The End of College? Not So Fast | Hacker News Reader

The End of College? Not So Fast

chronicle.com

43 pointsilamont43 comments

Course completion percentages are a ridiculous measure of success. Of course free online courses are going to have a lower course completion rate. I don't know how anyone could be surprised by this, or why anyone would care.

This isn't a problem. If I start a MOOC and don't finish, I have gained a little knowledge and lost nothing. College completion rates are a problem, because if I start college and don't finish, I'm saddled with thousands of dollars in debt and no degree. Completion isn't a problem with MOOCs, it's an advantage of MOOCs.

Missing out on the other aspects of college life is a mixed bag. Networking and lab work are definitely downsides to MOOCs, but these are potentially solvable problems. But there are serious downsides to college life too. I won't be sad to see the binge drinking and sexual assault culture of "greek" life die out, and I'm not going to shed any tears for sports stars actually having to focus on academics.

But the real crux here is that if trends continue, the cost of college is going to make this all irrelevant. It already is for many people: the author's quote that most students graduate with $30K of debt, as if that's a trivial sum, shows a stunning lack of awareness.

What none of the author's facts capture are the students who either don't start or don't finish because of cost. Tuition keeps rising and financial aid continues to be inadequate. There are only two solutions I see to this problem: funding gets better or we go with MOOCs. MOOCs aren't a perfect solution, but it may be the only solution we get.

I agree, comparing completion rates for MOOCs and 4-year colleges is like comparing the percentage of couples still together 4 years after a first date (MOOC) vs 4 years after getting married (college): I don't think we would WANT those numbers to be the same.

For me the real benefit of college was being put with others into the same class, the access to research labs, the face to face interactions with professors, and the access to face to face academic focused student organizations. (Honestly I should've done far more on the networking side of things.)

Those are not going to be easy to replace, but there should be a compromise where you can get these things without the greek life and less 'sports first and foremost' culture and with a massively reduced cost. MOOCs can definitely plan into this and even replace a lot of it for self starters, but I do think I would have been worse off without the social aspects of my university years.

For me, this was more the case in graduate school than as an undergraduate. While I did like my department (mathematics), my advisor, and the other students in my class, the majority of my undergraduate studies really did end up being direct completion of coursework (with the occasional group project). The social "coming into your own" aspect of university was there, but I'm not sure it's worth the price that we seem to attach to it. Maybe this is just the reality for state schools.

On the other hand, post-calculus/diffyq mathematics can be a difficult subject to grok on your own, without at least some face time with a subject matter expert to refocus. I can't predict with confidence if I really could have MOOC'd my way into graduate school. I thinking that it is unlikely. This is probably also true for the traditional physical engineering disciplines (electrical, mechanical, civil, aero, etc.)

There are plenty of schools without greek life and sports first cultures. Most liberal arts schools for example.

That is true. But liberal arts schools are also the least likely to be priced out by MOOCs. Liberal arts schools are more affordable than most, if you can get in.

> I do think I would have been worse off without the social aspects of my university years

What aspect of face to face interaction makes it worth seeking?

As an introvert, a remote worker full time, and a traditional BS in CS from a state university, I am more and more convinced the needs of many for "face time" is just because anyone over the age of ~20-25 grew up without ubiquitous communication without faces.

I'd cite my brother (age 12) and his generation, who seem to have no problem basically never seeing their peers from class anymore, but will play video games with them endlessly online. During summer break last year he barely went outside, besides what physical activity his mother compelled him to, which was not very much different from what I did, the difference being that his entire peer group is behaving the same way and from my impressions by him it is the norm rather than the exception now.

I fully expect that generation to have no issue doing collaborative learning using MOOC resources and the Internet instead of brick and mortar buildings in front of lecture podiums. Throughout history we have learned many a time how adaptable the human psyche is to its environment, and how much of ones assumptions on human behavior can be challenged and changed by the world changing around them. I think this is another one of those cases - because articles like this one are talking about college ten years from now, for children born ten years later. And that makes a huge difference in the Internet age.

MOOC completion rates matter if we live in a world where they have replaced traditional colleges. Not the overall completion rates, because that includes people who start them on a whim - which is possible since the cost of entry is so low.

