Could the KHAAAAAAN! Academy revolutionize *classroom* education?

You might be aware of the Khan Academy online videos that are available on YouTube. There are at present more than 2k videos, mostly on K-12 mathematics, but also as far as diffy-Q and linear algebra, some physics and biology, a bit of history, banking, and a few other subjects.

What you might not have yet heard is that the Los Altos school distract is experimenting with an interesting partnership with Khan.

The students do the Khan video lectures, often at home instead of hearing lectures from the teachers in the classroom, and in the classroom students work on “homework”, and other creative activities that focus on applying what they’ve learned from the computer. The Khan Academy has a set of problems online, and gathers data on what problems students are having most trouble with. Instructors can use that data to see where students are getting stuck with the videos and online exercises. They can then team up a student who “gets it” with a student who doesn’t, so that the students teach themselves – the student helper thus getting yet more instruction, because in order to teach it they have to know it even better. If there are still problems, then the teacher can try to help personally in any mental blocks that still persist.

The real teacher thus becomes more of a coach who can focus their efforts where those efforts are most needed. The teachers apparently enjoy it, because their precious time is put to its most valuable use. What Khan has found in his data, according to his 20-minute TED talk, is that many students who manage to break their blocks on a few specific hazy areas – these difficult areas precisely shown by the extensive data gathered – often speed up considerably with the subsequent material and can sometimes catch up and even overtake the other good students in the class. This method, by focusing on full mastery of each individual topic, apparently fills in all the gaps and thus makes things really click for the students as they progress to harder and harder material.

My question is: Will this method really catch on? Do you think it will work as well as they hope? There’s a super nerdy guy none of you have ever heard of before at the end of the TED video who seems to think this technique is a genuine innovation.

I am hopeful myself. I’ve been waiting for an internet-based sea change in education. This just might be it.

I think it’ll work as well as traditional home schooling or religious private schools. That is, I think it’ll work well for a few people and work poorly for the vast majority.

  1. For now, it’s being limited to arithmetic. In a traditional school setting, these lessons are typically the rote and mechanical portions of learning. Let’s see how well this type of program works in reading or even a science that often includes “fuzzy” bits, like biology or geology.

  2. Los Altos is one of the wealthiest cities in California and racially atypical of either the US or California (~80% white, 15% Asian, ~4% Hispanic/Latino). Just check out the Wikipedia entry on Los Altos. It’s unemployment rate is friggin’ 5%, well below the national and California averages. The schools are already really good. It’s basically a suburban enclave of well to do people.

  3. It’s basically one of the cities you choose if you work in Silicon Valley. It’s no secret that there’s a correlation between well educated, middle to upper class parents and high achieving kids. By virtue of birth, these kids will tend to be more intelligent and motivated than the average American student. Good program or not, the kids in this town are going to post better test scores than the average urban student and are certainly going to end up with better lives, on average, than their urban counterparts. Basically, this program works fine for Los Altos. Try it in the middle of Los Angeles, and you won’t see much improvement in math education.

  4. Remember how TV was also supposed to shift the education paradigm? Maybe this program has a shot, but hundreds of years of education in a number of different societies would indicate it will best serve a niche market.

Here’s a recent column/article about Khan: Khan Academy: Great Idea- With One Glaring Hole.

Some quotes critical of Khan:

I haven’t watched enough Khan to know whether these criticisms are fair.

I think ideas such as what the OP describes have potential. I’m not sure having students watch instructional videos at home and do online execrises and interact with each other and their instructors in class is all that much different from the old-fashioned model where students read their textbooks at home and do pencil-and-paper exercises and interact with each other and their instructors in class. And, given that people are willing to pay to attend live concerts and theater performances and comedy acts, go out to hear live public speakers, watch live sporting events, go to church to hear live preaching, when they could easily stay at home and watch videos of all of these things, I think there’s something to be said for live teaching, live lectures and demonstrations.

