Coursera’s Preview Mode(classcentral.com) |
Coursera’s Preview Mode(classcentral.com) |
I don't get this gloom and doom about MOOCs.
A substantial amount of people have transformed their lives by learning from MOOCs. I am one of them.
It’s not the usual suspects like, "people don't have self-discipline", "one learns much more in university", etc. that have limited the perceived influence of MOOCs. It is pure credentialism that is behind MOOCs not reaching their full potential.
Everyone is all about skills untill the hiring time comes. Now companies want the students of the best colleges, the best degrees, etc. Students with proper skills might not even get through the door without the proper degree.
I did projects with skills that I learned via MOOCs, I answered questions in interviews with the knowledge that I started getting from MOOCs. But it was my Master’s in CS which ultimately mattered in my getting interviews. In that degree, they still teach GOFAI, "soft computig", and fuzzy set theory, "expert systems", and more things from the 80s.
MOOCs matter, MOOCs are loved and studied by a serious set of students and professionals. But they still can't get you interviews for most roles in most companies. In frontier AI labs, they are now basically treating PhDs as the minimum qualification for most roles.
MOOCs + projects + self-directed learning, even if you are very good, offers you little in terms of career opportunities. That's why they have not been apparently "transformative".
I know the arguments about making hiring easier for companies, etc.
What hiring companies want is to get good employees who can do the job they are supposed to do, and they want to have as low a false positive rate as possible (hiring someone who can't do the job is very expensive).
I am not a hiring expert, but they probably get a lower false positive rate looking at school and credentials than any skill test they give.
Companies don't really care about false negative rates (not hiring a person who would do the job well) as long as they are still able to hire someone else who can't also do the job well. It sucks for the person who isn't hired, but not for the company.
- don't really care about false negative rates
That pretty much sums it up right there. And it's not hard at all to see _why_ this is the case.
There aren’t any metrics in this area that weren’t pulled out of someone’s ass, and we know this because collecting them correctly is prohibitively expensive and immune to automation.
Starting was fun, finishing was another thing. Credentialing another.
They helped get a lot of content out in the open.
The sharing of knowledge is great and should continue, hopefully with more modern digital interactions of 2025.
I remember on one of the first of these courses (AI in Berkley?), people were incensed that they would not get an official degree/certificate from the university.
Some were really mad, as if they had been defrauded or something.
Your point about certification is accurate, but I want to add another nuance.
Even if you manage to get a degree from a "good"/prestigious place, if you are "the wrong kind of person" that will only be held against you.
It's like succeeding at that is a violation of natural law and you need to be punished for the violation. It's almost religious and deeply guttural.
Of course that in places like the states, those people usually don't even get the change to graduate from those places since some excuse will be concocted during the application process.
MOOCs suffer from the online version of that.
However in places where admission is "blind" and the cost free or low this is very common.
Indian IITs (and other double-I institutions or similar) offer official certificates after MOOCs [0] and they are valued in regular colleges, and engineering students are often mandated to complete a MOOC or two every semester. The difference is that the final test is a proctored, on-site exam. The tests are held in major cities across India, and also outside India in places like UAE, for example.
Most MOOCs in the NPTEL platforms are boring traditional lectures, but some are extraordinarily good and at par with what Andrew Ng, Alfredo Canziani, Karpathy, or Jeremy Howard can offer. For example: Discrete Mathematics from IIT-R [1].
It's true, I took MOOCs because I wanted to learn things, and they were actually good at that. It was actually damn impressive how e.g. Andrew Ng's courses stripped away all difficulty which was not related to the thing you're actually trying to learn. The impact of that when you're self-directed is hard to overstate - nothing kills motivation like getting bogged down in getting software to work or confirming to some professor's favorite citation style.
Yet they seem to not understand that. They're all about my "career goals", they can't seem to comprehend that I would be there for any reason but impressing a fortune 500 company (which, from what I understand, it never does anyway).
> A substantial amount of people have transformed their lives by learning from MOOCs.
MOOCs are great and I agree: A lot of people have benefited.
The only disappointment is relative to the impossible hype cycle that was happening when MOOCs first entered the scene. You couldn’t open a MOOC relayed article or thread without some speculation that this would be the death of expensive university educations. That obviously didn’t happen, but MOOCs have been quite valuable on their own when well executed.
The primary disappointment I’ve seen is the half-baked courses that have been put out there. The first MOOC I tried was great and well run. The next two or three felt like they assigned some undergrad student a make-work project to put some old course materials on a website but they left out key parts. I remember it almost felt intentional, like someone didn’t want to put too much of the material online or the professor had objected to sharing their materials. They just wanted to say they got in on the MOOC train.
The later generation of courses that were made for the Internet were far better executed, of course, but they weren’t as plentiful and widespread as the hype predicted.
Maybe a bit early to declare that, as the wave of college closures has shown no signs of slowing after the Covid years [0] and is expected to accelerate further [1]?
[0] https://www.2adays.com/blog/college-shutdown-surge-update-th...
[1] https://www.fastcompany.com/91245055/higher-education-crisis...
But if we start benchmarking the effects of anything against its its Sillicon Valley-flavored hype, then, every tech/X-train will be judged as underwhelming, not reaching its "potential".
They also (at least in my experience) try to force you to watch via the browser. It's far more ergonomic for me to yt-dlp a full playlist and watch it locally.