But if we discover that a significant number of people who would have completed a traditional college program are not completing a commensurate MOOC program, then that would be a problem. That would mean that more people are getting even less education than they are now. I don't think we have the data for that, though, as I don't think we have a large enough population of college-age students who opted for MOOC-only programs instead of traditional college.

To be clear, I am not arguing that all of the problems you pointed out are not problems. They are problems, and they are huge ones. Student loan debt is a paralyzing social and economic issue, and the cost of college is an enormous barrier to social mobility.

"Even less education"

I think we definitely need less education and more learning done.

The point should never to be more "education", that just gives you ammunition to defend that you've done something meaningful in life at a dinner table.

The point is to use that education to do something useful. Education is a means to an end and not the end in and of itself.

I think a charitable reading of what I wrote would imply "less education" to mean "less learning." If you don't agree that one is implied by the other, then just pretend I said "less learning".

The MOOCs force one to "enroll" in the class to have a look at it or watch a few videos, to juice their enrollment numbers. Both the enrollments and completion rates are essentially meaningless.

As a person who took several MOOCs and completed none of them, I can honestly say that you're right on the money.

I think the only real problem that the other comments allude to is the social aspect of college life. People think this because College is the de-facto solution socialization with similarly aged peers.

I believe that many organizations will pop up with the death of the university to solve these problems. Think about it, all you want is a place to talk to others, and get info on jobs or other opportunities.

In previous ages, a lot of this was not done at university but at coffee shops or similar institutions. Just because you don't have college doesn't mean that you can't make connections. Heck, I believe if the college fad dies down, with the amazing social networks that we have now, it will be super easy to create meet up groups, and hey, it could even open up the potential for specifically themed coffee shops. For example, there is a coffee shop in my area that caters specifically to programmers, there's a podium if anyone wants to present a talk, there are private tables further away from the podium, there's free internet, the tables can be flipped, to give you glass surface that you can write on with a marker ( the list goes on ).

I feel that MOOCs will push towards students searching for mentors, who they really respect and hence can more effectively learn more. Furthermore, it will lower average student debt as well as make many aspects of selective college life publicly available. This will not however happen over night, but I see MOOCs as the harbingers of radical change in education that has been stagnated by the entrenchment of the ivory tower into the very heart of what education means today.

after running a few MOOC classes and contrasting them to brick and mortar classes the real problem in my opinion is assessment. the material we prepared was equal to what we would teach in a college class, but the assignment quality is not. this is I think a fundamental problem of scale. the only assessment tools that we have are multiple choice and regular expression matching short answer quiz questions. coming up with challenging yet fair multiple choice questions that let the learner express they have mastered a concept is extremely difficult. and to make the class rigorous, MOOC instructors would have to re-do that work for each class, which defeats another scaling point.

I think that MOOCs solved the wrong problem and because of that aren't the solution to the class. the MOOC is an iteration on the textbook. a domain expert puts a lot of time into the layout, design and presentation of materials and then creates a relatively static asset out of it. it's less static than the old textbook, but you can't walk into the professors office and ask questions for an hour face to face either.

we haven't figured out how to make assessment scale. we haven't moved far from the model we used for thousands of years, which is to have a student stand and talk to their instructors/mentors, and their mentors to make a judgement. how can you get a deep sense of a students capabilities this way, online?

I don't think it can't be done, I just don't think that MOOCs have really tried, and my current cynical understanding is that they don't really want to try. Platforms like edX, coursera, and OCW seem to stop at "you got video and text what more you want", and stuff like minerva is using technology but keeping the same scaling factor between instructors and students.

This is spot on. In fact (having TAed a 400-student intro course once) I'm not even sure if assessment scales properly in the larger brick and mortar classes. The professor was using an iClicker for pseudo-pop quizzes at the beginning of class but I noticed that a lot of students would simply give the same answers as their friends next to them. Likewise for the short essay questions in the final and midterm, they had created a massive Google Docs database combining lecture notes which (while on some level showing admirable initiative) also resulted in a number of identical (and, indeed, identically-wrong) answers. I've never taught a MOOC but it seems to me that this sort of hive-mind tendency would be magnified.

I think the hive-mind tendency would be mitigated by the fact that in a typical MOOC the students don't know each other, so that sort of coordination is difficult.

