April 19, 2012 3:44 PM   Subscribe

Coursera - free, online, introductory- to upper-undergraduate level classes in a wide variety of subjects, led by instructors from Princeton University, Stanford University, University of California, Berkeley, University of Michigan-Ann Arbor, and the University of Pennsylvania
posted by Blazecock Pileon (54 comments total) 138 users marked this as a favorite
I just signed up for several classes and totally looking forward to fall when they all start. For CS people, there is also Udacity. Class central aggregates the major providers.
posted by Foci for Analysis at 3:49 PM on April 19, 2012 [5 favorites]

I have signed up for the machine learning class. I totally expect to get my ass kicked, but it looks like fun.
posted by Blazecock Pileon at 3:50 PM on April 19, 2012

I just discovered Coursera's HCI class! Mood status: excited++
posted by Foci for Analysis at 3:56 PM on April 19, 2012

It would be fun to drop a " I remember taking this interesting course at Stanford " during cocktails .
posted by lobstah at 4:08 PM on April 19, 2012 [2 favorites]

While this is great I wonder if it's the thin edge of the wedge where second- and third-tier post-secondary schools will eventually close up shop because they just can't compete with freebie distance ed courses from the mega-schools. Why would the University of Guelph continue to run a CS department if Stanford is going to give away all its courses for free?
posted by GuyZero at 4:13 PM on April 19, 2012 [1 favorite]

Because you can't get a degree that way.


posted by snuffleupagus at 4:18 PM on April 19, 2012 [3 favorites]

You get a certificate but no credits at Coursera.
posted by Foci for Analysis at 4:21 PM on April 19, 2012

Somewhat tempted to sign up for the Gamification course, purely for the immense snark potential.
posted by adrianhon at 4:27 PM on April 19, 2012 [1 favorite]

You get a special badge if you read the first three units of the Gamification course on the first day.
posted by GuyZero at 4:27 PM on April 19, 2012 [11 favorites]

Alright, maybe it's time for me to become a machine learning learning machine.
posted by madcaptenor at 4:28 PM on April 19, 2012 [2 favorites]

Even if you did get credits, you'd have to be enrolled somewhere that you let you apply them towards their degree. External credits are always capped, and distance learning programs are often disqualified.
posted by snuffleupagus at 4:28 PM on April 19, 2012

You get a special badge if you read the first three units of the Gamification course on the first day.

...and your grade is based on how many new students you can refer for next semester.
posted by snuffleupagus at 4:29 PM on April 19, 2012 [4 favorites]

It's also worth noting that there are also free offerings from MIT and Harvard...probably many more.
posted by lobstah at 4:29 PM on April 19, 2012

You get a certificate, but it has a footnote that says (effectively): "We can't be sure who did the assignments or tests, or that they didn't cheat, and this is not official certification from X Uni". Also it's a PDF, and I imagine anyone who wanted to fake one could easily do that.

On the other hand as far as anyone can tell Coursera's business model will be to sell info about participants (with their agreement) to potential recruiters. If that's so, somewhere down the line, someone will need to trust what they say about participants almost as much as they trust the reputation of univs.

Also there are apparently some univs that are letting people count their Coursera work towards their own school's credits. I think I saw a guy in Norway whose uni was doing that.
posted by philipy at 4:31 PM on April 19, 2012 [1 favorite]

Signed up for Basic Behavioral Neurology. Yes yes y'all!
posted by moneyjane at 4:32 PM on April 19, 2012 [1 favorite]

All these computer science classes make me wish I was better at math. Signed up for some humanities classes.
posted by to sir with millipedes at 4:37 PM on April 19, 2012 [1 favorite]

The CSU system, or at least some schools, offered the option for students to "test out" of courses, if they could provide they were proficient in the topic of the course. If you really learned your stuff, and your brick-and-mortar uni was open to testing out, you could get out of some basic level courses.