People have fantasized about eliminating traditional instruction and holding classes by computer/TV for ages. It’s been tried time after time since the dawn of TV.

Every time, the results are the same: It works very well for certain students and certain subjects, not so well for others.

Some students really take to the independent studies approach, and find it gratifying to work through the stages. Others find it to be so much “busywork” and have trouble focusing on it or taking it seriously. I always hated it- it felt like an endless mountain of work that wasn’t really leading any place, and my objective became not “to learn” but “to get more pages done.”

Subjects like math, which have discrete “correct” answers and focus on rote learning are well adapted for it. Subjects like literature and history really need the accompanying classroom discussion, though. Learning to analyze, interpret and present information is an inherently social activity. A lot of the “learning” comes from the composite of different perspectives and clashings of ideas that happen between classmates.

I imagine the OP is a more methodical, less social type. That’s fine, but that’s just one way to be. Net based learning may be just thing thing for you, but it probably isn’t for everyone. Classroom teaching has been around for thousands of years, and chicken-littleism aside, it tends to produce educated people.

I think it can be a valuable supplement to classroom instruction. I recall my school getting computers when I was in junior high because OF COURSE a school needs a computer lab, computers are the future. Problem is we never did anything really constructive with them; we were introduced to the internet and that’s about it. Now we have a tangible, interactive study aid that a student can replay as much as he or she wants, and best of all it’s free.

When I was in business school, I had a couple of professors who used videos VERY effectively.

One had recorded all of his lectures as a voice-over his powerpoints. For any class session, you were to watch 1-2 of the sessions. Class itself was dedicated to discussion and group work. I loved it - I didn’t feel like I was spending time just listening to him read slides to me with 40 others. Classroom time was actually valueable.

A second professor would take any subject that the students were having problems understanding, and record a supplementary video that he would post. This REALLY helped when 90% of us got it, but 10% were bogging down the class. You could also go back to the videos to help study.

I think that there is something to say for taking standard lecture material and making it a video, and then letting the teacher / professor focus more on interaction. I could see this working as low as the high school level, but I am not sure how much further down the line it would work without some real Pixar level support of the lectures.

Education is a dish best served cold.

The criticisms are pretty fair. However, I think a few they named a few others are more problematic.

  1. First, the lessons can be boring and too procedural. It works okay for math, but they often don’t give you a full context of the importance and relevance of a topic.

  2. It only works if you know what you don’t know, and how to classify it. His video titles are fairly accurate, but if you don’t know which video to watch, it can become tedious.

  3. There is little continuity between videos. In a live class, there is often a bridge or a segue between subjects and topics that makes sense. His videos exist as in discrete parts. Even though they are organized in a fairly logical way, they often don’t seem to coalesce as anything more than separate lessons as opposed to an actual course.

  4. There is only one format and one teacher. Most of his lessons are him doing problems on a black screen with voice over. That can work well in some settings, but people often respond better when you can see a person, and make a visual connection.

  5. The article also inadvertently introduces the biggest problem. The article starts with the author plugging his own youtube video channel. Competition is generally good, but not when it prevents the adoption of a universal standard. The problem is that there is no uniformly accepted standard of what should be taught in say Calculus 1, let alone how online teaching tools should be formatted and disseminated. It’s only a matter of time before Khan academy sees several more prominent organizations doing the same. With that, you introduce all of the same redundancies that you sought to eliminate by putting things online.

Well, that explains why the Bohr model of the atom is still being taught, I guess.

I can’t tell what you’re saying here by “eliminating traditional instruction”.

This isn’t about eliminating the classroom. It’s not about eliminating the teacher, not about having a bunch of kids stay home in front of the computer all day. It’s about creating more time for classroom creativity and discussion. It’s a new technique that acts as a supplement, with perhaps the potential to increase the effectiveness of teachers who would have the real data to know what to concentrate on with their students.


It has not.

