Hacker News new | past | comments | ask | show | jobs | submit login
Are Courses Outdated? MIT Considers Offering ‘Modules’ Instead (chronicle.com)
150 points by cardamomo on Aug 25, 2014 | hide | past | web | favorite | 52 comments

No kidding.. Personally, I think for MOOCs [1] to really take off you need to let people choose how they consume the course. Look at Netflix and House of Cards. People can choose the pace at which they consume. MOOCs are trying to use the class structure and convert it to the online world, not to mention the price structure ($7k for an online course?!? [2]). It is a major pain point, but this seems to be the norm, at least for any of the courses I have looked at.

My personal use case is that I want to take a MOOC to learn a specific topic (after I've left school), so that I can use it in my real life, but you want to slow me way down (on some artificial schedule), and then charge me a crap load of money. Yeah, that's going to work great!

[1] http://en.wikipedia.org/wiki/Massive_open_online_course

[2] http://www.forbes.com/sites/troyonink/2013/05/15/georgia-tec...

I think online "course" is a tricky self definition to start with. A more natural starting point is textbooks. IE MOOCs should aim to use technology to make textbooks 10X better. Textbooks can be used in the context of a university course or for self study. If they are good enough, they can enable new kinds of use cases.

Imagine that there was an online resource for math that is perfect. Lectures, exercises, hints, spaced learning, badges, etc. Everything from high school level studies to the math classes required for Science and engineering Masters degrees.

It's so good that not creating your course around it is just being wasteful. You might be able to add value to that by creating a disciplined environment where people study in the same room or are required to have assignment X completed by 16:00 daily. Maybe you can improve (I suspect you can) on the results of self learning by adding in person tutoring.

At that point, a teacher might offer her own course. A 4-1 teacher to student ratio is much more affordable if you only need 5 hours per week. This is where the disruption comes from.

So basically I think that "disrupting Universities" is a kind of awkward way of thinking about this. It means online learning are looking at universities and asking themselves how can I replace this, instead of trying to do textbooks 2.0.

Your perfect "textbook" must be structure free in som sense. The classical textbook imposes a given structure which may or may not suite the teachers ideas of course form. The flexible learning resource is the problem to be solved.

I suppose thats debatable. There's probably some use for both. A textbook can also be a syllabus and plan for a semester. IE, perhaps a textbook is divided into 15 chapters, each representing one week of work and exercises. Lectures, tutorials and such can be structured around that. In other cases the teacher already has a structure in mind and they are looking for materials to support it. Both cases exist.

I suppose the ideal would be layered and modular so materials that can be snapped together like lego. But realistically, modular is a thing difficult to get right.

Coursera has a few self-paced classes. I've signed up for a few, and I've never come close to completing one. On the other hand, I've completed a few courses with a strict time schedule.

I think they've tested the self-paced idea out and found it just doesn't work as well under the circumstances.

Conversely, I've never managed to complete a Coursera course that was on a schedule. And I've failed more university course than I should have, for the same reason. I'll usually buy textbooks on my own and combine them with online resources to learn on my own time, and this has worked out better than almost any class I've attended.

The only exception to that is when professors took special note and allowed me to work differently. I once had a Java professor that instead of working through the weekly goals and quizzes, allowed me to pick a large project to work on over the course of the semester, and he would give me goals from the regular syllabus that I had to meet, but they weren't part of any strict schedule like the other students. Things like polymorphism, using an SQL database to store data, etc. I ended up making a tool for managing player stats and notes for a D&D campaign. It was both the most fun, and most informative course I'd ever taken. However, I understand it is unreasonable to expect a professor to split their time and attention in such a way. Especially if every student wants to pursue their own individual learning style.

Courses with a strict schedule are obviously effective for teaching a majority of people. But that doesn't mean it works for everybody, or even that it is the most effective method for that majority of people. I'm very excited to see an institution try out an alternative way of structured learning.

I totally agree. To me, one of the advantages of these online classes is that they usually impose some kind of pace. It's something difficult to enforce when self-studying. Otherwise, I may as well read a book which I think is a better format than a video.

The problem, at least for me, is that there's no real incentive to get a "good grade" in those courses. I can't imagine one single employer actually caring that someone has an actual coursera course completion letter.

