After I ran a panel at the South by Southwest Interactive Festival in 2010, Imagineering a Fully Digitized and Connected Future, I played around with writing a book about life in the 21st Century. The book didn’t pan out, but I liked the interview I did with Bror Saxberg, and I wanted to share it with you all. Bror is the Chief Learning Officer for Kaplan, but that doesn’t really give you a sense of the man. He practiced medicine after earning his M.D. from Harvard Medical School, and he’s got a Ph.D. for Electrical Engineering and Computer Science from MIT and a Masters from Oxford plus a couple of more degrees from the University of Washington … You getting the idea?
I first met Bror when he was the chief big brain for K12, Inc., an online learning company. See, education is his real passion, and that’s what I asked him about in this interview. Here are some of the things he said, in a format swiped from Esquire Magazine’s “What I’ve Learned” feature:
Put learning ahead of technology. The moment technology leads, you’re dead. It can take you in a spider web and who knows where it will go, right? That’s not going to help you. You gotta get the process right and then technology will help you do it faster, more reliably, and cheaper.
I don’t believe the use of technology for learning is going to be a stark break from the really slow progress that we’ve seen over the last ten years. My thinking is that you’re still going to have to rely on people to do a whole lot of things.
We have 60,000 university students at Kaplan University. It’s a virtual university where 30,000 students take Algebra every year. They don’t all start at once; we have like a dozen starts a year. That means, think of it, every year we have a dozen cohorts of more than 2,000 students each all hitting the same hard objectives during that calendar year.
We can do more than a decade’s worth of research in one year! We get innovation down to consumer electronic lifecycles. Conventional education institutions and funding sources think about a yearly cycle of improvement? I say “F” that. Every two weeks! Every six weeks! What’s your next idea? Get the data, you got a week, now roll it back out again. Let’s go.
We don’t have to wait for permission. We don’t have to wait for funding. We don’t have to wait for the Gates Foundation to love us. That’s part of the frustration, right? There are all these slow-moving, nice-feeling people out there, and I love them to death; but we gotta get rolling, man. I don’t want to wait.
Benjamin Bloom wanted to find the limits of learning. They got Bell Curves of students together and gave them highly professional tutoring for months, incredibly expensive tutoring.
What they found by doing that in several different fields, domains, age groups, etc. is that you can move the whole Bell Curve up by two standard deviations. Two standard deviations is mondo! It takes a 50th percentile player and moves them up to where the 96th percentile used to be.
Kaplan gets all these broken students who had trouble with school before. Benjamin Bloom’s work shows it’s not about your mind; your mind has two standard deviations to go. It’s OUR problem. The issue is not do you have it in you; the issue is how do we get it out. How do we make it happen because it can happen?
I don’t think we’ll be fiddling with genetics to become super learners. We have all that we’re going to get.
The mind is an association engine, not a computing engine. We’re very good at connecting stuff — that’s what working memory does. It’s this narrow but flexible part of our minds that can’t do very many things at once. It’s trying to interpret this enormously complex sensorium we have and find what it connects to that’s already on the inside that’s relevant. But if you have nothing on the inside, then seeing things on a screen is likely to just overload working memory.
There’s lots of English language training out there. You can order yourself a latte after a certain amount of training. But that’s not what you need in order to sit in a chemistry lecture, understand what the guy is saying, ask an intelligent question about it and understand the response.
You could say “Wait, it’s not that hard, it’s new vocabulary, but isn’t it really the same as ordering a latte?”
The answer is no because working memory is fine to be overloaded with ordering a latte, but it has to be almost completely unloaded for a chemistry course because chemistry is hard, right? I have to have all the language stuff – the listening, the speaking and how to connect the language. I have to have that wired down because otherwise I can’t do it.
I was doing an infectious disease rotation at the Mass General Hospital. We were pretty smart medical students, and we’d solve most problems. But every now and then, we couldn’t solve something, so you called the expert in. Infectious disease is one those bizarre fields where the experts are way out on the curve. I mean, they are huge experts. This guy shows up. He opens the door. He walks across the room. He waves at the patient. He walks over to the patient, shakes his hand, turns around and makes the diagnosis.
