Slashdot Mirror
The Rage For MOOCs
An anonymous reader writes "Ever since Stanford's Sebastian Thrun and Google's Peter Norvig signed up 160,000 people for their online artificial intelligence course last year, educators and entrepreneurs have been going ga-ga for 'MOOCs' — massive open online courses. A new article in Technology Review, The Crisis in Higher Education, gives a balanced overview of the pluses and minuses of MOOCs as well as some of the technical challenges they face in areas like machine learning and cheating detection. The author, Nicholas Carr, draws an interesting parallel with the 'correspondence course mania' of the 1920s, when people rushed to sign up to take courses by mail. 'Four times as many people were taking them as were enrolled in all the nation's colleges and universities combined.' That craze fizzled when investigations revealed that the quality of the teaching was poor and dropout rates astronomical. 'Is it different this time?' asks Carr. 'Has technology at last advanced to the point where the revolutionary promise of distance learning can be fulfilled?'"
I got nothing....
Where ya been ?!! I been looking for you ya mooc!!
The issue is not technology, it is teaching methodology. It is not clear if we have developed teaching methods that are appropriate for large online courses, or even for small courses.
Palm trees and 8
It's a faulty assumption that lack of technology caused high dropout rates in during the correspondence craze of the 20's. The real issue is that a low entry cost coupled with a lack of requiring people to attend a physical room or building means that walking away doesn't involve any walking. You simply don't watch anymore. It's as easy as changing the channel on the TV. Essentially you're commoditizing education. Without a requiring a large investment of cash, all but the most serious students students feel no remorse about walking away.
Focus on knowledge advancement.
The "honor codes" try to make the desire to share information dishonorable. They promote beating around the bush, and try to penalize clear explanations.
The "Honor codes" assume that there is only one way to learn. From personal experience, however, I know that I can learn effectively from the discussions of homework questions, which often include the answers. Often the discussions contain clearer explanations than the instructors can provide.
There's also the question of "reinventing the wheel". Why have thousands of students do the same problems, which already have fixed solutions? Instead, assign unresolved problems that no one yet knows the answer to, and see if students can collaborate to advance knowledge.
Interesting that this pops up in my RSS feed just as I'm browsing MIT OCW for a new course to take. I've taken several already, and really enjoy augmenting my knowledge with the course materials. I've also taken most of the Stanford and Udacity courses, so I'm well aware of what they have to offer.
I'd say the value of these courses is personal growth. I do not see any possibility of using these online courses for any type of credentials, and I certainly wouldn't put my online course experience in front of my actual degree on any sort of resume or job application, but I would say "I have some experience dealing with X." In fact, I doubt I'd have the skills or base knowledge to understand most of the courses I've taken in advanced physics, mechanics, and computer science without my bachelor's degree.
Maybe they could teach how to run a Massive Open Online Business?
I hear MOOBs are really popular nowadays...
1. Time
2. Money
Spending an hour or two studying at home in the evening is a lot more accessible to most regular working people than driving to their local community college and blowing their whole evening there. Money is also an issue, as taking the course in-person guarantees that you A) have to pay for it, and B) need to drive there which comes with its own costs.
Free MOOCs take care of 1) and 2) simultaneously, so all things considered, is it really that shocking that they're becoming more popular and in-demand?
Also,
'Has technology at last advanced to the point where the revolutionary promise of distance learning can be fulfilled?'"
Really? Is this an article from the 1980's? Distance learning technology has been sufficiently advanced and accessible for at least 10 years. Just because you don't have anatomically-correct personal telepresence devices in each classroom taking the place of human bodies doesn't mean distance learning technology isn't "advanced" enough. Web-based educational technology is pretty well-developed by now, and in most cases gives you the exact same amount of human interaction as you'd get today with most on-site college classes. By that I mean, if you have a question after you've listened to the professor drone on for an hour with no classroom interaction, you need to send him an email and wait for a response. At that point, the people in the classroom might as well have just stayed home and watched a video lecture in their underpants.
Seriously. MOOC? Seriously?