> I'm not even sure if assessment scales properly in the larger brick and mortar classes

you're right, it totally doesn't. the system lives with it because the huge intro/survey classes deal with such a superficial level of information that you can measure it with MC/short answer.

upper level classes deal with it by increasing the staff/student ratio, either by lowering the number of students or increasing the number of staff.

so the MOOC classes where the answers can't be graded by programs, MC or short answer, become this hybrid of advanced material with intro level assessments. I don't think this is to the students benefit.

I also think that the MOOC process overall has been corrupted by the same force that corrupted universities, so I don't see the system as providing a more pro-student alternative anytime soon. I don't know what would.

In some cases it works (beyond the multiple choice quizzes). In the algorithm courses I took, you have to implement a algorithm, and input a number that proves the implementation is correct. For example, on a graph with, say, 500,000 nodes, what are the sizes of the five largest strongly connected components. This is impossible to guess.

I also did a course with peer grading of one assignment (the Financial Markets course from Yale). It actually worked surprisingly well. You asses 5 other students' 300-word summary of a paper (based on 3 fixed questions). I didn't think it would work, but I've changed my mind.

Mind you, cheating is still a problem (e.g. somebody posts the correct answer online).

And, an even bigger problem is proving that you completed the course, not somebody else.

"the MOOC is an iteration on the textbook." Spot on, and this is what politicians and administrators don't understand. How many kids learned calculus out of a textbook alone? Not that many. But it's cheap.

MOOCs don't want to try, because the multiple-choice mode is cheap. That's the major appeal of MOOCs: cheap.

A few alternatives: SPOCs (small private), with discussion weekly and weekly face-to-face meetings between profs and students. I know some folks who ran one; it was effective yet very time-consuming. Programs like NetMath at University of Illinois: self-paced, hires real people to correct & provide feedback on written student work. Online coaching services like CoachingActuaries have adaptive exams and diagnostics, which is at least a step up from the static multiple choice.

None of these is free.

Adaptive exams can be a huge win. I know they can be hard/time-consuming to do really well, but it's fairly easy to make static multiple choice a little more dynamic.

For example, imagine an end-of-section quiz with 5 multiple choice questions with 4 answers each. The static way is each student sees the same 5 questions with the same 4 answers to choose from.

Instead, you could have a bank of 100 questions, in buckets of 20 covering the same material. So, each student sees one question from each bucket, drawn randomly. And then each question could have 10 answers, with 4 randomly displayed for any given instance of the question, sometimes with multiple right answers, sometimes none, etc.

This would require more work, but effort spent on assessment should scale with the number of students, just not linearly. (Right?)

The goal is not to make cheating impossible. The goal is to make cheating hard enough that it's easier to just learn the material and answer the question.

Furthermore, once you have that bank of 100 questions, you can start running analytics on it to see which variations are harder or easier for students to understand, which topics might need a better explanation in the teaching section, etc. You could eventually be able to generate adaptive exams from this data.

Sal Khan of the Khan Academy talks about all these things. I recommend his book: The One World Schoolhouse. (http://www.amazon.com/The-One-World-Schoolhouse-Reimagined/d...)

my problem with this is that it's still MC and short answer. the lifeless creatures from another dimension at ETS have been iterating on adaptive tests with the GRE for a long time, and yet, it's still the GRE...

Since he makes a lot of weaker arguments, I'll focus on his strongest argument:

> To think that someone [...] with only a high-school diploma [...] could motivate himself to complete a large number of MOOCs is naïve [...]

He comes up with some spurious statistics to support this assertion (about MOOC completion rates), but it's much more likely the reason he believes this is because of his experience as an educator.

I think this is a valid criticism. One important aspect of college is that you have something on the line (loans), and you're placed into an environment where the social context is "this is what we're doing now, we're completing courses and learning." And this social context propels people who would otherwise not have the self discipline to learn on their own.

The problem here is that the author treats this challenge as evidence that college is the best solution. If MOOCs can't answer it, college will continue to be dominant. Once I've spelled it out like that it should be obvious he's assuming a false choice.

What other options could we come up with to solve this interesting problem he's posed? MOOCs in their current iteration don't compel you to keep learning. But rather than assume college is the best option by default, let's keep thinking of other ways we could figure out how learning could be encouraged systematically.