This isn’t about sitting junior in front of the telly and drinking yourself into a stupor while his eyes glaze over from an hour-long educational video that has no backup instruction. The videos are short, most ten minutes long, and there is an actual program involved here, a data collection technique with respect to the problem solving, which might allow teachers to know what to focus on. This is something new. It could not have possibly existed with the advent of television. Instant feedback and data analysis were not available at that time, but with this new trial program, a teacher knows immediately after they log-in in the morning that little Billy was having problems adding and subtracting negative numbers last night. Now, that doesn’t mean that this will necessarily work any better than previous (entirely passive) efforts, but it nevertheless remains a substantive difference. This is the first convincing bit of technology that truly seems to act entirely in concert with the advantages of classroom instruction, instead of trying to replace the classroom.

Obviously, there’s a selection bias issue when we’re dealing with YouTube videos that are typically seen voluntarily, and Great Antibob’s point about the uniqueness of the current school district experiment is also well taken. Still, it’s an interesting question. This is a different kind of technological supplement than has been tried before, and it’s worth dealing with those differences on their own terms.

The purpose of this particular program, as it states in the OP, is to increase the amount of class time available for group discussion.

The point is more discussion. You mention that some subjects really need classroom discussion. Yes. That is the point. The theory behind this is that even math would benefit from more classroom discussion and opportunities for creativity. They attempt to accomplish that by getting most of the lecture work done privately, at the student’s own pace.

Well, I can’t answer whether the videos are “boring”. Different people have different tastes. It’s a certainty that some would respond to the technique better than others. But I can’t currently see how short video lectures, which students are free to stop and rewind with no fear of interrupting the teacher, would be a negative. Starting with the lecture at home, and continuing into real practical exercises in the class seems to me to be an unmitigated advantage. It appears to take advantage of the best of both worlds.

And many of the other criticisms in that article are factually incorrect. Take this one:

I haven’t hit too terribly many of the videos, but this is clearly wrong. He hits concepts constantly. He explains the concepts, often visually. He approaches the ideas from more than one direction, and he shows how they work. He has done this in every single video I’ve seen, without exception. He explicitly states that he’s aiming for an intuitive approach to learning. And that’s not the only mistake that they make. This next criticism is also wrong, factually wrong:

He quite clearly has his own distinctive style. It becomes unmistakable after very few viewings. In many of his sequences, he is able to refer directly to previous discussions (i.e. videos), and his online app even has a clearly outlined map of skills to learn from the most basic foundations. And you can hear him thinking his way through problems out loud all the time – he doesn’t have the lessons rigorously planned ahead of time, and so the entire process is him thinking his way through the problem.

These criticisms are just ego talking. This guy has his own videos, which he claims are “not boring”. I haven’t seen them, so maybe they actually are better and more interesting videos, but his evaluation of Khan is just silly.

Yes. It will not be just video taping a teacher’s instruction session. It will happen when there are many, many, tools available that a student can use seperately from direct teacher instruction. Tools that are interactive, educational networking, individualized course and content planning, all with the intent of allowing teachers to spend the most time at what they do best, and not repeating non-interactive activities. There have to be numerous tools for each topic that deal with the different levels and abilities of students, and the manner in which they learn efficiently.

I hear the Ulan Bator school district experienced marked improvement in horse archery and higher rates of female lamentation when they introduced Khan Academy.

Using the videos by themselves would be a terrible idea, since they’re not interactive. Khan sitting in front of his camera can’t tell when the students grasp an idea and he can move on, versus when they need a little more time with something. Until AI becomes vastly more sophisticated, that’s going to be something we’re going to continue needing real teachers for.

But from the OP, it doesn’t look like that’s what Los Altos is doing. They’re using the videos, but they also still have the classroom time with the live teacher. And since the general lecture material is covered in the videos, the teacher can spend all of the classroom time responding to the students’ individual issues. We’ll have to see how well it works, of course, but it looks to me like an excellent idea, and I’m glad they’re trying it.

