Now will be some rant and justification for the question, the question itself is in the last paragraph.
At the time (2012-2013) I was very hyped and enthusiastic, imagining new courses of the same quality and kind will be appearing at the same rate in future. By kind I mean university level subjects, fundamental topics that have high return of investment.
But it seems to me it went quite differently. I don't really see any new courses being published and recommended on the lists among good oldies. My own experience with stuff that came later also wasn't as pleasant, I started dropping courses much more often.
The platforms themselves changed significantly: 1. Switch from fundamental to hands-on subjects like technologies and frameworks. 2. Switching to paid model (not that I have anything against it, although being poor in a poor country I avoid paying as much as I can and haven't paid for a single certificate). This leads to less people checking out the course and giving it a "media coverage". 3. Switching from strict start-end date to "take anytime you want". Because of that the social element for me has been essentially lost, the forums are half-dead. 4. Increased amount of courses. Together with (1) this makes it hard to find something by random exploration. When I open edx computer science section I have to go through pages and pages of microsoft courses about their technologies.
This made me gradually lose interest in MOOCs and switch to books and self-learning. I occasionally go back to MOOC platforms or MOOC aggregators, thinking maybe I just missed something or something interesting came out recently.
Did you take any good (valuable, mind-expanding, long-term rewarding, intellectually stimulating) courses that aren't famous and have a high chance of being overlooked?
It's not necessarily a MOOC in the exact same vein, but given that the knowledge you gain from a MOOC is the valuable part (as opposed to any completion certificate or check mark), it's still extremely valuable and a great opportunity to learn.
https://pdos.csail.mit.edu/6.824/
One of the best courses i had taken on Coursera was the Algorithms series by Robert Sedgewick. Apart from going through the lectures, i had obtained a great satisfaction from solving the assignments. But this year, i have taken up the lab assignments for the following: https://cs144.github.io/ https://tc.gts3.org/cs3210/2020/spring/info.html
Both have been recommended earlier on HN.
Next on my list is 6.824
So far, it's allowed me to dive deep into how Map Reduce works, I even presented GFS in a local Distributed Systems group, and was able to understand Raft. I'm currently doing Raft labs (which is when the Pandemic happened). I've designed some sharding schemes in my previous company and worked a lot on Streaming systems but learning the theory behind everything is amazing.
If there are people here who would like to put a serious joint effort into doing this course, I would absolutely 100% be interested in it. Doing this alone from my apartment in a foreign country gets a bit tricky vs doing it in a group setting. :)
I want to do this in rust. Hit me up on twitter. Same user ID.
https://www.csail.mit.edu/person/robert-morris
https://en.wikipedia.org/wiki/Robert_Tappan_Morris
For anybody who has completed the 2020 spring version (the lab code recently changed), did you have any issues with the lab 1 test script? I think they left out the merging files part when porting to the 2020 version which causes tests to fail.
https://www.reddit.com/r/mit6824clojure/comments/hfm0hk/rmit...
https://devopswithdocker.com/
My first time doing online-learning, and registering for the university, but now I've done it I'm gonna work my way through a few more of the courses.
(You don't need to register, you can just start reading/completing the exercises, and use the telegraph chat-group for help. Mixture of English and Finnish, if you ask questions in English people will use that happily. You only need register if you want official credits, which might be useful in the future for me, but probably not people outside the country)
https://kodekloud.com/p/docker-for-the-absolute-beginner-han...
Now I know how people shopping around for a coding bootcamp feel.
Plusses:
* neat environment to work with: there's a drone and plane simulator based on Unity3D's Unity Player and NVIDIA PhysX that you connect to from Python, as well as another full simulator implemented with OpenGL (in C++).
* Considerable ground covered: representation, path planning, drone control, estimation, even basics of sensor fusing. Optionally, you can implement a full autopilot for a fixed wing plane.
Neutral:
* All the boilerplate code is provided (Python, C++), and you're asked to fill in the blanks here and there. So, one avoids a lot of hassle. On the other hand, you know, the instructions say "Compute the new variance matrix after the transition using the formula \Sigma_t = G \Sigma_{t-1} GˆT + Q", and then you plug in the line
and now you've implemented an Extended Kalman filter in C++. Right.Minuses:
* quite some mistakes. The course seems to get little love and few updates or bug fixes. If I had paid for it, I'd be miffed.