I hear BOOBs are more popular
There is certainly a threshold below which technology is a fairly likely candidate for your educational problem(the development and widespread availability of the printing press isn't a bad option to designate, though I'm sure one could make an argument for others); but once you hit that point, it seems like the marginal return on throwing additional technology at the problem starts to degrade pretty rapidly until you get into the realm of sci-fi stuff like pedagogical AIs or brain interfaces, or possibly-available-sooner-but-still-rather-tepid stuff like performance enhancing drugs.
This is not to say that technology hasn't made some of the logistics of education more convenient or cheaper(pushing PDFs is easier and lighter than pushing paper, email tends to arrive faster than the USPS, etc.); but something like one of these 'massively open online courses' is really a lower-latency version of what people were doing through the mail in the '20s(if you simply must have sound, bump the timeline to the cassette tape era, if you must have video, videocassette tape era).
The weak links are still people successfully reading/listening to acquire the material, with both tutoring and assessments posing a real scaling challenge because both are comparatively labor intensive(unless specifically shoehorned into a scantron type format) and frequently necessary to keep less auto-didactic members of the class on track.
I've seen a few cases that seem genuinely novel in terms of solving this problem, there's some neat music-teaching software out there that can(once a given piece and instrument is added to its library, which only has to happen once) analyze and provide feedback on a student's playback of the piece. Probably not a major producer of improv jazz geniuses; but an actual improvement over the conventional 'practice, practice, practice' with much less frequent feedback.
For any subject that hasn't been conquered by an expert system suitable for telling students useful things about how they are doing, though, all the technology we can throw at the problem seems to amount to little more than a slightly cheaper, slightly faster, book.
I suggest investing all your money in BOOBs. I hear they're great investment opportunities.
In any case, the worst ROI you could expect is -50%.
If you just wanna learn, all you need is a good textbook, and some patience.
Or you could watch a video or read some shit on the internet, or whatever.
Learning isn't really that hard to come by.
If, on the other hand, you want to have evidence showing that you do, in fact, know the material, then it gets much trickier.
It's particularly tricky to automate, since it's intrinsically an arms race between students and testers.
The usual approach for automating decision problems is heuristic + blacklist + whitelist.
We can't really whitelist anyone, since generating a whitelist is, itself, the whole point. Your college diploma is the whitelist entry.
We can blacklist people, but only if they get caught, and it doesn't work all that well if they can just retake the course at no extra cost.
We can't use heuristics because with so much at stake, the students are highly motivated to cheat, and will exploit any weakness they can find.
The heuristic will quickly be broken and the whole thing goes to shit.
So basically, we go nothing that works here.
The traditional solution is tests taken in a controlled environment, under supervision of paid humans, with harsh punishments for cheating.
So far, I've yet to see any alternative to that, regardless of computers or the internet.
There is no breakthrough in sight.
I didn't see a dropout rate from the Stanford class, but I'd be interested to know if it matches that of the correspondance courses. The article mentions MIT's class with a 5% completion rate. Doesn't sound that great to me. But for 'free education' it's not too bad.
Here are some reasons, in random order:
1. The courses are "immersive" with frequent short quizzes, explanation of answers, etc. (in case of udacity, it is almost like once every couple of minutes). This is a big plus compared to correspondence courses.
2. There is a strong online community, instant access to reference material, forums, discussions, etc., which is a big plus.
3. Most of the material is free (I do not have any experience with non-free material).
4. The teachers are top class -- I mean, really top class, and the material they teach is high class and very unique [*].
5. The classes are massively scalable, archivable, easily made available, etc. (correspondence courses aren't).
6. There is an Indian saying "knowledge is wealth". So far, the top 1% have rarely helped the bottom 99% (and made them think that they should only "occupy wall street"). The MOOCs help in making the knowledge available to the 99% (turns out, it is a simpler problem to solve than the financial one).
The only major point people make is with respect to evaluating the credentials of a student who has taken these courses (and any types of cheating)... It is not a problem of the educator -- my belief is that the job of evaluating a candidate is mostly that of the interviewer. Employers that rely on lazy interviews in hiring people help the society at large -- they take away people that game the system out of the pool! And, slashdot should be the last place where education becomes secondary to grades (mind you, there are still grades for the MOOCs, and one can repeat the courses multiple times -- so one actually learns and deserves a top grade).