I like MOOCs, but I am using them as continuing education while working, not as a replacement for college. For that purpose, it works quite well [1]. However, one problem is when you study a subject and get a bit stuck (currently taking Pricing Options [2], and there are a few tricky parts so far). Ideally I would like to ask questions and based on the answers continue to ask questions until I understand. As it is now, it is pretty painful to try to use the forums, and maybe get an answer. In person (or live video) with a professor or TA would help a lot. But of course that doesn't scale well.

Also, the fact that the completion rate for MOOCs is very low doesn't mean much. It is free to sign up, so many (including me) sign up only to check out a course, or to just watch the lectures.

[1] http://henrikwarne.com/2012/05/08/coursera-algorithms-course...

[2] https://www.edx.org/course/v2/pricing-options-mathematical-m...

Henrik_w, those are some good points. The potential of MOOCs to deliver low cost education has been hurt somewhat by poor design choices. Classes have been simply "scaled up" by professors without much change to their structure. Without physical classroom interactions, learning online is much more difficult and needs a a collaborative support community to aid that process.

A few friends of mine from high school and I build a company called StudyCloud that plays in the space. To solve the issue of lackluster collaboration and low retention, our product builds a social community for each individual course, having other students and teachers there for support and encouragement. These individual social communities are integrated together for the student into a complete course overview. This allows a student to keep track of all their classes, seeing easily how well they're progressing at a moment's notice. This at-a-glance functionality is coupled with extensive notifications that help bring students back to the platform regularly to engage with one another more frequently than might typically occur.

We've been working on StudyCloud for over a year now, with most of our growth in the Midwest. We are working with a few universities on hybrid course design to further improve not only the product but best practices for teachers. I would be happy to share more if you interested.

www.mystudycloud.com

Very interesting. I have been on a kick recently on how to really attack the college market with a different offering, and had thought of some of these same points. (Which you are executing on! Well done!)

My concept was to try to create a study abroad program, which combined a concentrated "learning community," along with some MOOCs, and on-site tutors/navigators. I think you could run this at a RADICALLY different price point than most colleges, and therefore nip at the edges of the market. Basically take a semester or two share from traditional colleges.

Of course, the challenge is in accrediting and ensuring credits would transfer back to the college. But it just seems when colleges are charging $40k / year tuition that there is some room to attack this on the edges. And the college students then get the benefit of a fun study abroad which would be even cheaper than home....

This seems like a lot of cherry-picked data that ignores many other possible explanations.

MOOCs may have a low completion rate, but if there's no cost to enter, and no penalty for leaving, then it's easy to attract a lot of people who might have only had a passing interest and decided the subject wasn't for them, or who got busy with other things and stopped paying attention to the course (I'm guilty of that). It's not really an accurate comparison to a class where people have thousands of dollars invested in getting a grade.

Also, the fact that one other country has college tuition prices that are approaching the US ($14,000 is still relatively low, even for many public schools if we're talking out of state tuition) is not a good argument that US schools are reasonably priced.

Looking at free MOOCs only is indeed cherry-picking data. Instead, one interesting example is remedial math at San Jose State University: a class students paid for and needed for college, implemented by Udacity. It went so poorly that the experiment was stopped [1]. The pass rates and test scores were so low that the university couldn't, in conscience, continue.

There were some implementation problems, which have been addressed in a summer 2014 implementation [2]. It's worth noting that SJSU let 40% of students drop the summer course, which really increased the pass rate!

[1] http://www.mercurynews.com/ci_23688069/mooc-mashup-san-jose-... [2] http://www.informationweek.com/software/mooc-math-students-b...?

They tried a MOOC for remedial classes? That's really baffling to me.

Yes :( Unfortunately, a lot of politicians and administrators figure MOOCs are best targeted at the classes that have vast herds of students coming through, like remedial math and writing, rather than those little upper-level classes for advanced students (that profs love to teach!). But you can see the problem: the major predictor for success in a MOOC is study skills + maturity. The major predictor for being in a remedial math class? The lack of effective study skills. If these kids had great math self-teaching ability, they would have taught themselves out of remedial math a while ago (like half the kids in calculus).

The end of this article describes a "future situation" that is actually the current situation exactly:

"Students with means of their own will have the option of attending a four-year institution and obtaining all the benefits that come with it. Poorer students will obtain their postsecondary education from MOOCs and their credentials through badges."