* not too much going on in the chat and discussion forum.
As to your points:
1. more "vocational" courses: Yes. Some of it seems more or less designed to quickly train the next generation of code monkeys.
2. paid model: Absolutely. Without the COVID-19 free month, this would have cost EUR 400 per month. Steep.
3. anytime you want: Agreed, the social element was pretty thin. I had the one month deadline before the billing would kick in, so that was useful for self-motivation.
4. more variety: yes, hard to identify good ones, particularly in combination with #2 (pay to study).
I was commenting to a friend the other day that it's insane that we hire 1000s of teachers to teach roughly the same math class 2x per year for 40 years. Instead I'd rather a fraction of those make a truly fantastic video course and then the remaining fraction could be hands on helpers/instructors for those who need help. This is kind of khan academy, but I'd like to see that happen in the public school system.
On the other hand, making a good mooc is a massive undertaking and it's unclear where does the money should come from. IIRC the initial idea behind moocs was that the certificate is going to be prestigious and people would want to pay for it. But what happened was that far more people would drop off early, and those who complete rarely bought the pdf.
Also I don't think online remote learning can substitute offline on-campus. Even very immaterial things like CS or math can benefit from in-person interaction with professors, which also further depend on the personality of the student. Immersion in the studying atmosphere is much deeper (you being among other students) which, for me anyway, is a huge motivation boost. Then there is also networking and the baggage of useful acquaintances that you come with when you graduate.
And if more teachers would do it like this, and they would put their videos online, it means that if your teacher's explanation didn't work for you, for whatever reason, you can try another one. Having the same thing explained from 2 or 3 different angles would probably help.
The usual response is: "Teacher's job isn't just explaining; it's also making sure the kids understood properly, explaining questions, giving problems, checking the answers to the problems, etc." Yeah, so we could have 1/2 of the lesson on video, and another 1/2 solving problems and answering questions. Still an improvement over how things are done now. And a part of "giving problems, checking solution" could be done by a computer, by the way.
The part of the response I agree with, is that you cannot eliminate the helpers completely. No matter how well the videos would explain it, some students would misunderstand it creatively, and if there is no one to notice and fix the problem, the outcomes will be bad for a fraction of students. Also, for kids, someone needs to be there to make sure they actually watch the videos, and don't play with their phones during the lesson instead.
Khan Academy is fantastic! My daughter loves it, but she also sometimes gets stuck on a lesson, and needs to ask me. In principle, this could be solved by collecting feedback, and creating optional extra slow explanations for people who got stuck at some point. Like: if you didn't understand the short video, watch this long one.
The way we do teaching now is just... dunno, pretending that we live in 19th century and can't do better. It still works, but has a lot of problems. For example, if your teacher sucks, and you can't change schools easily, you are screwed; and I have heard stories about incompetent teachers.
(Also, all textbooks for elementary and high school should be open-source, and freely downloadable from one web page. With the option for volunteers to fix mistakes, add more explanations, make translations, etc.)
You don't see any benefit to having one mega amazing course?
You think there's no way to overcome the limitation you mentioned (bad information) at all? How about a peer review process?
Millions of institutions exist. Even if you take 1% of them and tell them to create a freely available course, it's good enough. Beside, I don't understand how many public funded universities don't release everything they do or teach. It's just weird imo.
Why does education need to be expensive? Why should you have to pay for it?
It's one of those things that I don't understand the reason for being behind paywall.
I’ve also wondered about why publicly funded institutions don’t publicly release their courses and other resources. When the COVID lockdown started, I naively hoped that we would get some more pure math courses since now universities are teaching remotely but haven’t really found any universities doing that. I’m tempted to reach out to my alma mater and ask if I can audit their pure math courses that they’re teaching remotely in the fall.
It definitely seems easier to find CS courses online than for pure math. Maybe because there’s a wider audience for that, I don’t quite know.
Same as software libraries: we do invent new ways of doing the same thing from time to time.
Still, switching to using one of the few top courses by default would be a huge improvement.
I really enjoyed this class. It provided me with several very useful mathematical tools to think more clearly about important and frequent problems. Scott is a great teacher.
[1] https://www.coursera.org/learn/model-thinking
Can you please summarize what you took from the course? How would you use outside of the course the information you've learned? I would imagine something like "I've learned a model A that can be applied at real life situations such as a, b to get better insight", or maybe even "I can write a simulation to predict how some system will evolve in the future"?