So when I do those time sensitive classes, if life gets a bit busy and I can't keep up, I'll just skip that week. And then of course I'm kind of screwed as I likely don't have time to do 2x the coursework the following week and it snowballs.

If instead it was self-paced I could go ahead of pace at will and behind pace at will. If I fell behind pace there I can still complete it AOK.

And perhaps this is just me, but for some reason once I fall behind in the time sensitive classes I rarely manage to complete it on my own time. There's something subconsciously working on me with the time schedules such that I throw my hands up and say screw it.

Self-pacing or very loose pacing also much makes any sort of discussion board or peer to peer interactions pretty much useless. Discussion boards have pretty horrible S/N ratios and other problems under the best of circumstances but with self-pacing you almost might as well not have them.

And then, as you say, for a lot of topics it's then worth asking exactly what function the MOOC is serving that isn't served equally well or better in many cases than an appropriate book.

That's a good point and it touches on some of the intangibles that universities provide. OTOH, just emulating a university is probably not the best solution. Self pacing has a lot of advantages and it would be great if commitment devices worked on those too.

I think both self-pacing and and scheduled courses have their own advantages. For a self-pacing course, it's more flexible to learn, but it might also be too flexible so that students cannot finish it. And the forums for those self-paced courses are just like a ghost town. On the other hands, a scheduled course with some due dates make all the students in the same rate of progress. It might have more interaction between students. It definitely helps if online students can have some interactions. Also, under the stress of the due dates, students might have more motivations to finish those assignments.

I used to sign up many courses online (Coursera, Udacity, eDX, Code School, Treehouse, etc ... ) but finish few of them. I'm now trying to hold some small study groups (3-5 people) and see whether we have higher probabilities to finish those courses or not. In my experience, social power is unlimited.

My second cents.

Sure but Netflix is entertainment. I agree, you may get more users wanting to participate and "stick" with a module if you make the modules smaller and even more up to the user's leisure. By virtue of making things smaller, you make completing an individual unit "easier." I think if your goal is to have a higher "completion rate" per "unit of education" the sure that solves it. However, that is not an interesting metric in and of itself.

That being said, I think MOOCs definitely have their place; I just don't think they displace traditional universities. More often than not, I think for very engaged students, they complement an existing traditional student's education. I think a big problem with the format offered by Coursera is really just this idea of "This course starts this time and ends at this time," which is, like as you say, trying to mimic traditional class structure for not much apparent reason.

What if I did not know there was going to be a Cryptography class on April 10th (I made that up) but I really wanted to go through the course and solve all the problems and get a grade rather than being "penalized" for only finding out about it months later or even after the course completed? Again, though, I see MOOCs as additional education but not a replacement for traditional college education, and probably more beneficial for autodidact learners.

MOOCs are great for autodidacts who want to learn small lumps of information, and who generally already know how to educate themselves. They're not great for creating balanced professionals, or for non-autodidacts.

Plenty of universities offer some flexibility in their courses anyway, so there already is some modularity. Some base subjects (if you're doing medicine, you need biochemistry, for example), and some electives.

What I often do is sign up for any Cousera class that I could conceivably be interested in. I do nothing when the course is live, since I don't care about the certificate (paid or unpaid). Having signed up for the courses, though, I have access to the content in the future and can pick and choose what to watch. I then come back to the course and can consume it rapidly and with a focus on the materials I care about.

For example, https://www.coursera.org/course/money . I had no idea what in this course and its sequel were important to me. But then a bank mentions how Standard Treasury should understand repo markets, and then I watched the lectures on that topic.

So the idea of consuming courses in modules is very appealing to me, and honestly how I consumed most classes in college: the material prescribed was basically incidental to what I ultimately learned about the topic. I'm the type of person who gets really focused on some applied problem and then teach myself the, for example, CS, stats, measure theory, machine learning, etc, to do what I want on the problem.

Tangent: My only compliant so far is that the (pure) math I can find in MOOCs goes exactly to the point I stopped: advanced calculus and theoretical matrix algebra. I don't expect it to have the types of statistics classes I'd like (I majored in the topic) but I'm annoyed that I can't really find an analysis course...

