But "accessibility issues" could completely change the way people learn. What if each student could access the kind of instruction one gets at Philips Exeter or Princeton University?
Or what if the tailored aspect of private instruction could be scaled up? For example, if you give a tough math problem to 1,000,000 students, and want to have a tutor walk them through the problem, you'd need to have 1,000,000 tutors. But there are probably only 20 types of mistake one can make on a given math problem. So if you had a "choose your own adventure" solution, one math teacher could, in a month, record lessons that reach a hundred million students at a personalized level.
Or, one could make a cost-effective way to learn from the absolute top practitioners in a field. Most people cannot afford private basketball lessons with an NBA basketball coach, for example. But let's say such a coach watched 10,000 people play basketball for five minutes each, and recorded five minutes of feedback for each person (about one year of full-time work). Then, you could upload a video of you playing, you could algorithmically analyze the user's playing style, and match them to the person in the set of 10,000 that matches them best. The five minutes of feedback might be really useful to that person, and can be delivered at scale.
> if you give a tough math problem to 1,000,000 students, and want to have a tutor walk them through the problem, you'd need to have 1,000,000 tutors. But there are probably only 20 types of mistake one can make on a given math problem. So if you had a "choose your own adventure" solution, one math teacher could, in a month, record lessons that reach a hundred million students at a personalized level.
Not quite. There may be 20 different types of mistake you could make on the problem (I would make a lower estimate, personally), but you can make them in several different places, and the influence of a mistake will be felt in the rest of the problem.
Udacity deals with this currently by never assigning complicated problems that might see a mistake in one step show up in a later step; this isn't quite ideal as instruction, but it makes it possible to automate handling the assignments.
I'm not sure math is a good example, but for grading small code exercises there are systems that map "equivalent" solutions on each other, ignoring whitespace/variable names/simple equivalent operations. This allows the tutors to spend more energy on the long tail of "strange" solutions which needs more individual attention.
It depends on how you look at it. Thanks to Wikipedia (this was back in 2002-2006) I was able to bypass the problem that half of my university lecturers were researchers first, and teachers a distant second.
Key computer science concepts went from being incomprehensible to being remarkably easy, and I didn't need to spend a small fortune on textbooks either. So to some extent, a revolution has already happened.
I remember a huge controversy while I was in college over whether it was even possible to get reliable information off wikipedia. At the time, my primary use for wikipedia was math articles, so I found this somewhat amusing. The math articles were fine.
Accessibility definitely, but I think it's naive to assume that just because the centralized teacher-student model of education has been around for a long time necessarily means it will never be supplanted.
I think the evolution of education is probably technology supporting traditional teaching, along with some way of easily varying the curriculum or the learning mode to accommodate different learning styles.
For that to happen I would guess that making designing curricula, lesson plans and activities easier to do would be a good start.
Agree, technology will (and already has to some extent) revolutionize learning on your own, but there still is a lot of value in going to a class, learning together with real people and listening to an experienced teacher. It is kind of similar as to why virtual meetings will never fully supplant real life meetings. Our mind just reacts differently in real conversations and i think peer pressure actually makes some sense in a learning environment.
It's similar to how i feel about working at home after years of doing it and now working in an office again. It's just a different feeling of commitment to a project.
Real gains in educational technology will come from mediating in-person social interactions (focused on knowledge building rather than general purpose interactions, of course).
In many instances, the teacher actually knows what should be done to improve the learning of a student - but is logistically constrained from providing that level of service (Bloom's 2 sigma problem). The low hanging fruit is removing those constraints.
We will need better tools for understanding the knowledge/abilities of our students in a much more refined manner.
I think just about everyone would be delighted if recorded lectures completely replace traditional giant lectures.
But I agree, there's much more to the structure of education that technology can help with, but not supplant. Universities aren't going away (or radically changing) any time soon.
Or what if the tailored aspect of private instruction could be scaled up? For example, if you give a tough math problem to 1,000,000 students, and want to have a tutor walk them through the problem, you'd need to have 1,000,000 tutors. But there are probably only 20 types of mistake one can make on a given math problem. So if you had a "choose your own adventure" solution, one math teacher could, in a month, record lessons that reach a hundred million students at a personalized level.
Or, one could make a cost-effective way to learn from the absolute top practitioners in a field. Most people cannot afford private basketball lessons with an NBA basketball coach, for example. But let's say such a coach watched 10,000 people play basketball for five minutes each, and recorded five minutes of feedback for each person (about one year of full-time work). Then, you could upload a video of you playing, you could algorithmically analyze the user's playing style, and match them to the person in the set of 10,000 that matches them best. The five minutes of feedback might be really useful to that person, and can be delivered at scale.