Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Coursera’s Preview Mode

classcentral.com

Post reply on HN
Print view View original

41–50 of 61 posts

Re: Coursera’s Preview Mode

#41

Post by recursivecaveat »

Sometimes I wonder what an MOOC-like education that went all-in on signalling/credentialism would look like. You convince some real down-on-their-luck university to lend you their name to provide eduction for lower income adults, you rent a big empty room with folding chairs in downtown where people actually live a few nights a week, a contracted PhD with a tweed jacket and a blackboard lectures in person, curriculum straight from an old edition textbook with pen-and-paper tests.

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.

Re: Coursera’s Preview Mode

#42

Post by __rito__ »

fc417fc802 wrote: post #32
__rito__ wrote: post #22

> MOOCs never achieved the transformative potential promised during the early hype. 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 credentialis…

I never liked MOOCs because I didn't see the value add over youtube. At the end of the day I just want free and unfettered access the full list of video lectures. Accounts, course enrollment, fixed sequences, it's all just friction that gets in the way of actual learning. 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…

YouTube + a maintained webpage with slides, assignments, problem sets, midterm and final papers etc. is a fine combination.

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.

Re: Coursera’s Preview Mode

#43

Post by rustystump »

RHSeeger wrote: post #39
cortesoft wrote: post #29

Earlier quoted context omitted.

> 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. 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…

- a lower false positive rate looking at school and credentials than any skill test they give. - 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.

Is there data that backs this? In my anecdotal experience it is all over the place. Unless i specifically know a given program it is a total crapshoot on if it means anything.

I find pedigree to be about predictive of performance as a d20 toss.

Re: Coursera’s Preview Mode

#44

Post by vintermann »

__rito__ wrote: post #22

> MOOCs never achieved the transformative potential promised during the early hype. 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 credentialis…

I have also taken many Coursera courses, and paid for them too. I was considering taking another, but I can't reward this sort of "leadership". Congratulations, they sold out on all their promises and in return they're about as profitable as a mid-sized construction company?

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).

Re: Coursera’s Preview Mode

#45

Post by vintermann »

t_mann wrote: post #38
Aurornis wrote: post #37

Earlier quoted context omitted.

> I don't get this gloom and doom about MOOCs. > 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 o…

> this would be the death of expensive university educations. That obviously didn’t happen 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...

It isn't MOOCs which is killing them in any case, though. Or do you disagree?

Re: Coursera’s Preview Mode

#46

Post by mgnn »

When did Udacity fell off the MOOC discourse? That's my first experience with MOOCs; have fond memories learning robotics with Sebastian Thrun there. I used to prefer its short lesson and frequent quiz too.

Re: Coursera’s Preview Mode

#47

Post by gopher_space »

cortesoft wrote: post #29
__rito__ wrote: post #22

> MOOCs never achieved the transformative potential promised during the early hype. 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 credentialis…

> 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. 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…

Nobody gets in trouble for a bad hire from a good school, and nobody wants to read your resume. End of story.

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.

Re: Coursera’s Preview Mode

#48

Post by __rito__ »

77pt77 wrote: post #30
__rito__ wrote: post #22

> MOOCs never achieved the transformative potential promised during the early hype. 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 credentialis…

> 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 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 poi…

Thank you for adding the nuance to the conversation, although I would have liked some elaboration on that.

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].

[0]: https://onlinecourses.nptel.ac.in/

[1]: https://nptel.ac.in/courses/106106183

Re: Coursera’s Preview Mode

#50

Post by __rito__ »

rustystump wrote: post #43
RHSeeger wrote: post #39

Earlier quoted context omitted.

- a lower false positive rate looking at school and credentials than any skill test they give. - 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.

Is there data that backs this? In my anecdotal experience it is all over the place. Unless i specifically know a given program it is a total crapshoot on if it means anything. I find pedigree to be about predictive of performance as a d20 toss.

I have been on the hiring side, too. And the difference between a good enough college and an elite college is the density of really good students.

If you have a fraction:

    really good students
    --------------------
     all students in CS
Then this number tends to be much higher in elite colleges than good enough colleges.

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.

Post reply on HN