The biggest problem I foresee is that, with the video-watching being done outside of class time, some students will slack off and not watch them, hoping to just coast through the supervised time with the teacher. But then, this isn’t qualitatively all that different from students slacking on studying or doing homework in a more traditional educational model.

Short television instruction has been used before. i can remember when “interactive DVD” came out, and the format was rather similar but with the “chunks” stored on DVD rather than YouTube.

Teachers have, since the begininning of time, had a “data collection” technique that provides instant feedback. It’s called “schoolwork.” Teachers have always been able to look at what their students were getting wrong and tailor their lessons accordingly. Teachers have always asked “Okay, did anyone have trouble with problem five, because it was a bit tricky” and then the ability to (even more instantly than looking it up on a computer) address that. That’s what teaching is. A computer interface may make the data collection a bit easier, but we’ve had scantrons for a long time, and generally it hasn’t been worth using them in everyday classroom instruction.

I’m not saying this is a useful tool, but there is nothing new about the “present information and then evaluate student’s work on it,” concept.

A potential weakness is that it may be able to show where the class as a whole has weaknesses, but isn’t at good at pinpointing where individual students are going wrong. You may know that the class as a whole is having trouble with multiplication, but it could be that Timmy is mixed about how to carry, Zoe gets mixed up because her handwriting is messy and the numbers look the same, Rich hasn’t memorizing his timestables, Keisha keeps adding incorrectly, Marco is making careless errors and not checking his work, and Ann is doing it wrong on purpose because she hates school and likes to make problems. All of these issues are going to look exactly the same on the morning printout. Only close interaction with the students as individuals will let the teacher give the guidance needed for each case.

He tasks me. He tasks me!

Teacher, make him stop tasking me!

In my own math learning and teaching experience, the competition of different available resources with different perspectives is generally a good thing, providing some measure of relief from the unfortunate reality that exceedingly often, the standard way of teaching some area of mathematics (or other subjects, for that matter) isn’t very good; again and again, I find that “the standard” perspective on a subject is often just obfuscating cruft which persists purely by the inertia of tradition, long after the discovery of better ways of looking at it. On top of which, of course, different students find different approaches more to their taste anyway. The way I see it, an educational world bountiful enough to provide such “redundancy” is far more enjoyable than the alternative.

Well, for example, using a random Khan Academy video to illustrate this criticism, consider this introduction to matrix multiplication.

I think this video is a good illustration of what people mean when they claim Khan Academy is more about the computational element of mathematics than the conceptual element. In the video (and its sequel), the definition of matrix multiplication is given, in the sense that one is told the procedure by which one could calculate the product of two matrices, but no motivation is given for why one would care about this highly non-intuitive definition of matrix multiplication as opposed to any other. It’s just some crazy made-up rules handed down from above, so far as the student can tell (and, indeed, so far as the instructor explicitly stresses, for the time being). The instructor spends the entire video saying “I’m going to define for you the rules of an arbitrary algorithm” and then in painstaking detail showing how to execute that algorithm, without any effort at teaching what I would call the underlying concepts that make the value of those rules obvious and their derivation clear: the concepts of linear transformations (i.e., functions which preserve addition), of matrices as representations of linear transformations, and of matrix multiplication as composition of the corresponding transformations. Without that conceptual background, the algorithm being memorized is just some voodoo incantations.

Now, for all I know, the instructor eventually reveals the conceptual underpinning of matrix multiplication in some other video way down the line, but why languish in confusion in the meanwhile instead of introducing the conceptual understanding upfront?

But in his defense, I don’t think he’s any worse or different than traditional teachers here (because, well, traditional math education is also all about the computational element rather than the conceptual one); as I recall, I was also taught matrix multiplication in this unfortunate fashion myself.