Otherwise, having a solid understanding in some of these topics could help leagues when you're juggling a number of other courses, plus whatever extracurricular activities you are involved with, too.
posted by filthy light thief at 4:40 PM on April 19, 2012

Btw, I am signed up for Machine Learning, but I don't know if I'll have time, esp as NLP (which I am in the middle of) will overlap with it. NLP is fantastic I have to say!

Interestingly I found out today that you can look at all of Udacity's material without even registering. A strange contrast with Coursera where it's often hard to find out even roughly what a course will cover, how long it will be, or what the essential pre-reqs are without actually signing up for the course in question.

Most of the Udaciity stuff seems very intro level and slow paced compared with Coursera though. But you can do the Robot Car course any time at your own pace, and that's more the level of the Coursera ones.

A part of my brain thinks that the unintended consequence of all these courses is people will end up studying lots of fun courses in their spare time instead of... maybe actually building stuff in their spare time.
posted by philipy at 4:50 PM on April 19, 2012 [3 favorites]

Wait, no hard sciences? WTF?
posted by Chekhovian at 4:54 PM on April 19, 2012

A future full of badges.
posted by blucevalo at 5:22 PM on April 19, 2012 [2 favorites]

Wait, no hard sciences?

They have biology, math, and computer science. You have a very narrow idea of "hard sciences" if none of these count.
posted by grouse at 5:24 PM on April 19, 2012

It's actually because no one wants your boring old thermodynamics course stinkin' up their super-sexy online website full of gamification and machine learning hotness.
posted by GuyZero at 5:25 PM on April 19, 2012 [3 favorites]

To be fair, fake science is pretty easy.
posted by Blazecock Pileon at 5:27 PM on April 19, 2012 [3 favorites]

if none of these count

None of the mathematicians I know would be pleased to hear you describe math as any kind of science. They would never dirty themselves with anything that might touch the real world. As for biology and cs, the less said, the better.
posted by Chekhovian at 5:38 PM on April 19, 2012 [1 favorite]

They just announced more than half these courses in the last few days. We have no idea what they might have in the pipeline.
posted by philipy at 5:40 PM on April 19, 2012

I know there's a lot unknown thus far in Biology, but not a hard science? Really?
posted by mccarty.tim at 5:45 PM on April 19, 2012

something something physics and stamp-collecting something something
posted by Chekhovian at 5:49 PM on April 19, 2012

As for biology and cs, the less said, the better.

It's okay. I'd be bitter, too, if the government was shutting down all the funding for particle accelerators, space exploration and telescopes and there was no hope of getting a job that doesn't involve modeling finance math for crooks. But, hey, if you ever get bored come join me in my online rocks for jocks class. Next week, we're looking at YouTube videos of pebbles!
posted by Blazecock Pileon at 5:50 PM on April 19, 2012 [3 favorites]

YouTube videos of pebbles

Alright, that's pretty solid. I'm not terribly concerned about particle accelerators, but it does sadden me that america doesn't seem to do anything anymore, just create bizarre financial instruments and program social software. I rather liked cs back when it was solving logic puzzles about the most aesthetically pleasing implementation possible and calling graphics.h to write cute little games. As far as I can tell that doesn't seem to bear much relation to real world cs, which seems to primarily about discerning meaning from terse and incomplete help files or hunting for compatible versions of the whatits you need at the moment.
posted by Chekhovian at 6:12 PM on April 19, 2012

I'm going to repackage an AA presentation with the title "Debugging ANT"
posted by GuyZero at 6:16 PM on April 19, 2012 [1 favorite]

As far as I can tell that doesn't seem to bear much relation to real world cs

That's because you're confusing computer science with software engineering. It's a little bit like confusing astrology with astronomy. Real world software engineering requires similar levels of tribal knowledge, faith and alignment of planets.
posted by rh at 6:23 PM on April 19, 2012 [2 favorites]

Real world software engineering requires similar levels of tribal knowledge, faith and alignment of planets

My cs buddies back in college weren't actually taught real programming in any of their classes. It was all theory and abstract algorithms and discrete math. They were left on their own to figure out how to get any of the stuff they needed actually working on their home machines for their homeworks and final projects etc. Then again some of my friends at another uni got four year degrees in cs and new all the theory but somehow couldn't manage to open up an IDE and make something do something. Poor bastards.