Yea, it's really tough to find an analysis course. I looked around online for a long time and never found one. Ended up picking it up as a regular college class and learned a lot. I don't understand why nobody has bother recording analysis lectures and posting them as they have with other subjects. It seems the only autodidactic way to learn analysis at this point is to download a book and dive in. If you're looking for pure math though, http://www.extension.harvard.edu/open-learning-initiative/ab... is a really great Abstract Algebra course taught by Gross at Harvard though.

While not a course in the udacity/coursera sense, I found these real analysis lectures by Professor Francis Su extremely good. He breaks down every step of the derivations. And while sometimes the questions he asks, or the things he states may seem trivial I think it helps to focus on these simple but very important details.


There's a functional analysis course at https://www.coursera.org/course/functionalanalysis. I've only skimmed the notes but they look pretty good, if a bit terse.

In general I think the lack of higher mathematics classes is because it would be very hard to evaluate proofs in a MOOC setting.

This is just so simplistic. I'm surprised it comes from MIT. The idea that students pick songs and not albums seems a silly analogy to coursework. The use of MOOCs as evidence of anything seems very thin. I'm sure many students report that they only studied the parts they were interested in of a course rather than 1) they did not have the time to devote, 2) had some free time but did not take it as seriously, 3) it was hard an gave up, or just that maybe the student genuinely only interested in learning the pumping lemma but not much more about push down automata; however I'm not sure that sort of cherry picking does a college education make.

We can talk about costs and all of this kind of thing but I feel like we dumb down higher education because it seems hip (like picking songs for your iPod because we are all have hyper attention deficits, or something).

the very concept of structured courses in education is because we don't know what to study to understand XXX things.When it comes from a teacher , it comes with a dependency graph drawn by a person who knows the course.If proper knowledge gathering is considered an objective.This dependency is a natural consequence of subject-matter than an optional one .

For e.g. for someone to learn abstract algebra ,its highly suggested that they take up linear algebra first.As for me, i've always the dependency graph quite seriously .May be the gifted brains can leave linear algebra and directly take up abstract algebra , is that so?

I'm not sure about abstract algebra specifically, but in other subjects that I have tried to take up on my own, I have found that the dependencies become obvious once you realize you are in over your head. In fact, my preferred method of learning these days is to start with your goal, then traverse that so-called dependency graph to the dependent, and often simpler, concepts until you are able to work your way back to your original goal again.

This is a nice way of priming your mind for the concepts you're interested in--and is probably another example of why the current educational system languishes under standardized testing for metrics: when the goal isn't in the students' direct interest, the system can only provide an indirect benefit for their minds. It's very inefficient.

The class structure has grown out of the physical limitations of presenting course information to mass groups of people in physical classrooms. Take the MIT Modules, make them asynchronous and add accountability (limitations and commitments: something like, only 5 active modules at a time until you've completed them) and mentoring (which is your next module? If you struggled with X, maybe you should go back a couple dependencies?) and you could have a really effective system that creates even more amazing people--at the pace that suits them best.

They do not suggest that dependencies are ignored. Not everything in sequentially dependant. And sometimes some material may apply to multiple courses. The idea is to break things out so they can create the dependency graph and take advantage of places where there actually aren't dependencies or know for sure when there are.

I think you're taking this the wrong way. Ideally, most of what I've learned could be used under different scenarios. I mean, you can learn programming to hack music, understand data or create an intelligence to do some of your work for you. There are different applications, and everyone has a different end goal in mind.

If you allow for this kind of modularity, it allows people to adapt their courses to what they want to make in the end.

With regards to the pumping lemma or the master theorem, the point is, no one is forcing people _not_ to learn bout these concepts. However, people should have the right to choose what they learn to meet their own goals. If you can't make an informed decision by the time you're at college regarding what you need to study or what's important, then you need to think long and hard since life's about making decisions like this.

I mean, I had a choice to take automata theory. I did so, because I wanted a better understanding of how computation originated. I'm pretty sure others would be interested as well.

It's not that simplistic, data from MOOCs shows that short MOOCs (3-4 weeks) do better than long MOOCs (> 4 weeks), probably it's because the attention wains and there need to be other incentives for people to continue.