[*] To give a perspective, I am old, not from comp.sci background, didn't know python as of January (and have been destined to amount to nothing much!). I completed two courses on Udacity (CS101 -- thinking they'd focus on search, but they taught me python; and Peter Norvig's course). I had a phone interview with a "big deal" company where I gave a one-line answer based on what Peter Norvig taught [which impressed the interviewer -- and I explained him that their guy taught me the stuff]. I also took a course with Tim Roughgarden on Algorithms, and that helped me re-discover the joy of math and formal treatment of problems. I met him [Roughgarden] recently when he was visiting a nearby university, and his point was, if someone spends one hour on his class and learns something, he is more than happy. Without these courses, I'd still be wondering, "where did I screw up". Not any more.
I hate repeating myself, but what's this craze to attribute someone with such a reputation as Peter Norvig merely to his current employer? He's much more than a mere Google employee, IMHO. Can't we credit people with their real achievements instead of their employers?
cpghost at Cordula's Web.
It takes time to hone courses and online teaching methodologies into effective systems of education, so it's no huge surprise if the quality is below standard at the moment.
That first AI course of Thrun and Norvig's was nothing short of a didactic disaster, full of unexplained inconsistencies in the material, very limited coverage of the area, and no effective authoritative means of answering queries and misunderstandings. The many online fora were just the blind leading the blind. In summary, it was not a Stanford quality course. Many people still benefited from it, but that's a testament to their own individual perseverance and not to the quality of the material nor the teaching.
It will take time to get it right, and I'm sure that that particular course has improved already. But even more importantly, it will require a lot of experience to flow under the bridge before MOOCs earn significant respect, much of that experience based on trial and error. This should be no surprise. Physically attended courses didn't become perfect overnight either.
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
I know two women well, my wife, and a family friend. My family friend got a credential attending Phoenix University, while my wife is getting the same credential attendance a California State University. The University my wife is attending is a mid-range school, nothing particularly special, and definitely not a first tier school.
The difference between the two is rather stark. Wife is easily passing exams for accreditation that Family Friend (a very sharp gal, mind you) struggles with, making multiple attempts at without passing. Family friend is frustrated and in debt, Wife is blazing through excitedly. (Still in debt, but it's pretty clear she'll be able to handle the debt load when out in the field)
My respect for the CSU system has risen dramatically, and my contempt for the college system has vanished. I still think that the combination of computer technology and education holds tremendous progress, but it's pretty clear to me that online-only learning is still a work in progress.
We should be using computers as a way of transferring information, as a richer replacement to textbooks, but teachers very clearly still need to be a robust part of the mix.
I have no problem with your religion until you decide it's reason to deprive others of the truth.
I think interaction contributes a lot to learning. So in the online classes, I can interact with other students in the forums. And the in-video quizzes provide some interaction while I'm learning the material.
Someone posts an article on slashdot saying it is balanced. That is enough to determine it is highly biased.
( No Ihaven't read it. I have no idea which way it is slanted. I stopped reading when I saw the word "balanced". )
Most people don't want to learn, they want to be taught. Thus the hype about interactive textbooks (the ones with actual stuff you have to read are too boring) and online videos (because reading is hard and listening is just as bad).
There's this idea that if you just hit on the magic teaching method you'll be able to pour knowledge into people's heads and they can just sit in their lazyboy watching TV, I mean, the computer, and absorb it.
They're really not...
College has become far more about the degree than the experience, sadly. You can meet scores of graduates that have shining transcripts and dismal educations. And this is one of the reason the cost is so obscene -- as Thomas Frank said, "An annual pass to Disneyland would also cost $54,000 if society believed that what it took to make you eligible for success was a great many hours spent absorbing the subtle lessons of the Finding Nemo Submarine Voyage."
Until there is prestige associated with online learning, an online education will never be as valuable or acceptable as a brick-and-mortar degree mill experience. However, to those who actually want to learn and to do, access to high quality education experiences from anywhere in the world is fantastic and will only continue to improve with better technology and pedagogy. Though it's no surprising that the breakthrough course was in a geeky subject that attracts genuine curiosity.
Hey mate, spare a sig?
Article by a professor who took the course along with a small group:
http://cacm.acm.org/magazines/2012/8/153817-will-massive-open-online-courses-change-how-we-teach/fulltext
Nicholas Carr offering a "balanced overview" of anything?