How is this any different than what exists right now? The availability of student loans allows low-income students to attend college but brings with it a lot of the ills of MOOCs. Low completion rates being the most common example.

Why should anyone care about low completion rates for MOOCs?

Completion rates don't matter as much when the content is free. I've signed up for 10+ of MOOC courses but only completed 2. I meant to just point out that if you consider a low completion rate a negative factor than the current system of easy debt for low income students has a similar problem.

I don't think this future is implausible, but I think the MOOCs will not be able to be free. There are costs associated with creating a good online course, and those costs have to be deferred somehow.

Perhaps these MOOC developers will be able to attract patrons who will cover the large up-front cost, and ads could cover the ongoing support, but one way or another they won't be completely free.

As for whether recent highschool graduates will be able to complete a traditional college degree worth of MOOC courses, it will depend on how much they actually help those graduates get jobs. We went to college on the insistance of our parents that it would improve our quality of life above theirs, and the promises from colleges that a degree would make our careers.

For some of us, that was true. For others (particularly in recent years), not so much. MOOC badges will have to offer more than just promises to justify the time and money we put into them.

Right now, Coursera offers certificates for degrees that cost money. They also use this to create capstones, consisting of 4 to 5 MOOCs and a final project.

I believe there was someone who completed all the necessary open courseware for an MIT "degree" (obviously substituting courses when not available) but can't seem to find the link.

It's not clear to me how you prove that you did the work for the course. As far as I can tell with Coursera, all it takes is for you to input the correct answers in order to pass (i.e. you may have found the answers online, had a friend help you etc). Identity is still unsolved.

Coursera often changes the numbers/units for math problems and switch up problems on every attempt. You can cheat at a 4-year physical college, but it's harder. If you are trying to cheat on courses, you probably shouldn't be in college :)

You are right that MOOCs can't be free forever. In particular, right now MOOCs are subsidized by tenure: it's often tenured profs who can get a course release who develop MOOCs, and they do so because they're interested in education, rather than from a profit motive per se. Most universities swallow the cost because it's good for marketing at this point. This won't be true in the future.

Many universities are looking to develop their own branded online programs, which may or may not look like MOOCs, so that they can have an additional revenue stream. This pulls them away from platforms like Coursera. I am curious what will happen as the ed companies' interests diverge from those of the university.

There are costs assorted with creating good software as well but the community have found ways to still make FOSS a viable reality. Is there any reason the same methods can't be modified for MOOCs?

There effectively is a model today where you can take the course for free but you pay for a "verified certificate." The problem (as is often the case with freemium) is that it's not clear what the value of a verified certificate is for the vast majority of people. There may be scenarios such as employers wanting to see proof of continuing education but they're the minority.

Most FOSS software is supported by corporations who use that software, and contribute their changes back.

If we parallel this to MOOCs - basically companies interested in accepting MOOC coursework would want to produce & manage the MOOC content, or hire it out.

It could work, but would it? Tough sell.

I feel author somehow saying just badges for crediblity is wrong, it has to be compared what we have today in university as credibility. It is just a pdf certificate and a grade point with university name. It doesn't distinguishes student very well.

With people moving towards online education there will be much better education data gathering and much better way to present them. Kevin carey pointed out this https://accredible.com/10000005 for replacement of current education certificate for data gathering about student learning and then using the data collected in better way for analyzing student performance and for employablity. P.S (I work at accredible as SE)

I don't disagree with any of the other criticisms here, but seeing the headline and source together was enough. It calls to mind a harness-maker sneering "Horseless carriages? Nobody will ever want to drive those!" Who are they trying to convince?

"To think that someone almost three decades younger, with only a high-school diploma (or perhaps even less education) could motivate himself to complete a large number of MOOCs is naïve, and is not borne out by the evidence."

Why is this naive? I think this attitude is prime evidence of the REAL problem of the education system today - that we don't motivate younger people enough to be curious. no amount of MOOCs will solve that, only good and devoted teachers combined with a renewed curriculum will. MOOCs are great for subjects involving computers and mathematics, but for subjects like sociology and biology I would deign to argue that even traditional college is better equipped.

Reply on news.ycombinator.com