.

> But what the school also overlooked is that students with academic problems may not be the best candidates for distance learning. "They need people almost literally holding their hands, watching them do work," says Newfield, "being there when they get frustrated at the moment when that happens."

No one cares if you fail courses, and no one holds your hand in the University. The professor doesn't know you, and he isn't going to check up on your progress. The only difference is that you the professor might have office hours where you can ask him questions personally (something not many students do, by the way).

The article doesn't really convince me that there are perils. Turns out the class where 50% of students failed was one were lots of students had already failed the course before, and a a major other part of the students didn't have access to a computer. Surely MOOC isn't for everyone, as it probably provides less hand holding. It's still a great resource - I have already learned a lot from such courses.

As someone who has taken large lecture classes both in person and online, there's really no difference. A seminar is one thing, but no one gets anything interactive out of a class in a 300+ person lecture hall.

More students failed because they lowered the bar; I sure as hell won't fail a class I pay $2000 for but I might lose interest in one I paid $150 for. There's nothing wrong with a "low" pass rate in this case, it's just a different situation.

At my university, quite some classes have 50% failure rates, especially introductory. What's so special about that that its deemed a failure?

"They pointed to the risk that the edX model would lead to two classes of universities: a top tier "in which privileged students get their own real professor; the other, financially stressed private and public universities in which students watch a bunch of video-taped lectures and interact with … a glorified teaching assistant."

Unless a college professor is a thought leader in the field, with influential publications and books, aren't most of them "glorified teaching assistants" already, teaching from other people's books and distributing other people's papers for further reading and discussion?

> "Having a scholar teach and engage his or her own students in person is far superior to having those students watch a video of another scholar engaging his or her own students," the professors wrote.

I beg to differ. Lecturing has been demonstrated to be a flatly terrible method of instruction/learning, and I see minimal difference between watching a professor 50 feet away versus on a screen. When the educational method is so poor that a minimal number of students learn anything from it they couldn't learn studying textbooks on their own, digitizing it can hardly make it much worse.

http://www.npr.org/2012/01/01/144550920/physicists-seek-to-l...

This article illustrates the flaws with lecturing and, by proxy, with the current efforts made by MOOCs, Khan Academy, and so on. Ironically, this is not new information. People like Maria Montessori developed teaching methods more than a hundred years ago by studying children and constructing tools, classrooms and methods of instruction around how they learned.

From the linked article -

> Mazur's physics class is now different. Rather than lecturing, he makes his students do most of the talking.

> At a recent class, the students — nearly 100 of them — are in small groups discussing a question. Three possible answers to the question are projected on a screen. Before the students start talking with one another, they use a mobile device to vote for their answer. Only 29 percent got it right. After talking for a few minutes, Mazur tells them to answer the question again.

> This time, 62 percent of the students get the question right. Next, Mazur leads a discussion about the reasoning behind the answer. The process then begins again with a new question. This is a method Mazur calls "peer Instruction." He now teaches all of his classes this way.

> "What we found over now close to 20 years of using this approach is that the learning gains at the end of the semester nearly triple," he says.

Maria Montessori recommended teaching children in age groups of 3 years at a time (3-6 year olds in one classroom, 7-9 in another, etc.) The final stage of learning material is when you can teach it to others, and requires a greater mastery; hence, a classroom that spans ages/grades provides opportunities for older students to instruct younger ones and strengthen their own knowledge of the material.

This is effectively what Mazur has rediscovered and applied; by having students debate among themselves, the ones that are correct and hence, have a better understanding of the material, can explain to the others their reasoning behind their choice, as well as improving their grasp by having to defend against questions from the others.

Montessori also developed a wide variety of tools and "works" for students to use, allowing for a kinesthetic learning experience (something I would consider vastly more effective than lecture or video at younger ages, and of continuing importance even through college) - note that when Mazur's students largely got the question concerning gravity wrong, even after learning the mechanics of it in lecture, he resorted to practical demonstrations. There are few better examples of this than the Binomial Cube, which is introduced in Montessori preschools(!)

http://www.montessoriworld.org/sensory/sbinoml.html

Preschool students are not expected to understand the math behind the cube, which is after all algebra, but the work is meant to prime them for when they do encounter it several years down the line, as well as providing a physical construct that the math can be related back to at that point.

Which brings me to my final point: rather than trying to move traditional, ineffective methods of instruction online, edtech really should be aiming to maximize the learning capabilities of computers, which can allow things and methods of learning that simply aren't possible in the physical world. For example, the binomial cube, as good as it is, is still a fairly complex concept; a computer representation of it could extend it by allowing a student to reshape pieces in a 3D space, or provide values for the x, y and z variables and see how the cube transforms. This would allow for an even more powerful and intuitive understanding of how the pieces, sides and full cube relate to the binomial equation; and its manipulation that you could only do in a virtual environment.

That said, I don't really expect this to happen any time soon. Newspapers are still being dragged kicking and screaming into the digital age; I'd guess genuinely effective new teaching methods will similarly take years to come together. (It might be useful to start by discarding the concept of a "class" and look to MMOs for inspiration - material created for people at a variety of levels of understanding, with no real required time frame, where students can form groups to tackle difficult concepts/projects, or mentor others...)

Failing a class doesn't tell me anything. What does it even mean? Did 50% of the student not learn the material, didn't they learn but parts of it, what were the requirements, the criteria for "passing"? Who made that criteria, and what was it based on?

A single pass/fail ratio from a single case study doesn't tell me anything. Especially if the data is not explained.