Still not as bad as my ChemE friends. The department policy was that only excel was allowed for numerial analysis, no matlab, mathematica, nothing. So somehow they'd do these convoluted nonlinear fits in excel, manually changing parameters and tweaking it by hand iteration by iteration. That was voodoo. Impressive, but sad really.
posted by Chekhovian at 6:43 PM on April 19, 2012

All very interesting, but what does it have to do with Coursera?
posted by grouse at 6:44 PM on April 19, 2012 [1 favorite]

All these computer science classes make me wish I was better at math.

If you'd like to learn about computing, Udacity 101 looks like it would be accessible to anyone, and probably needs no math at all. You can check it out as much as you like without registering.

Also Statistics One has the aim of being accessible to anyone, and says it is "designed to be a friendly introduction to very simple, very basic, fundamental concepts in statistics."

And if you actually want to learn more math from the ground up, Khan Academy seems to be worth a look as well. But there are a lot of things you can do in computing that don't involve much math at all, so don't let that put you off if you're interested in learning.
posted by philipy at 6:51 PM on April 19, 2012 [3 favorites]

> Why would the University of Guelph continue to run a CS department if Stanford is going to give away all its courses for free?

Because Guelph is in the top 15 research universities in Canada? My husband said he learned more from the classic MIT online course "Structure and Interpretation of Computer Programs" than all his actual computer courses at college and university, but going on and doing research is something else again.
posted by Listener at 7:10 PM on April 19, 2012

Why would the University of Guelph continue to run a CS department if Stanford is going to give away all its courses for free?

Hahaha. Ha. Ha. Aha. Ha. Ha. ha. Good one.

Why even attend college courses? Sit in your underpants and watch videos and get the feeling like you understand what's going on without ever having to do any work! Then, go talk to someone who does understand the topics and see how you really don't know what's going on, because you didn't really understand it at all!
posted by King Bee at 7:20 PM on April 19, 2012

Somewhat tempted to sign up for the Gamification course, purely for the immense snark potential.

I have to agree, although without seeing the syllabus, who can really say if this guy *really* knows his stuff in the empirical sense, or if he's blowing it out his cheeser because it's a hot topic. Interesting. I just searched him in the Harvard Ed School's research database and came up empty. However a Google search shows he's an associate prof at the business school. He's definitely connected (Wharton, FCC, etc.); he seems to know his law and business (I am in no position to judge), but from a teaching and learning standpoint, he doesn't exist. So there's a possible grain of salt, depending on your point of view and objectives.

However, as an interested party in open courseware and learning, I will keep tabs on this with enthusiasm.
posted by smirkette at 7:49 PM on April 19, 2012

Hm I don't know, should I stay with my Night Elf Shadow Priest next expansion, or reroll as Combinatorial Analyst I?
posted by polymodus at 7:52 PM on April 19, 2012 [3 favorites]

I rescind my last remark. I fiddled in my search terms and came up with 4 search results. All law journals.
posted by smirkette at 7:54 PM on April 19, 2012

Been reading about it a spin!
posted by huckleberryhart at 8:13 PM on April 19, 2012

Half through Coursera's NLP class. It's been challenging but rewarding.
posted by yerfatma at 8:23 PM on April 19, 2012

A part of my brain thinks that the unintended consequence of all these courses is people will end up studying lots of fun courses in their spare time instead of... maybe actually building stuff in their spare time.

Personally, I plan to get out of the garden and off the internet long enough to read the books required for the course on science fiction and fantasy, watch the lectures from a popular prof, involve myself in whatever interaction is offered (some is) and otherwise enjoy the hell out of something I could never otherwise hope to do in this tiny little town at the end of Nowhere, U.S.A.
Spare time? Honey, I've got nothing but spare time and the closest I come to building stuff is the baloney sandwich I had for lunch.
posted by kemrocken at 9:05 PM on April 19, 2012

On the Udacity home page there's a quote from a student:

"I feel this is something as fundamentally new to education as the invention of printing."

