The Stanford Education Experiment Could Change Higher Learning Forever | Hacker News Reader

The Stanford Education Experiment Could Change Higher Learning Forever

wired.com

111 pointsnigma30 comments

I want to address the following quote from the article: "During the fall term, the Stanford students taking CS221 preferred watching the KnowLabs videos. Thrun says this improved their performance."

I took the official Stanford offering this past fall. I am not a Stanford student (anymore), but I enrolled as part of SCPD. As a result of the official course starting two weeks earlier than the public offering, the course staff usually didn't post up the YouTube videos in time for me to do the assignments, so I relied primarily on the lectures. (Actually, as a distance student, I relied upon the lecture recordings.)

I eventually did catch up on the YouTube videos, and I can attest that they are indeed better. However, they produced superior results mainly because they covered more material. For example, in one case, I did over two pages of error-prone calculations because I didn't know the mathematical shortcut that was taught in the YouTube videos. And I gasped as the professors spelled out in detail how to do problems that I had to think hard about how to do. Indeed, one problem that I was pleasantly challenged by on the midterm was simply not covered in lecture; in contrast, the professors explained it tutorial-style on YouTube.

Using the terminology of the course, I would say that the YouTube videos "overfitted" to the homework and exam problems -- so of course the students preferred the videos!

I do suspect that short, interactive, and extremely polished videos will be proven to be better in most cases, but unfortunately this experiment cannot be used as a basis for forming that conclusion. Clearly, the lectures were handicapped.

And to be honest, I am skeptical of how videos would perform against some of the tremendously excellent lectures out there, such as the introductory computer science lectures that are posted at Stanford Engineering Everywhere.

With all this said, I had a great deal of fun taking this course. I applaud the course staff for doing a stellar job, and I am grateful to be part of what I am certain will be a historical experiment.

'Only 10 institutions in the world delivering higher education'?

Not unless universities abandon the teaching of history, literature, philosophy, politics, economics, whatever etc., or perhaps Thrun is thinking they'll be a gigantic leap forward in developing machine intelligence able for instance to assess students' essays in these subjects. If anyone's seen the slightest evidence that this might happen I'd be more than interested to see it.

Indeed. And not just those subjects. Computers are still a very long way off from judging the correctness of say a human written proof or solution to a physics problem. Even in technical subjects universities will be impossible to replace without major major leaps in AI.

I think education at a great University provides primarily four benefits:

1. Great instructor-driven education 2. Great student peer-group-- folks that are truly smarter than you and teach you to think in ways you hadn't imagined (and those you pay attention to, because they might someday change the world) 3. Great networks/connections and maybe even a brand 4. An opportunity to immerse in great research.

I think online education courses do a great job of approximating 1. They sort of try to do 2. And they fail at 3.

I don't point this out to say "yes, online ed. could change higher learning forevermore" or "no, this is a flash in the pan", but to highlight that traditional Universities do have a role-- they may simply be a place where people watch online videos and build stuff together, but they add value beyond a guy in his den watching a video (to be poetic predictions of the death of traditional learning are greatly exaggerated).

e.g. As a Stanford student, (1) above is certainly lowest on my list of great-things-Stanford-provides. (2) is high, and so is infrastructure. On (4), I think the Gates building has a astoundingly high great-ideas/sq-m. None of these can be replicated online easily.

But not all colleges and universities are at the same level as Stanford. The idea is to provide higher quality education to a larger group of people who might not otherwise not be able to receive it.

As you state, this cannot replace the benefit of putting a whole bunch of smart and motivated people in close physical proximity, but can complement other programs or help other students achieve a better education then would otherwise be possible.

I think this is great and having such high quality level instructors is very good. However, I'd caution the typical view nowadays, especially around here, that this signals the end of traditional college and that we should just throw it out.

I agree. I don't know about computer science students, but as a civil engineering student college means a whole lot more to me than just being lectured at. I'm involved in half a dozen student organizations, and a lot of the time I find myself learning more outside the classroom working with professors and other students on competition teams and organizations than I am in class. It's sad that so many people think watching videos can teach them as much as actually plugging into a real life program with other students and experience faculty to interact with. I think it shows there's something wrong with how most people are approaching college in general.

