Arnold on the Current State of Computers in Education
By Bryan Caplan
Here is how I size up the current state of computers in education:
My reactions, point by point:
1. Note that in the music industry, the Internet has put record
stores out of business. It has not put composers and musicians out of
Point taken. But one of the main benefits of online education is supposed to be that everyone starts learning from very best teachers on the planet. If that’s so, then why wouldn’t existing professors be more analogous to record store employees than composers or musicians?
2. Computers have a comparative advantage in repetition,
distribution, and data retrieval and storage. Humans have a comparative
advantage in interpersonal skills.
3. Lectures are a very weak teaching method. Distributing lectures
on line costs little, but by the same token it provides little benefit.
I agree, but this raises a huge puzzle. If lectures are so useless, why do professors spend more time lecturing than in all other teaching activities combined? My story, of course, is signaling. Lectures plus tests are a bad way to impart skills, but a great way for students to signal their intelligence, conscientiousness, and conformity to employers.
4. Teaching equals feedback. Some forms of feedback are repetitive
(grading multiple-choice tests), which argues for computers, but many
forms require interpersonal skills (grading essays), which argues for
humans. Yes, computers are starting to learn to grade essays, but this
is not yet their comparative advantage.
Most classes provide very little feedback of any kind. In a typical social science class, TAs grade the homework, and the professor grades one midterm and one final per student. How valuable can this homeopathic dose of “feedback” possibly be?
Feedback admittedly seems more useful when students write papers. But when professors actually make students repeatedly rewrite their papers, they tell me that students largely ignore professorial feedback.
5. Students learn from one another. Yes, entrepreneurs are trying
to reproduce “social learning” over the Internet, but this is not its
A minority of serious students learn from one another. Most students barely discuss their coursework with fellow students. I’d say that social networking websites stimulate far more social learning for intellectually engaged students than dorm room discussions ever have. When I was an undergrad, I had to struggle to find students interested in discussing ideas. And I was at UC Berkeley, one of the most elite public universities in the world.
6. I think that for (some) individual teachers, a “hybrid model”
that combines human feedback with effective use of computers will win.
Does using PowerPoint count as a “hybrid model”? If so, we’re already there. But the change seems pretty superficial.
7. In terms of effective education using a “hybrid model,” I think
that existing educational institutions will lose. Institutional
adaptation tends to be inferior to individual adaptation. Tyler and
Bryan will still have jobs in 15 years, even if GMU is effectively dead.
If you define “effectively dead,” we have a bet.
8. Education has a very large Hansonian
component. Politicians show that they care by rhetorically supporting
education. Parents show that they care about their children by
arranging for them to attend high-status schools. As I have said
before, I think this is the element that makes the future of education
so difficult to predict. Right now, the status of online education is
low relative to the status of in-person education. If that “tips” at
some point, so that it becomes higher status to give your children
tablets and sign them up for online courses, then the institutions
True, but I continue to think that Arnold’s focusing too much on elite parents and too little on garden-variety employers. In our society, college is first and foremost how students convince employers that they’re worth hiring for good jobs. We can imagine a tipping point, but I see no reason to expect tipping to actually happen.
As usual, I’m open to bets.