Zach Wissner-Gross – TRANSCRIPT
How many times have you been in lecture halls, like this one? Maybe this reminds you of some of you college classes or if you’re a little younger, this is what you have to look forward to, when you get to college, right? Hundreds of people, packed into one room, listening to one person give a lecture and in this case, that’s me. Have you ever been bored or confused during a lecture? What do you do when you’re confused during a lecture? You kind of sit there quietly, looking around, hoping that somebody else is just as confused as you are, and might slowly raise your hand, to ask a question, and then you stop and you realize, “Man, this is a stupid question, and if I ask this question I’m not wasting my own time, I’m wasting everybody else’s time.”
So, you pull your hand back down and you focus again on the lecture and you realize at that point that you’re completely lost. Right? Have you had that experience? I know, I have too. So, if you really want to learn something lectures are a really, really inefficient way to learn, and there is research to back that up. So, here is Benjamin Bloom and he was an educational researcher and back in the 80’s he found that, compared to other students, students who learn in lectures actually do two standard deviations worse than students who get individual, personalized attention. So, if you get a private tutor, that is basically saying you’ll go from being a C student to an A student.
That’s huge! So, what should we do? So, one approach might be to tell every professor out there, “You need to spend a couple of hours with each student one-on-one and do that for your whole class.” Professors just don’t have the time to do that, right? Another approach, might be to get a private tutor for every student out there and that costs a lot of money. Personalized instruction is really expensive, both in terms of time and money, but lectures are scalable. If you can give a lecture to one person, then you can give a lecture to a hundred or a thousand or as many people as you can fit into that room, and you can stream it online for no extra cost. We give lectures because on a per-student basis, they are so cheap and they take so much less time than personalized instruction.
So, I’m going to move ahead from the 80s and the work of Benjamin Bloom to close to present day, 2012, when the New York Times declared “the year of the MOOC” What’s a “MOOC”? It sounds kind of funny, right? “MOOC.” It stands for “Massive Open Online Course” Harvard, MIT, Stanford, Berkeley, all the top institutions out there give online courses to millions of students around the world for free; these are “MOOCs.” So, you can go online and you can take a course with a hundred thousands students around the world simultaneously, that sounds great, right? But if you remember how little attention you got in that lecture hall with hundred other students and you divide that by a thousand, that’s how much personal attention you get in online MOOC with a hundred thousands other students.
MOOCS are lecture-based because lectures are what scale. So it shouldn’t be a surprise to you that as of, still today, more than 90% of students who will start a MOOC will drop out. So, over the last few years I’ve been asking myself a question and we use all this great technology for MOOCs. We use high-speed internet connection, cloud computing, cheap data storage, we need all these technologies to scale up lectures, so we can give them to millions of students, and my questions is this: can we leverage these same technologies to scale up not lectures, but personalized instruction? One-on-one tutoring and then maybe instead of having 90% of students dropping out, we can have 90% of students completing the courses that they signed up for. That would be something awesome.
So, what are the barriers to scaling up personalized instruction? So what I like to do sometimes is imagine you are a tutor and you have the student sitting next to you. What do you do? Do you give the student a 10-minute speech about how to solve the next part of their homework? No! That is really bad tutoring, you would never do that. What do you do? You talk to the student. You have a conversation. You ask questions like, “What are you up to in class? What are you struggling with? What is giving you the most trouble? This problem here, what approach would you take to solve that problem?” Lectures are one-way communication from teacher to student, while tutoring, personalized instruction, is a two-way conversation.
You have to listen to your students and that’s why they’re so hard to scale, that’s why tutoring is hard to scale, because students can and will say just about anything I used to teach Introductory Physics at Harvard, to some of the brightest students in the world, and one of the first lessons in that class is, “You throw a ball in the air. At what angle should you throw it so it goes as far as possible?” So most of the students say 45 degrees, some say 60, some say 247. We had one student who said, “You should throw it straight up,” and then he thought about a little bit and he changed his answer later on, so that’s the challenge! How can you build software that gives intelligent, useful feedback to students who are giving you so many different answers? That’s the challenge, that’s why you can’t scale personalized instruction.
So, my team and I were up to that task and we think we found the best way to do it, the best way to scale personalized learning. Here is one of our lessons we made a couple of lessons online, this is an early one on quadrilaterals. Quadrilaterals are shapes with four straight sides. We asked the students, after we’d given that definition, we’d say, “How many of these shapes here are quadrilaterals, can you find them?” And only 44% of students are able to do it, which means 56% got this question wrong, that’s a lot of wrong answers! So, we looked at the data, we tried to see what’s going on, and there were really obvious trends, there were patterns. They weren’t actually confused about what quadrilaterals were, they just might have missed them, 85% answered it wrong this way: They found the two on the bottom and they missed the one in the upper left.