I guess a lot of people sense that something like this might really turn out to be true, and that's why we're excited. And that's why a lot of these Profs are taking on a bunch of extra work for no pay to get their courses out there.

If it does turn out that way, it'll have all kinds of unexpected effects. At the moment it looks like having high quality free lectures to watch might lead to "flipping" the classroom, with students watching lecturers on their own, but working on problems in class.

Salman Khan's TED talk explains that concept, and they've been seeing that happen at Stanford with their own campus students.

When you think about it, learning by watching videos from a Stanford prof isn't any more of a radical concept than learning by reading a book written by a Stanford prof. No-one thinks that's unusual or "not real learning". That would be like thinking that reading The Feynman Lectures is valueless if you couldn't be there in person.

In fact I guess there are plenty of us that learned more from books we read than ever we learned from sitting in lectures.

Actually I did the online courses I did as much to check out what the potential of this way of doing education might be as to learn anything specific. Initially I wasn't sure how well it was going to work, doing the courses was an experiment to find out.

From what I've seen it actually works really well, and in some ways even better than traditional methods. You can't raise your hand to ask a question, but you can stop the video and rewind and review the point that you didn't quite get. You can stop the video and check something out in a book. You can stop the video and Google a term that you don't know. You can go read discussions of the topic in the class forum. You can stop and do a bunch of calculations and convince yourself of a point. If you hear something really interesting, you can stop the video and go read the paper that was mentioned while it's still fresh in your thoughts.

I should mention that the lecturers I've seen are very good teachers. They're engaging, enthusiastic, know their stuff, and know how to communicate it. Dan Jurafsky on NLP is one of the best teachers of anything I've ever come across.

The quizzes and assignments are very well designed too. The videos come in short 8-15 min segments, and most have some self-test question included, and those work well for testing that you're really taking in the material. Most classes also have weekly problem sets and assignments, and they're pretty substantial. So if you do them, you do get a good grasp of the subject.

Also the forums are a revelation. They have all the pluses and minuses of online communities. But just like you do on Ask Mefi, you get people who are pretty knowledgeable on some topic explaining it to others who are confused. And just like Ask Mefi, it is pretty fun and satisfying to help out your classmates on the forums, and nice to be able to ask for help when you need it. From time to time the profs will check in and explain a point that seems to be troubling people. Sometimes they'll even make an extra video Q&A session where they'll answer some things that have been raised. It's not like being able to raise your hand in class or discussing a point back and forth in a tutorial, but it's considerably closer to either of those than reading a book would be.

And compared to reading a book, you also get all those "gamified" aspects of a doing a course. There are deadlines and scores, and you care about your score, and how you're doing compared to everyone else. You stay up late reworking your model so it can get a perfect score. And when you get that certificate at the end that formally means nothing, psychologically it means a lot. You got through all the challenges, and you achieved something that you feel good about.

Overall it's a surprisingly complete learning experience.

And when you see the number of adults with full time jobs and families that are making time week after week to keep up with these courses, maybe 10-15-20 hours a week, well, that is a sign of something pretty powerful going on.

At this point it's hard to know how this will evolve. Maybe it won't work well for some subjects, or maybe the funding for it will dry up, or everyone will get bored when it's not an exciting new thing. But maybe it will be really important for the future of education, for how people live their lives, and what opportunities people have.

I could certainly see these things growing and developing and in a not too distant future becoming as pervasive as video games, and seeing a lot of adults and kids immersed in regular online learning the way they're immersed in regular online gaming today.
posted by philipy at 9:16 PM on April 19, 2012 [7 favorites]

baloney sandwich

I think we could adopt that as a good term for a lot of tech projects.
posted by philipy at 9:22 PM on April 19, 2012

My security class at Berkeley is run on Coursera, and honestly, it's been pretty bad.