(Lest I be misunderstood, though, I do greatly admire the efforts of the Khan Academy, and clearly it is a valuable resource for many students. I just get the impression it generally falls into the same “How to be a calculator” trap that infests math education more generally.)

The program keeps track of their video-watching process. If they don’t do their home videowork, and the accompanying problem sets online, that would show up in the teachers’ records.

The YouTube data also is said to include more general “attention span” markers, which indicate the exact moments in the videos where attention seems to drop off. I assume this is based on viewers stopping and rewinding different sections, but I’m not sure.

And that would be an excellent point if this program attempted to use scantrons, or the computer equivalent, as the basis of classroom instruction.

It doesn’t.


They are not.

The entire feedback system records individual student performance, including time spent on problem sets and videos; specific concepts which they’ve “mastered”, with ten questions in a row answered correctly, categorized very neatly and narrowly; the online problem sets they had most trouble with, recording to the second how long they spent on each question; and other discrete nuggets of precious precious data. The point of those problem sets is to ensure that students have full mastery of a concept before moving on. If little Billy is having trouble understanding adding negative numbers, then he is given help on that topic until it clicks. He isn’t supposed to move on until all the gaps are filled. And little Susie can’t dick around and do poorly on purpose without setting off all kinds of alarms. You can sign up in their “Practice” section to see their data service yourself and check all this out.

Even a good student can have Swiss-cheese-like gaps, little problems they might not even realize they have, and traditional instruction is potentially going to miss those gaps if the students don’t ask questions. But this Khan thing is specifically designed to give teachers exact data on those little moments of haziness, before those Swiss cheese gaps become a problem in understanding harder material. The purpose is to help teachers focus their attentions more effectively. And in fact, the feedback and data system is still being re-jiggered, based on teacher requests. The cooperation between KA and the school district teachers is, they believe, allowing them to refine their data gathering process even more, specifically according to the guidelines of what information teachers think would be most helpful.

Now, that’s still no guarantee that this would work better. There is going to be an asynchronous element in the classroom, and I’m not sure how exactly that would be managed.

But there’s kind of already an asynchronous element in classroom learning. Some follow along, and others have troubles and lag behind, and I don’t see that changing with this. What’s different is how the program attempts to home in on specific trouble spots for the teacher to help with. I haven’t seen any rigorous data here, but the few teachers whose comments I’ve read on this genuinely seem to like it as a supplement that’s powerful enough to allow them to re-engineer their classroom schedules to maximize useful interaction.

The revolutionary point here is not instructional videos, scantrons, computer-education, or whatever other preconceived strawman notions you have. The difference here is that the whole package has been put together in such a way that this technology seems, finally, to put itself effectively in service to the wishes and needs of the teachers. If this works as well as they are hoping, then much more of our teachers’ precious time in class can be spent on those very issues that you rightly think are important.

Good point. I got two things to say 'bout that.

First, I would repeat your observation that even if there are substandard teaching methods that creep into the videos like matrix multiplication, they are likely to be substandard in exactly the same way that traditional teaching is likely to be substandard. I would add the point that if the teacher doesn’t know that, then the students haven’t lost out on much, but if the teacher does realize, then that can be compensated for. (Related to this: you’ll be sad, though unsurprised, to hear that his presentation of the Euler Formula is through the Taylor series. At least from what I saw, anyway. I didn’t watch the whole thing. The BetterExplained guy has a much BetterExplanation of that one.)

Second, I would refer again to that YouTube “attention span” data they’re gathering. There’s nothing stopping you or me or Khan himself from creating a new sequence of matrix multiplication videos that are taught with a different method, based on a more intuitive and easier to follow approach. If the new sequence of videos has better metrics, then that can be identified conclusively. The original sequence can be quickly marked as substandard, and then replaced. This is unprecedented math education data, which means there is the possibility here for the videos to just get better and better and better. They will never be enough by themselves without genuine classroom interaction, but they can nevertheless be continually improved.