Surprisingly, the US military embraced online learning a little more than a decade ago. This was a boom for annual sustainment courses and the ability of highly active units to maintain OSHA, EPA, and other more job specific annual training requirements was greatly increased.

This great success in high level reporting percentages for training ended up driving even more training online.

This concept is a very slippery slope. The amazing statistics that accompany online training will provide very direct motivation to administrator(generals). Those administrators will easily come to the conclusion that training should be greatly standardized and moved online, due to the high level of efficiency.

In reality the DOD has moved some entire training programs to online. The navy has moved entire A and C schools(basic job fields) online. These are resident schools, and they are primarily schools based on hardware, that may also have strong software/theory based systems. The result is highly drilled students with almost no real world or problem solving skills in the field that they have been schooled on. I know this is anecdotal, but I have seen the change in the students first hand and the ability to teach critical analysis on a specific subject with out labs and 2 way live debate can not be underestimated.

Given the level of tracking of job performance the military does, and pretty much everything else, it seems quite surprising that the result of lacking "real world or problem solving skills" should show up somewhere. Has it?

If it hasn't, what would the military need to track for the alleged harmful effects of this to be visible?

Well, how do you define a course? For me, it's a series of individual lectures / modules that each teach you a given concept belonging to the same topic (that of the course). Concepts will often be dependent on each other, i.e. you will need to first understand concept A in order to be able to understand concept B.

So, taking apart a course and splitting it into modules might make a lot of sense, since it allows students to only learn what is most relevant to them at the time and be rewarded for their learning successes more easily and more often. Considering the horrendous dropout rates of most MOOC platforms this might be a very good way to incentivize students to "keep at it". However, this makes only sense if the individual modules are independent enough, so that students can learn about them individually (which, for many courses might not be the case).

For me, the concept of a "course" is mostly obsolete when delivering educational content online, and is probably mainly still in use because everyone understands what it means. It might even be a good example of skeuomorphism in this sense: http://en.wikipedia.org/wiki/Skeuomorph

I currently work on a (real-world) course platform that allows people to take "micro-courses" which can be completed in one single evening session and which (mostly) build upon each other. Like this, students can decide more easily which skills they want to learn, get results faster and spend less money on the course.

A concept which is similarly outdated (in my opiniion) is that of a scientific journal: These were created to (physically) group articles from a given field together so that they could be distributed to people working in that field. Digital distribution through the Internet makes this obsolete, since articles can be simply grouped using tags (or other meta-information). Like this, the concept of a "journal" gets replaced by that of a "view" on a group of articles that are similar in a given way, such as "all articles in the field of biophysics published last year", or "all articles [across all domains] that investigate sleep patterns". Using views instead of a journals is vastly superior (again, in my opinion) since it allows you to search for knowledge spanning different disciplines and fields of research.

Considering the horrendous dropout rates of most MOOC platforms this might be a very good way to incentivize students to "keep at it".

I don't think a high drop rate for MOOCs is a problem in the slightest. They should be evaluated like any other funnel on the web, where even 1% can be a good completion rate, rather than like paid lecture courses.

The dropout rate in higher education is a big problem because the marginal cost is high for each student (e.g classroom space, heating, cooling, electricity, parking, etc). The marginal cost for each additional MOOC student is almost zero so it doesn't really matter if there's a high dropout rate.

Yeah, re: the dropout rate, I think registering for a MOOC is more like browsing a book than registering for a college class. If I get through week 3 or 4 and take a midterm, then I find I tend to stick with the MOOC and finish it.

Without that momentum, it gets saved for later and I sometimes return to browse the content long after the class ends. More like a book.

The remaining "value" of journals is as a brand. One journal may have a reputation as being more selective than another, so a reader might assume the quality of the research is higher, or more novel, or whatever other attributes are associated with the brand.

Whether this is true or a helpful distinction is another matter, but it does distinguish the journal approach from what you describe.

"But the professors on the MIT committee that drafted the report argue that the numbers show that larger percentages explored significant parts of courses, which may be all they wanted or needed."