I agree that what students do outside the classroom is as important, if not more important, than the lectures. However, going through the lectures, in person or through a video, lessens the learning curve for critical skills that are a prerequisite to outside the classroom learning experiences. In other words, these videos give everyone who watches them an opportunity to engage in the next level of learning.

I believe it is possible that with these prerequisites moved outside of universities, it becomes possible for more group learning experiences, not unlike those you experienced with professors and other students, to become available to everyone. Currently, the latter is only available (mostly) to students at a university.

People don't (read: shouldn't) think that all it takes to become a computer scientist or civil engineer is to watch a few videos. They definitely know that they can't without learning the concepts in those videos, allowing themselves to become more involved with real challenges in the field and further their educations.

I agree that there's more to college than just the classes, but why can't we try to get some of this environment to exist outside of a college environment? What's sad to me is that once you leave undergrad, a lot of the things you talk about become closed off to people.

I agree. I think if it does happen, it will be a slow transition(several decades) because that's how long culture takes to change.

Thrun seems to be getting into this mode as well with this line: "He’s thinking big now. He imagines that in 10 years, job applicants will tout their Udacity degrees. In 50 years, he says, there will be only 10 institutions in the world delivering higher education and Udacity has a shot at being one of them. Thrun just has to plot the right course."

Why is it that people (even apparently faculty like Thrun) seem to forget that universities deliver more than just undergraduate education? "Delivering higher education" also means research labs and centers for scholars, graduate education, professional education, executive education, etc. While undergraduate education may be ripe for disruption, there is a serious leap needed to go from there to the complete disappearance of thousands of institutions.

Just look at the numbers: dederal research grants and endowments will sustain at least several hundred universities for the long-term, and other universities that are currently teaching-focused will move more of their efforts to research as students take online classes and stop being paid customers.

The institutions that should be worried are technical colleges (depending on major), community colleges and perhaps state college systems. They are the most likely to be disrupted, particularly if Udacity could offer a more comprehensive curriculum. But Stanford? Or even schools like the University of Florida or UC-Santa Barbara? They have plenty of other income sources, and still have a lot of life in them.

In your world view, places of higher education and places of advanced research are strongly coupled. This does not need to be the case. Historically some great research labs had nothing to do with education (Bell Labs). I think Thrun's prediction is that there were be great institutions of higher education that have nothing to do with research. I certainly agree with that point.

Of course, all disruptive innovation starts at the bottom of the market. I am sure the first big wins will be giving educational access to talented and bright but poor kids in the developing world. For them, even something like access to a first world university is far from affordable, and the fledgling online classes are much better than the alternative, even if they aren't yet competitive with traditional university education.

I took the online ml-class last semester, and I am now taking the pgm-class this semester. I have or am taking graduate versions of the same classes at a traditional university.

The big discovery (thanks to Sal Khan) is that teaching fundamentally scales. It's a big waste that all over the world, there are hundreds of smart researchers each teaching 10s to hundreds of students basically the same subject. What would be best is to get one great teacher for each of the subjects of those 100s researchers, make him focus really hard (10x the effort he would have put in, but 1/10 the total effort of all of the previous combined effort) to make a ridiculously good online class. Then you have freed up a bunch of time for the researchers, and most of the almost thousands to 10s of thousands of students are actually experiencing a better class.

I think you exemplify the more likely scenario, in my view, of the typical real, down and dirty use-cases for this stuff. That is people that are already undergraduates and graduate students using this as supplementary material either to their current classes or more probably extra-curricular.

As a replacement for the university, I think this is actually moving in exactly the wrong direction. We've already tried more centralized education (e.g. standardized testing), and it doesn't work. The quality of education any given student receives is proportional to the level to which that education can be personalized to match their interests, abilities, learning styles, etc. Decreasing faculty-student ratios means decreasing personalization. It's just a losing proposition all around.