The quality of the lectures leaves something to be desired. I can barely hear one of the professors over the noise of his recorded lecture (I'm not sure what's going on in the background) and I really wish there were transcripts to supplement the videos.

The quizzes also aren't functioning correctly, because a number of their questions have text boxes where they ask you to type in the answer. They use a regular expression (or so I've heard) to check to see if you've typed in the right answer, which doesn't work well. Sometimes I've had to type the exact format (periods and all) to get the answer to be right. The answers for multiple choice questions also bleed into each other due to misshapen formatting.

Also, the labs/homework upload section of Coursera were so non-functional that we had to switch to a different website half-way through the semester. We also no longer use the forums, because Piazza happens to have a cleaner and more user-friendly interface.

However, I will say that it's nice to have a set of video lectures online that I can go back to before midterms and finals. My school already has webcasts though, so I'm not sure this is much of an improvement from that, except for the fact that I can speed up and slow down lectures at will.
posted by movicont at 9:39 PM on April 19, 2012

I did the Berkeley SaaS as well. I think they had some tech issues with getting the Berkeley things to work with the Coursera (Stanford-based) platform.

Also they rolled SaaS out on schedule, unlike many of the Stanford courses, but that maybe meant it started with more glitches. I know on SaaS they weren't able to use the kind of presentation technology they wanted and had to fall back to what they could get to work.

There's plenty of room for improvement - the forum functionality isn't all you could want, there were glitches with autograders etc.

But the course materials and teaching were very good, and they did call it an "Alpha Version" of the course, which turned out to be an accurate description.

So I guess that is an important thing for people to know before they decide to do a course. You may experience all manner of tech problems, as it's still early days for the platforms. If you're not feeling like dealing with that, maybe wait for a later course.

I can't generalize too much as I only have one Berkeley and one Stanford course to compare, and the Stanford one started later. But FWIW the Stanford course has been going noticeably smoother, though not without some issues.
posted by philipy at 10:32 PM on April 19, 2012

> At the moment it looks like having high quality free lectures to watch might lead to "flipping" the classroom, with students watching lecturers on their own, but working on problems in class.

My university has been putting more and more stuff online, it seems, and while some of it is good I feel the real motivation much of the time is making things cheaper for the school. Students pay the same price for an online course marked by an online system as they do for a lecture/seminar course with labs or thoughtfully marked long writing projects. Seems obvious which makes more money for the school. Expanding their income streams!
posted by Listener at 12:05 PM on April 20, 2012

Here's the strange thing about cs courses at Stanford: they've been sort of reversed already. The English department will demand total attendence, but the CS instructors are already teaching remote students(Stanford has some remote student programs) so a lot of people don't go to lecture.

I'm in CS109, or basic probability for CS kids, for example. 160 people enrolled, about 40 people attend. Class is taped, all the many TAs have office hours, meaning like 25 hours of office hours a week, and you can attend those without a question and just do your problem set while getting questions answered.

I would watch out for CS161 and CS124, because the big criticism of CS221 by the actual Stanford students was how it got watered down in the service of the online people. Well,161 and 124 are fairly significant, really challenging classes (161 more so than 124) and we should be able to see if it has that problem.
posted by curuinor at 8:54 AM on April 21, 2012

All of my law school classes are taped. I habitually review the class recordings to solidify learning, and in preparation for finals. The school's policy and most professors' informal practice still requires and enforces attendance (although some profs may allow more leeway than the school claims to.)

The policy is most strongly enforced in smaller classes, but even in the giant 100+ student single-exam memorization-heavy lecture courses Profs will get in your face for missing too many sessions. In fact, I'd say from what I've seen law schools are getting MORE uptight about in person attendance at the same that it's become less important in other areas. Even classes held online insist on having a live lecture component w/an attendance requirement. From what I've heard from previous generations of lawyers, this is something of a reversal from the traditional approach, where many students would attend infrequently at best and just show up for the final.
posted by snuffleupagus at 7:09 AM on April 22, 2012

big criticism of CS221 by the actual Stanford students was how it got watered down in the service of the online people

I'm curious to know how they would be able to establish that it was watered down, and in what ways they think it was watered down. It's a pretty common social phenomenon that someone who did a former version of something glances over the new version briefly and declares how it was all so much better in their day, and that becomes the standard story everyone tells each other, with no thorough examination and no real basis of data.