Extraordinary.
Parity: What to do when the weekend comes.
The old degree system needs reworking.
It need to move to a more of trades / tech / badges like system.
The older systems is in big blocks loaded with lot's of fluff and filler and is slow to update to new ideas in fast moving areas.
in a older colleges lot's of the teaching staff has been in education for all of life and have little real world experience of what they are teaching in some areas.
We need a system that is
* not tied to the college time table.
* is open to drop in / continuing education
* more skills based / trades like
* real apprenticeships in the tech field not internships tied to college
After taking a few courses from Coursera, a high dropout rate is not surprising. The CS courses are mainly math courses in disguise, which works when you are teaching CS students at the high end of the intelligence spectrum, like at Stanford and other top-tier colleges, but simply loses most students otherwise. Even the NLP course was very focused on the mathematical models, much less so on the linguistics.
I suppose many might say it's not computer science without the math, but you can still teach much about computer technology and software design while being gentler with the math.
Personally, I've enjoyed the courses because I like math (except the quantum computation course, which was dreadful), but I know most of our CS students would be buried by the math. For the record, I'm at a state univ with some good research, but nowhere near a flagship. We do want to graduate some students, and the students we do graduate are in demand in our area.
Yes yes yes.
It is rather disingenuous of Thrun to complain about the use of filmed lectures in online teaching, while still himself using what is essentially a lecture format, when ignoring the work of one of the world's leading distance institutions who effectively ditched the video lecture years ago in favour of carefully planned, scripted and edited pedagogical lectures. Thrun has taken a massive step back and is well behind the state of the art in many respects.
Sadly, though, the OU is buying into the online "revolution" and moving more and more of there tutorials online. They even have courses with no synchronous tutorials, instead relying on text forums.
The fact that some people take no active participation in discussion isn't acknowledged as evidence of a problem, but heralded as proof of the superiority of the medium, by invoking the unproven idea of "learning styles". Yes, "lurking" has now been redefined as a learning style in online education land.
It's sad -- the OU risks destroying itself in the name of austerity... :-(
Got them moderator blues I blieve I walk out the do', With these mod-points I been gettin', I 'most never post no mo'
One think I've noticed so far, it seems that pretty much all the courses on offer are introductory. And when you think about the drop out rate, this kinda makes sense. If only 10% of people starting your course finishes it, what's the point of putting up a follow-up course? Already the market for that course is tiny in comparison to the introductory course. You can probably expect a lower drop-out rate on the follow-up course, but it'll be less than 100%. And not even all the students finishing your first course will take the second. So maybe 10% finish the first course, 75% of them take the second class, and maybe 50% of them finish it. Now what's the point of the third class in the series? You pretty rapidly go from MOOC to TOOC (tiny open online class) with a handful of students.
So my fear is, while this is a great way to get a broad introduction to a lot of different topics, you'll never see it really get into depth. For that, you will absolutely still need a traditional university.
BTW, I took the original AI class, the Udacity web engineering class, and I'm currently taking the gamification course of Coursera. I've also done the traditional university thing in the past for far too many years to get a Ph.D.
That's a bit harsh. Learning is much easier if you're taught.
In fact, everyone likes learning. Learning is the purest form of mental stimulation, and mental stimulation is the source of the sensation we call "fun". What people don't like is not learning. If a textbook is boring, it's because you're not learning. Most interactive textbooks fail to address this problem, and just add pointless bells and whistles that don't address the problem of poorly paced material. What online education does offer is a much tighter, better audited feedback loop, that should allow educators to build-a-better-textbook. That's what companies like Knewton are doing with their adaptive courseware. Unfortunately the big names -- Coursera, edX and Udacity aren't: they're saying "we're taking lots of data, and we'll you know, analyse it. At some point." They're not writing adaptive courses, which is a shame. They've had thousands of students follow a single iteration of the syllabus. In a real university, you'd revise the syllabus on feedback every year -- 20-100 students per iteration. If they'd iterated for every hundred students in the AI course, the course would now be damn-near optimised!