For me the main reasons for quiting a MOOC were:

1. I wasn't sure whether the course is for me or not. So I did the first weeks, but then decided that I'm not so much interested in the topic/course. Note that this mostly concerns courses which I would otherwise probably never taken.

2. Lack of time. As a doctoral student in mathematics, even having the time for MOOCs is rather a luxury. Thus it can easily happen that suddenly something important is due (for my obvious main obligations). Because MOOCs have a schedule to follow, I mostly see no other option than quiting the course (and feeling sorry for it).

Educational institutions needs to get with the times. Their business models aren't what they use to be. Governments have issued major cutbacks in funding here in Canada (I'm sure elsewhere in the world). These Universities are already largely funded by donations. They need to stop throwing all their money in buildings and start putting it into their technology stack online.

I do like the idea of modular degree's. Most are already fairly modular where you can pick and choose a variety of courses with a set of core required courses. The problem is they can't figure out how to do it online.

There is absolutely no reason why these Universities don't offer complete degree's online. Many offer a handful of starter/intro courses over correspondence but that's about it. They don't need to turn into MOOC's, they just need to offer quality online education - not 10 year old material. If they offer it, people will take their programs.

The only University in Canada doing complete online degrees right now is Athabasca. And some of the material is horridly out of date (COMP 268). Business opportunity? Hello?

> they just need to offer quality online education - not 10 year old material.

Quality and freshness are orthogonal, at least at the undergrad level. I earned an electrical engineering degree a couple of years ago. I also collect older texts in my areas of interest. Using only material from the late-1950s I could have passed my classes. Using material from the mid-1970's, aced them. (If you are curious, the reason for the difference is the development of the MOSFET.)

This is not to say that the material taught was 40 years old. Rather that 40 years ago the body of core knowledge hit what could be reasonably taught in a four-year time span. None of this "old" material has been proven wrong (at least at the scales undergrads deal with). The material won't need revising until either the silicon transistor goes the way of the vacuum tube or Laplace transforms are part of the standard high school education.

Indeed. For chemistry, for the basic knowledge you could go back quite a bit further, probably to the 1930's when quantum mechanics was applied to it. More recent for some of the lab work, there are some really neat analytical toys to play with.

On the other hand, anything touching on biology, including to some degree biochemistry, needs to be up to date. Very complicated systems about which we're continuing to learn major things all the time. At the other end, at least for MIT's core science requirements, physics of the 19th century are fine (mechanics and E&M).

In terms of offering entire degree online, I'm not sure existing universities have a substantial advantage that makes it important they do it as opposed to some new startup. OTOH, using online/digital resources in innovative ways to do what they do better is something which universities are uniquely positioned to explore.

These "zero to entry level developer in x months" startups seem to be getting results. Imagine an alternative semester, summer term, using MOOC technology, etc. They have the students. They have the teachers. There is a lot of tech out there waiting for good opportunities to apply itself to. All that's needed is some initiative and some stones.

Exactly - this is the first thing I thought of when I read this. Why not start by getting your existing degree programs online? Then they try breaking up into modules, experimenting with different teaching methods, etc.

Affiliated with Georgia Tech, Udacity offers online master computer science degree. They've mentioned that the graduated certificate would be same as those onsite students'. If that works well, I think more and more university will offers degrees via Moocs

In some sense, isn't what universities do already modular? To get a degree you don't take a single course consisting of 1200 hours of classroom time; you take 30 courses of 40 hours each. There's no reason to assume that 60 20-hour courses, or 120 10-hour courses, wouldn't be better - or perhaps a mixture of different lengths.

MOOCs and the archaic university structure in general should shift to this.

There's obviously something really important to have in mind and it's that knowledge is cummulative, meaning that in order to learn something you need to have some previous knowledge of something else, which would be a disadvantage for what MIT is proposing.

The articles and songs analogy is really accurate and necessary to point out. It's not that the internet is creating an attention deficit disorder in us, it's that it's giving us new alternatives on how to consume different products that satisfy us more.

The best excercise to take learning to a whole new level in any kind of field is "learning by doing", building something. Modules should not be focused in throwing lots of information to the student and ask him to replicate it, but it should be more about both teacher and students learning together, discovering and inventing new things.