This sort of system seems more like a replacement for the textbook to me. Traditional textbooks serve the purpose of giving both the instructor and students in a class access to the knowledge of the experts on a subject. This is simply a more interactive way of getting that expert information. Then the personalized part of the education can kick in and not have to worry about helping students memorize basic facts.

But online courses actually give you more options as far as personalizing education. At most universities, unless they're very large or very specialized, you're likely to have one professor from a specific subject in a field. With online lectures, even from a reasonably small number of sources you can mix and match. Two personal examples I can think of is supplementing my algorithms class with MIT's algorithms class on OCW, and watching Tom Mitchell's lecture on kernel functions to add depth to the topic as covered in Ng's Ml-class.

Even at great universities there are professors who are better researchers than teaches that still may be a students only option. If you took the top 10, or even 5 teachers on a given subject you could give students the freedom to mix and match as suited their learning style.

Personally I actually find the lectures a poor substitute for a text. I find the Probabilistic Graphical Models textbook a great addition to the pgm class on coursera. Even though the lectures for pgm-class are long and detail, there's still only so much information you can get into a few hours. The information density of a textbook is hard to beat.

One of the dirty little secrets (ok, not really a secret) of the education system (higher or otherwise) is that it's not just about helping the students learn, but also verifying that they have done so and credentialing them. Online education can potentially be pretty good at helping students learn (although, even at that, it's much weaker at helping the less-self-motivated students learn; you might regard that as either a plus or a minus, I suppose).

It's a lot less good at verifying it, especially if there's credentialing involved. As soon as the verification becomes high-stakes (due to a resultant credential---grade, course credit, etc), cheating becomes a big problem, and one not easily surmounted within the context of the online learning environments that have sprung up so far.

So if something like Thrun's class is really going to "change higher learning forever", one of those changes may have to be a separation of the teaching/learning component and the verification/credentialing component. The teachers could then say, learn it or don't, I'll help but I'm not your taskmaster... and then if the students want to prove to anyone else that they learned something from it, they have to figure out some other way to do that, because transcripts won't cut it.

Actually, I could kind of get behind that.

Do credentials even matter to anyone?

* You have Joel Spolsky writing about CS grads who cannot even pass the FizzBuzz test

* You have companies like 37signals talking about some of their best hires not having any credentials at all

* You have Google (and now everyone else, it seems) giving intense testing to all potential hires -- if you already have the credentials to prove you can do the work, why do you need to be tested?

I feel the online schools have popped up at just the right time. The world is moving on from the traditional education accreditation model, as the idea of giving credentials has proven to not work (see above). However, you still need to be able to walk the walk, and taking courses online is one way to get there.

There are already systems in place to handle verified testing, like GRE and CPA exams. There are online courses that have homework and reading components online, but test offline.

For non-tests, students can and do cheat at regular Universities.

I completely agree that for this system to work, a verification/credentialing component is needed and I actually believe that the money will be there. Give away the lectures and problem assignments but make money off of more personal tutoring and the credentials.

These classes aren't simply verified intelligence tests that employers can use to screen job applicants. If you learn something useful, you can use that knowledge to make money.

What's most interesting and beautiful about the concept of free, open education is that people that would not have access to a college education can now have it. I'm talking about people that are the poorest of the poor who would never see a university let alone be able to attend one. A lot of folks on here are talking about certification...certification doesn't matter, rather creating wealth matters! We're on YC Hacker News right now, I'm sure everybody has read Paul Graham's "How To Create Wealth" essay (http://paulgraham.com/wealth.html). A person with a college degree or higher is useless to society if they can't produce something with it. However, somebody with little or zero formal education that can create something world changing IS valuable! Just giving people access to information allows them to change their world and create wealth. Some of you may have heard of William Kamkwamba (http://en.wikipedia.org/wiki/William_Kamkwamba) who created a windmill that could generate electricity just after seeing a picture of a windmill on the cover of a book written in a language he couldn't read. I saw an interview with him on "The Daily Show" where he joked that when he had the opportunity to access the Internet for the first time, he typed "windmill" into Google and was like "where was this when I was trying to build that damn windmill?!". People like him can do so much just with a picture -- imagine if he had access to science and math lectures from an amazing university like Stanford.