So I stop him at the door, right? I’m thinking this is an amazing bravura display. I say, “Well how’d you do that?”
He looks like he has no clue! And I’m thinking I can throw darts, too, you know? I didn’t just pick something out of my hat, right?
But then he said, “Well, I opened the door and the room didn’t smell of these six things, and so that knocked off these parts of the tree of the things that could be wrong. As I walked across the room, his eyes tracked me. And when he raised his hand to wave at me, it didn’t wiggle in a certain way so that knocked off some neurological stuff. As I walked closer to him, I could see his skin color, his eye color, the shaking of his hand, the muscle …”
He was a frigging, walking diagnostic machine! The data was just pouring in from every sense he had, okay? The cool thing was that it was apparent that he had no conscious awareness that that’s what he did until I asked him to walk through it.
He literally had to ask, “Yeh, what did I do?”
This is what happens with experts: They get a rapid, fluent ability to look at patterns and chunks of information. Things are wired in at a subconscious level, so they don’t have to occupy working memory.
The speculation is that pools of specialized neurons are interconnected differently so that the rate at which they process things just happens faster because it’s neural network that’s been trained with stimulus response over many, many iterations. It’s becoming apparent that this is right, yet almost no one is bothering to use it in any learning environments.
Cognitive psychologists have been looking at how experts work. When experts are teaching novices, they essentially never tell them more than 30 percent of what they need to know to actually make an expert decision. It’s not because experts are bad guys. It’s because a huge amount of expert decision-making is subconscious.
A technique cognitive psychologists use to study experts is called cognitive task analysis. You get two or three experts from a domain, and you get them on their own and you say, “Give us examples of some really hard problems that would demonstrate to you that somebody else is an expert; now tell us how you would go through those.”
You take notes. You watch the experts execute their work, and you have them talk aloud. Then you consolidate the stories because each expert will only tell you bits because they don’t consciously know how they do it. Put it together, review it, and you have something cool.
This is funny and frustrating at the same time: They’ve found that it’s actually quite inefficient to use teachers who are experts because teachers will fill the empty spaces they don’t think about with stuff that is irrelevant, but which they convince themselves is really what they do.
You have to show them video where they see themselves doing the task and they discover that what they tell their students is totally irrelevant to how they do it themselves. They’re always wrong.
You need to go to real experts who are not confused by teaching. What you find typically is that you can knock out anywhere from 20 to 40 percent of the training that is conventionally done to become an expert because most of it is a waste. It’s useless. It’s just air-filling that is not tied to expertise. So that’s fairly shocking, but you also find that you dramatically reduce errors in the first year job performance of the people who go through these programs.
Less time; fewer errors. That’s pretty cool.
There are still two standard deviations out there waiting to be realized.
Training surgeons is dangerous. For every ten residents you train in surgeries related to the superior vena cava, for example, you have one fatality and three or four major complications. They did a cognitive task analysis of the procedure and sure enough, training time shrunk by 25 percent.
They followed participants for a couple of years after they left the training programs and found that yes, the original group with the original training program had the same rate of problems. The other group: no fatalities, no major complications.
You need to every two to five years you need to repeat a cognitive task analysis. The beauty of this is that it allows you create a very efficient feedback loop. You can go back and alter the parts of your training that are no longer keeping up. You can now be really focused. You can spend $100,000 on a tenth of the course, because you have the evidence that that is the part with the problem, so now you can really afford to build a simulation environment to nail that sucker.
The CEO of Kaplan came in a couple of years ago and hired me last summer. He announced that for Kaplan to be advantaged as a company we will need to be the world’s best educator by 2020. Then he looked at me and asked, “So what did I just say?”
To me what he means is that we have to put in place techniques, technologies, and hardware by 2015 so that we can run feedback loops for five more years after that, collect the data and run experiments.