Luckily MIT OCW exists and an almost unimaginably wide variety of educational material is constantly being dumped on youtube.
MIT OCW is great in this regard.
But can't say that I didn't appreciate the interactive quizzes in MOOC platforms. They are really effective for me.
Fill-in-the-blanks type programming assignments are useless. But the written code can teach you.
In-video quizzes in Coursera is also good for learning, since they ask you something that you should be able to solve with video content up to time t, but they explain it later at t+1. I think the co-founder also mentions this in her TED talk. This is good. I apply it when I am teaching someone in-person.
I also did a lot of online group-studies. These are really really great. All you need is a Discord server or email group, and Google Meet. This approach is better than the usual MOOC setup.
I also deeply value cohort-based MOOCs with video meetings (like Neuromatch Academy) or without (like Complexity Explorer).
And I agree with your sentiment against 2-4 minute videos; slow web design, sometimes requiring interventions with mouse is not something I cherish in some MOOCs.
Edit window is gone.
While not revolutionary, a recent improvement is AI-based review, which is much appreciated for it's near instant review.
From a transformative perspective, I like the AWS Skill Builder SimuLearn classes. They say teaching is one of the best ways to learn, and I found the chat-based role play where you are the expert to be very interesting.
When I started learning to code for data science I appreciated the lectures and examples. Once you reach a certain technical level these MOOCs aren't really helpful though.
I would disagree, I saw a lot of people, especially in the Data Science field that got up-skilled by back then free Coursera.
But in both cases, if you learn something from that course or video, and then go off and create something based on that, that something can be something employers care about.
These days I don't do any personal projects or contribute to OSS, I just list my previous jobs and can ace any on the spot questions.
Expecting to be able to list the completion of an online course as an achievement is kind of silly. They are very easy to skim through without learning, and are only really useful for yourself to learn, not to prove anything.
What would it cost without the overhead of a whole university around it? Would adults attend?Would employers accept it? Hard to change eduction more than a little at a time, since the buyer is literally buying a "you did it" certificate for 10s of 1000s of dollars, they are justifiably risk averse. The person examining the credential also doesn't want to do individual research on every single person's eduction.
- 5m subscribers, almost 2B views on youtube
- 30m enrollments
- 600+ courses every semester in 22 disciplines
Anyone from the world can signup. Proctored exams are optional and cost about $11 per course. Not taking VC funding and setting up local chapters for supporting students seems to have worked out well for them.
Website: https://nptel.ac.in/
ACM report about this from November 2022: https://dl.acm.org/doi/pdf/10.1145/3550473
Former-director of IIT Madras has talked about how NPTEL came together: https://www.youtube.com/watch?v=LV-QoGegFLY
If that was all they claimed there wouldn't have been early hype around transformative potential.
Coursera had 1.7 million "students" in under 2 years and was growing faster than Facebook. The President of edX talked about "disrupting" the entire higher education system. Stanford, MIT, Princeton, Brown, Columbia, and Duke all offered MOOCs with the idea that anyone anywhere in the world could now get an MIT/Stanford level of education without needing to be on-campus (or pay a lot of money in tuition).
From the 2012 NYTimes article on MOOCs:
'Dr. Agarwal predicts that “a year from now, campuses will give credit for people with edX certificates.” He expects students will one day arrive on campus with MOOC credits the way they do now with Advanced Placement.'
Never happened.
'Dr. Stavens promises more change, and more disruption: “We are only 5 to 10 percent of the way there.”'
Turned out we were already 100% of the way there after 24 months of MOOCs being launched.
What happened was primarily due to personalities and misaligned incentives.
The hype was massive. Everyone was supposed to be going MOOCs way. It was supposed to restructuring education system grounds up.
Now all I have seen is many did these big data/ data science courses and joined that great enterprise IT boondoggle of data processing/analytics.
In the early days, Coursera was transformative. I took some amazing courses and learned a lot.
...but the vision was always automation and as courses were rerun, the instructors were more likely to be disengaged, and live interaction via the forums fell. Finally when courses went to ongoing enrollment, there were no longer cohorts, and the experience was a correspondence course.
So in the end it was not transformative. It is pretty much just Youtube plus a website plus a payment platform. A less expensive, less interactive, unaccredited version of Phoenix University.
There's nothing wrong with teachers getting paid. It is the standard model. But of course the standard model is the standard model, not transformative.
And this is what the early success really was.
It'd be interesting to see if what they "up-skilled" on is now common knowledge in Data Science.
I find pedigree to be about predictive of performance as a d20 toss.
In my experience with hiring, though, it tracks. The number of times I’ve seen a MOOC on a resume is already a small number. Of those, it was usually 1-3 courses, not an entire program. Taking a couple courses does not compare to the repetition and layered learning of an entire program.
If someone showed up with an entire MOOC learning program and certificate of completion that I could verify then I would look into it. In my experience you don’t see this. You see people listing a couple MOOC courses right next to their other certifications. When you ask them questions about the topic they usually don’t remember much because it was a one-time course they took.
If you have a fraction:
really good students
--------------------
all students in CS
This is my personal experience.
(Past) company did hire from non-elite colleges in case-by-case basis, but one time we wanted 4-5 freshers, we did go to the elite college.
All students of elite colleges aren't better than all students of good enough colleges, but the fraction is what is different.
In a a typical CS dept of 50 students, you can find 40-45 really good ones in elite colleges versus 3-8 in good-enough colleges.