Examples:
I mostly use MIT open courseware for self learning. A lot of course websites are also online and if I hear about an interesting course at some University I'll just google the course number. Stanford online and MIT ocw have a lot of lectures on their YouTube channels and I've found some interesting lectures just searching YouTube for a topic. When I wanted assignments and lectures for Russell and Norvig's AI textbook, I googled "Russell AI course" and stumbled on Berkeley's AI course. I've been enjoying it.
How has the social aspect + taking the course with others helped you in the past? I never found it that important. But I have always preferred self learning even in college.
3blue1brown for math (https://www.youtube.com/channel/UCYO_jab_esuFRV4b17AJtAw)
Everyday Astronaut if you are casually interested in rockets (https://www.youtube.com/channel/UC6uKrU_WqJ1R2HMTY3LIx5Q)
Bret Victor (https://www.youtube.com/watch?v=PUv66718DII&list=PLS4RYH2Xfp...) or Rich Hickey (https://www.youtube.com/watch?v=f84n5oFoZBc&list=PLXsqD83He-...) for programming and design.
Doing a full college course is such a time investment and it usually takes a few weeks to start proving the interesting stuff. I find it more fun to look to shorter content for inspiration. I might realize I don't understand the fundamentals of the topic and then seek out a way to study it.
I have never used much websites for the on-campus courses, they are rarely self-contained and require more organization and research to use. There is one good feature of those: they often have excellent assignments that you often won't find in textbooks, so I think can be a good complement to a textbook. One thing that bothers me though is that these can go down at arbitrary time and you can't be sure it works properly. E.g. a website for 2013 course may have some of the links broken and you will only find out when you get there. Still, better than nothing and I appreciate greatly such initiatives.
>How has the social aspect + taking the course with others helped you in the past First of all there was just a general excitement "A course just opened and there are like 40 thousand of us". If you spend time around people that would do things like that than I guess the effect won't be as striking. But 40k was like a half of the population of the town I lived in. It felt like a part of something huge.
In general, thinking that many people go through the exercises right now and struggle just as I am was somehow motivating and uplifting. Reading that something had the same issues as you 5 years ago vs right now somehow make quite a bit of difference for me.
Then there is interaction. Getting your questions discussed and helping others on the forum is totally different for me than just browsing through a dead forum section for a possible answer to my problem. I had a technique: when I completed something I would go to forums and read questions, and then try to answer them at least in my head. Doing so with a "real" questions that are waiting for an answer is more engaging because in addition to practice you also help someone.
And the most insane thing that only happened once: competition. I was doing Sedgewick's algorithms and the automated grader itself was brutal. I loved the course so much that I took it twice, the second time doing all assignments to 100/100. But in addition to the grader there was forum section where people would show their stats (how fast their solution worked). I remember cutting my time only to discover next day that someone did better, and then half excited half suspecting that the person simply trolling with fake numbers would go and try to optimize my solution even further. I would wake up at 5am to work on it before work, the whole place is messy with drawings on paper, etc. I cannot imagine doing the same thing when reading a book.
I've experienced links not working. In fact, Berkeley's AI course (https://berkeleyai.github.io/cs188-website/) includes 3 contests but the server no longer accepts submissions for testing let alone actually competing against other students.
There have to be a lot of other people like you and I out there reading textbooks and looking at OCW but never talking to each other. A simple discord server for MIT OCW would be something. I'm currently frustrated because I'm working through the text, lecture notes and problem sets for MIT 6.045J: Automata, Computability and Complexity[0]. But there are no solutions to the problem sets! It's a proof based course and I just want to sanity check if my logic is actually valid.
[0]https://ocw.mit.edu/courses/electrical-engineering-and-compu...
I think it's quite solid and valuable, has a good amount of exercises and I really like the frameworks/technologies they teach (React/Express/TypeScript/GraphQL).
Also trying to expand on my own the topics and concepts they introduce, so I think at the end it's gonna be a very good long-time investment.
For more specific stuff like Algorithms and Data Structures or Programming Languages knowledge, I usually prefer books over MOOCs.