We have finally reached that point (or are at least damn close to it) where virtually every person on the planet, no matter how poor, will have access to a computer and the Internet. That is all a person needs to access these resources. I'm sure everybody has been following the Raspberry Pi, a full-fledged computer with video output for $25 - and it's only going to get cheaper. People like Thruun and Khan are dead-on. The only way to get people out of poverty is by empowering them with knowledge and education. In my opinion, the work these folks are doing are the seeds of a coming worldwide renaissance where poverty and ignorance will be eliminated.

For me to get a Statement of Accomplishment, will I have to go along with the course or can I start later and still get it. I'm interested in CS101 but the course is already in its fifth week.

You can still join. The exam will count for 100%, instead of 50%.

A new class of CS101 starts April 16th

What is your source for this? I signed up for the class before the new year and got an email in February saying the class would be delayed until the second half of March and I would be notified two weeks in advance of the class start. I haven't heard anything since then and the cs101 site doesn't appear to have any more information.

I assumed samrat was referring to Udacity's CS101 course, because it is in its fifth week. I don't know about Coursera's CS101 course.

I will say that my friend just started taking Udacity's CS101 course, and he really loves it. I think they have a much better model for teaching a 101 course than Coursera* because you transition immediately from learning something to coding it and having your work checked.

* Disclaimer that I haven't looked at Coursera's CS101 course, only their other courses, but am presuming they're using the same technology.

OK, that makes sense. I wasn't aware of the Udacity course. I will have to check it out, as it is looking like the timing of the Stanford one might not work out for me. Do they start new classes fairly often? I can't seem to find the schedule on their site.

Coursera vs. MITx. Any views?

MITx supplies a textbook which is very nice. A Jeff Kent hacked together a scrolling textbook viewer as well. MITx needs a bit more moderating on the discussion boards. Many people are going to fail the class when they find out that there is more to the class than finding the answer in the forums.

MITx suffers a bit from small assignments and small exercise sets. They've compensated for the shortcoming already by adding tutorials. (I didn't notice them when I first took a gander - I think they were added.) The textbook also counters this deficit.

The first assignment for 6.002x was due last weekend but they take the best 10 out of 12, so it's still worthwhile to sign up!

Coursera's offerings are also nice. One of the features that was there that is no longer working for me is increased video playback speed. The +/- buttons do not work. Playing video at ~1.4x was quite nice. MITx can play at 1.25x and 1.5x. Dr. Aggarwal's (MIT) Hindi-Bostonian accent at 1.5x chipmunk speeds is a bit too much for my brain to process, but at 1.25x he seems even more excited about his teaching!

I'm also having issues with the video crashing or hanging with Coursera. To counteract that I am taking advantage of downloading the videos and using VLC to chipmunk them. Coursera is more portable. But in both cases students have hacked together downloaders for the entire course anyway.

And as always, check how you are evaluated. Just like a real course, the way you are evaluated changes from course to course.

I'm really enjoying the Natural Language Processing. The first assignment is due on Sunday I think.

Other than the SaaS and Model Thinking class, everything else is still just starting or hasn't started yet. I am going through the Model Thinking videos course even though I registered way too late. Really really worth it.

My TL; DR: I narrowly prefer the MITx setup over Coursera.

Coursera, I'm doing the NLP class but it is just a guy writing equations on a chalkboard. The class would be much better if he used plain english to explain his equations instead of reading off the slide.

MITx, I gave up on the electronics course because I can't read the guys handwriting and I didn't like the labs. The book is also in the worst format I've ever seen but at least it's free.

> Coursera, I'm doing the NLP class but it is just a guy writing equations on a chalkboard

I'm doing the NLP class too. The professors go through the material pretty rapidly and sometimes you have to fill in some gaps yourself, but "just a guy writing equations on a chalkboard" isn't an accurate description.

Reply on news.ycombinator.com