Got them moderator blues I blieve I walk out the do', With these mod-points I been gettin', I 'most never post no mo'
The answer to this is an obvious YES. For MOOCs like Udacity or even Codecademy, the lessons are free. That's really all that's needed to make it different, but there's more to it than that. Students can communicate with each other and get feedback much more easily and quickly than with 1920's correspondence courses too.
Taking guns away from the 99% gives the 1% 100% of the power.
Absolutely. Learning is much easier if your taught, and even easier if you're taught well. Teaching, particularly good teaching, is a service that is limited in supply and I think it should be reserved for people who want to learn. Free education for all (books, videos, etc.) and someone to spend their time helping you do it for those who are willing to put some effort into it. Education should absolutely be publicly funded but should not be free.
Everyone does not like learning. Yes, it's incredible to me too, but it seems to be true. Actual scientific studies have shown that the average (adult) person much prefers "learning" things he already "knows" (believes, rather) to learning things he does not. People like watching videos online and saying "ah yes, that confirms what I already knew." Videos that do that, or are designed so that the watcher can *think* they do that, are popular. Videos that challenge incorrect beliefs and confirm that the watcher actually learned something are not.
Learning is hard. Most people prefer not to do it unless they have a very strong motivation to do so.
...a lot of education research is barely research, relying as it does on very small sample sizes.
Actually I think the problem with sample size is not really solvable. If you increase the sample size of students you then must typically introduce more instructors. At this point you now have a new sample size problem: the instructor sample. Ironically MOOCs are really good at addressing this: one instructor can teach 160k students which is a large enough sample size that you can divide it randomly into several groups to act as control samples. So while I have strong doubts about the quality of the educational offerings of the current courses they do at least have the tools now to do real, scientifically meaningful, studies on the best way to teach so I have hope for significant improvements over time.
You don't need a course to be a "MOOC". In America many Universities offer videos of their lectures and some offer notes and exercises too. Buy the textbook - secondhand, show some gritm put aside the time and you can learn anything. If you get stuck there are forums like 'mad scientists' where people will help you. Connect with others interested in learning the same thing. If you get stuck on a particular concept, check another textbook for an alternate explanation or check out Kahn Academy. Many tutors post short clips explaining concepts on Youtube too.
Some people ask if you get credit for these. Of course you don't, and I question the motives of those people: If you're more interested in buying a piece of paper, then buy the piece of paper. But if you want to learn, yes, it can be done.
http://ocw.mit.edu/
http://webcast.berkeley.edu/
http://ocw.uci.edu/
http://www.youtube.com/education?category=University
Some universities only make the podcasts available only on ITunes, but there are alternatives to Apple's walled garden: http://www.copytrans.net/copytransmanager.php
Can I add: A Pox on Australian degree factories (also known as 'Universities') who won't make their content available online because they are more interested in squeezing every last cent out of their students. The idea of accidentally helping someone terrifies them.
Everyone does not like learning. Yes, it's incredible to me too, but it seems to be true. Actual scientific studies have shown that the average (adult) person much prefers "learning" things he already "knows" (believes, rather) to learning things he does not. People like watching videos online and saying "ah yes, that confirms what I already knew." Videos that do that, or are designed so that the watcher can *think* they do that, are popular. Videos that challenge incorrect beliefs and confirm that the watcher actually learned something are not.
Learning is hard. Most people prefer not to do it unless they have a very strong motivation to do so.
I think you're heading off on a slight tangent here. Yes, confirmation bias is a very real phenomenon. I see it in others all the time, and I recognise it in myself. But my point about everyone liking learning still stands -- the actual process of learning is genuinely, universally enjoyable. Confirmation bias isn't a mechanism that's "more enjoyable" than learning, it's a mechanism that simply blocks us from receiving new information, and prevents the beginning of the learning process. And there is very little involved in, for example, a machine learning course that is likely to be subject to confirmation bias. Therefore, if someone isn't learning from a machine learning textbook/course, it's the textbook/course's fault, not a student's lack of desire to learn or lack of enjoyment of learning.
Got them moderator blues I blieve I walk out the do', With these mod-points I been gettin', I 'most never post no mo'
I didn't mention confirmation bias. Perhaps you're demonstrating a little of what I was talking about. When presented with evidence contrary to your beliefs, unless it's correct interpretation is pressed upon you (which is not enjoyable) you bend it to support (or at least not contradict) what you believe.