I have let go of edx and moocs and now fish for online university courses which are put up on their websites. Notable ones are MIT, UCB, UC (davis/san diego i don't remember, but one of them puts up all videos as podcasts), CMU (some videos), Brown, etc. Then you have top profs putting up stuff online, like Sedgewick, Pavlo, etc. Personally I feel this is the way forward, pick and choose courses online directly from universities, instead of mooc platforms, for each subject. There is this github repo - https://github.com/prakhar1989/awesome-courses/blob/master/R... which lists out a lot of courses. Occasionally, if you browse enough, you might find some rare links like these courses on database systems - https://bigdata.uni-saarland.de/datenbankenlernen/ and https://www.youtube.com/channel/UCDIJAkbAr53I4fggNsbzdrA/pla...
NP-hard, at a rate of one problem per week.
https://www.executeprogram.com/
Also, Dan Boneh had a YouTube thing up a few months ago where he said that Cryptography II would be coming to Coursera this year, so that will be interesting too.
https://youtu.be/eQdkZRLD09M
Most of the replies here are more suitable to "what are good moocs", which is fine and what I more or less expected anyway.
As to regularly resubmitted post, this kind of touches the issues I asked. If there aren't many (or none) good new materials then there is no much point reposting same things again and again, similar threads from 2016 or 2018 will do. Some communities post a compilation of what new courses are gonna start soon. I don't find these useful, because they end up being just a huge dump of everything. I guess what I would find interesting is a recurring post about learning that goes like like "what are you learning/learnt recently/this year/this season" and why a mentioned source of learning was good. This naturally will focus on quality rather than quantity.
There might be some good content, but the movement is definitely dead.
Sebastian Thrun / udacity never actually delivered. You still need to get a bachelor's degree by a regular university to get into oms-cs.
Sad.
I feel like this is absolutely not the case. For software development and adjacent fields (UX/UI design, 2/3D graphics, editing, ...) you definitely need the skills and the degree is a benefit (a door opener) but not a necessity.
That said, if you expected a coursera course to replace a university, you'd first have to deliver on reducing the amount of people who cheat (i.e increasing the prestige of the degree). Nowadays, many IT professionals don't even know what CCNA, AWS CSAP, or RHCSA mean (and all of those are somewhat prestigious in the IT field)
The idea of getting online certificates instead of degrees is dead. The idea evolved into learning skills instead of gathering certificates, and demonstrating your skills. Think Pluralsight, LinkedIn Learning, or LinuxAcademy, which are very much alive and kicking.
But i am doing 'mountains 101' during the pandemic https://www.coursera.org/learn/mountains-101
[0] https://cs193p.sites.stanford.edu/
This answer doesn't answer the question that was prompted, but I volunteered for Stanford's Code in Place (introductory CS) and it was kind of amazing getting to be on the other side of the MOOC experience as a TA. I lead 5 assigned class sessions with a small group of people (never more than 10) and held 5 other office hour style sessions as well.
It was a really rewarding experience and I think a lot of people who can volunteer for it should consider doing so if Code in Place continues.
CS Video courses: https://github.com/Developer-Y/cs-video-courses
Math/Science video courses: https://github.com/Developer-Y/math-science-video-lectures
Some of Electrical/Mechanical engineering courses: https://github.com/Developer-Y/engineering-video-courses
If you don't have problem understanding Indian accent, India's NPTEL publishes University level courses on this site: https://nptel.ac.in/course.html
You can take Indian IIT courses online with assignments (and proctored exams if you are in India) at following site: https://swayam.gov.in/
MIT's Intro courses in this department have long been excellent and very neuroscience oriented.(The caveat is most of what's on OCW here is fairly old.)
There's also at least one course on edX https://www.edx.org/course/minds-and-machines although I don't know anything about it.
The creator of fast api hangs out on HN. You can find his posts if you search HN for 'fast api'
The course is project oriented and quite concrete (focusing on practical/directly usable knowledge, before diving into impl details and math) so I liked it a lot more than anything I had tried before (courses on Coursera, studying on Kahn Academy).
Not only did I actually learn, the course gave me the skills and motivation needed to build my own toy project[0]!
@OP: Give Fast.ai's course (part 1)[1] a try. You'll train your own image classifier by the end of lesson 2. At that point you'll know if you like or dislike the course's style.
[0]: https://ml-hot-or-cold.projects.chrisjeakle.com/ [1]: https://course.fast.ai/