Not saying that it wasn't watered down, I wouldn't know about that. But "what everyone says" is not that strong as evidence goes.

And given that according to surveys a large majority of the online class were already graduates, a large majority were also experienced tech professionals, and a sizable fraction already had a grad degree, it's not so clear that it would *need* to be watered down.

Just look at the kind of people on Mefi who are saying , "I think I'll do that class". And consider what kind of people generally would self-select to put themselves in such classes.

FWIW, the Berkeley SaaS class was apparently in no way watered down and the identical campus version was considered to be one of their more challenging classes by the Berkeley students. Performance is hard to compare because the conditions are so different, but in as much as there was any evidence at all, the UCB and online students seemed to perform about the same.

One thing that is true about the online classes though is that the online student population is much more diverse than the college student population, and diverse along a great many dimensions. It'll be interesting to see how well a single MOOC can meet the needs of diverse groups, versus whether what's needed is a variety of courses aimed at different segments.

My guess so far is that the best answer will be a mix. i.e. Design each MOOC so it can be used in varied ways depending on need, as well as having a variety of MOOCs from a variety of sources for each topic, with different intended audiences. One thing that's already happened on Udacity is that as well as the standard option of doing the courses as a formal class with scheduled assignments, it's now possible to do some of the courses entirely at your own pace, and it's also possible to just dip into only those segments that are of interest to you.

Without the constraints of college logistics, MOOCs could be something closer in flexibility to books, which you can use according to your need. For example you could have a lot more depth than any single class could cover, and you could have more exercises and suggested projects than any one person would ever do, and leave it to people to go deep in the areas they need, and skim or skip what they don't need.

I'd guess that MOOCs might (if they continue to exist!) eventually evolve into something much different than traditional campus-based classes.
posted by philipy at 9:04 AM on April 22, 2012

I did see this article about Udacity a while back, which said:

Thrun was eloquent on the subject of how he realized that he had been running “weeder” classes, designed to be tough and make students fail and make himself, the professor, look good. Going forwards, he said, he wanted to learn from Khan Academy and build courses designed to make as many students as possible succeed — by revisiting classes and tests as many times as necessary until they really master the material.

Now changes made because of that could well look to some people like "watering down" without actually necessarily being watering down. There's a difference between not teaching important things because they're difficult versus finding a way to teach those same things that makes the formerly-difficult much easier to grasp.

And if you're proud of having done it the hard way, and you've got ego tied up in believing that you can do things that are too tough for most people to do, it's kind of hard to take that maybe a lot of people could learn to do it.
posted by philipy at 9:47 AM on April 22, 2012

So i'm taking the machine learning course. Does anybody know what the relationship between the course and Stanford CS229 (same professor, similar-looking syllabus) is?
posted by madcaptenor at 2:01 PM on April 23, 2012

So I've just now finished the first week of lectures and quizzes for the Stanford Machine Learning course I've seen a few people say they've also signed up for. I have to say I'm quite pleased. Although the first week only covers the most basic of techniques, I've already learned something I can put to work right away. I'm excited enough about the rest of the course that I think I'm going to start week two tonight.

Obviously none of it "counts", but I've always been a, Let every sheep keep its own skin, kind of guy anyway. What matters more to me is actually learning something new, and so far Coursera and Stanford have delivered, at least for this course. I'll almost certainly be going over the catalog and attempting another one in the not too distant future.
posted by ob1quixote at 6:36 PM on April 25, 2012

Glad to hear it, ob1quixote. My schedule for this week has been pretty front-loaded so I've been putting it off until later in the week.
posted by madcaptenor at 6:40 PM on April 25, 2012

« Older Better red than dead   |   Julie Taymor not involved Newer »

This thread has been archived and is closed to new comments