I always thought learning should be more modular, which is why I broke down educational content on my site into modules. https://www.learneroo.com/modules (Some people were confused by it though.)

I like that MIT is willing to embrace the internet as a teaching channel and open minded enough to be willing to question its approach.

After reading the comments posted here I wanted to add that, as I'm sure most hackers/entrepreneurs here will appreciate, innovation is not a straight path and you cannot know where you'll end up from where you started. The important thing is to be willing to go out into the unknown and find a path to what works.

I can appreciate the comments about how this approach to teaching may not be the best way for everyone, but then again this would also apply to the "traditional" face to face teacher to students approach. I'm not going to be drawn in to making a judgement of which one is better than the other, but I sincerely believe that having diversity in how content is taught is better for everyone.

Also with regards to comments about MOOC completion I'll use my own experience as a guide when I say that some MOOCs are taught better than others and some courses I've been more interested in the content than others. This creates a situation similar to when I was at University and I would say that my grades reflected my interest in a subject and not my ability in it. The more I was interested in a subject the more I would invest extra time and attention to the course content. I have come to the conclusion that this was a lazy way of thinking and now I have a different approach to learning, I can see how this applies to MOOCs.

My take on modularisation is that it can be a good thing and I would point towards Khan Academy and how modular the concepts are and how they build as you go through a subject area. There you can find a supportive community willing to answer student questions focused around the module. Also as someone who is unwilling to go back to University to learn specific areas of knowledge taught at University I'm happy to hear that I would be able to pull out the areas I want to need to know when I need it.

Lastly, I would also like to say that perhaps there should be a MOOC on learning from a MOOC. I'm not sure these comments apply to the Hacker News community, but I've noticed that many people don't acquire an approach to learning. I myself have self taught on a great deal of subjects, including programming, and at times struggled through learning how best I learn. Amongst the insights to I've had into my learning style I would humbly recommend "Learn Anything: Hacking Your Education with Dale J. Stephens" from Creative Live. Besides learning a new portmanteau word "hackademic", I learnt a lot of new and useful strategies for tackling learning. The link is here for those interested: https://www.creativelive.com/courses/learn-anything-hacking-...

There's a MOOC going on that explores the "learn how to learn" theme:


Thanks Icbiazon, I wasn't aware of this!

I can also confirm that the Khan Academy model works, at least for younger children.

I'm still waiting for and/or planning to create a directed graph of dependencies between different subjects.

The Open University here in the UK has been doing something like this remotely for 30 years or so: http://www.open.ac.uk/

I'm doing a BSc through them and while there are certain groups of modules I have to pick from, perhaps half of my course could be done on modules that other BSc grads in my subject didn't do.

In undergrad, you can also take an "open" BA or BSc (depending on what modules you take) on almost any configuration of modules. You get the degree of BA/BSc (Hons) Open assuming you did the right mixture of module levels and got the right marks: http://www.open.ac.uk/courses/qualifications/qd

> Many professors said they were game to try the approach. In a survey included in the appendix to the report, 25 percent of MIT professors suggested that their classes “could benefit from a modular approach,”

Is this another way of saying that 75% of MIT faculty are opposed to this?

Or perhaps their courses wouldn't benefit from a modular approach.

A whole lot of STEM builds upon foundations, plenty of which are built up as a course runs. There's also stuff in many courses for which the long term motivation is not clear, and might not even be clear if an attempt at explanation was made.

Once you get past certain points I can think of various fields where modularity might work. Math, e.g. when you have to learn a particular thing to attack another subject. Engineering where, after the initial principles, there are various applications. Biology, a wild, wild world which after getting a very solid foundation I study in a "just in time" manner because it's changing so quickly and it's not my field.

I think we'll see a variety of solutions appifying different aspects of education that work best, while minimizing bureaucracy and barriers to entry. MOOCs are a clear progression from the university system, but it's also increasingly clear that a human element is needed in personal education.

That's why we're working on www.skillsesh.com

I had good traditional classes and bad traditional classes. Does this mean that traditional classes are archaic or only the bad ones?

What if some courses work better with the traditional format and those were the good traditional classes?

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact