Arnold Kling on Education and the Internet

Episode Summary

Executive Summary: Russ Roberts and Arnold Kling examine online education’s promise and limits, arguing that the biggest gains will come not from scaling lectures but from personalized feedback, adaptive learning, independent certification, and self-directed learning. They debate MOOCs, clickers, flipped classrooms, and whether college is mainly transformative, sorting, or signaling, concluding that education’s future likely lies in lower-cost, more flexible systems that better match students’ needs.

Main Topics: Why education technology changed so little for so long (Priority: 5/5): Kling argues that teaching methods remained largely unchanged for centuries because classrooms were built around a professor speaking to a room, with only minor tools like chalkboards and projectors altering the model. MOOCs and the limits of scaling lectures (Priority: 5/5): The speakers debate massive open online courses, with Kling skeptical that simply moving a traditional lecture online creates durable learning, while Roberts points to real access and completion gains for some learners. Personalized learning and feedback as the core of teaching (Priority: 5/5): Both emphasize that teaching works best when it provides immediate feedback, hints, and tailored guidance, not passive information delivery. Kling argues teaching equals feedback and that one-on-one tutoring is the ideal model. Adaptive textbooks and automation (Priority: 4/5): Kling discusses adaptive textbooks and AI-like tutoring systems that adjust difficulty, repeat material, and respond to student mood, but notes that current tools remain far from truly personalized human tutoring. Independent certification and separating teaching from grading (Priority: 5/5): Kling strongly favors independent certification to reduce credential monopolies, improve incentives, and avoid conflicts of interest between teachers who instruct and teachers who grade. Clickers, flipped classrooms, and social learning (Priority: 4/5): They discuss classroom technologies that increase interaction, especially clickers used for peer instruction, though Kling remains skeptical of tools that simply reinforce one-to-many teaching rather than genuine dialogue. College as sorting, signaling, and transformation (Priority: 5/5): The conversation broadens to whether college adds real learning or mostly credentialing and sorting. Roberts frames college as also a time for exploration and identity formation, while Kling stresses the signaling and sorting functions.

Key Arguments: Massive online courses are overhyped if they merely scale lecture-based teaching; the real promise lies in individualized instruction rather than broadcasting to thousands. The students who benefit most from MOOCs are either already close to the course level and well supported, or far away and otherwise excluded; many others will drop out or gain little. Teaching is fundamentally about feedback: students need to try, fail, get hints, and revise, which is hard to replicate in passive video or textbook formats. The best educational technology should mimic one-on-one tutoring by adapting to the learner’s pace, confidence, and mastery level. Separating instruction from certification could improve rigor and reduce grade inflation, while also creating room for alternative credentials outside traditional colleges. Clickers can be useful only when they provoke discussion and peer explanation; otherwise they are just a technological wrapper around lecture-based teaching. Much of higher education may function as sorting or signaling rather than durable human capital formation, especially given weak long-term retention of many course outcomes. College may be too expensive and overused for some students, while not necessary for others who could learn more efficiently through self-directed, online, or alternative paths.

Data Points: MOOC dropout rate: 90%-95% - Kling’s estimate of the share of students who do not benefit from massive online courses Stanford AI course enrollment: 100,000+ students - Roberts cites Sebastian Thrun’s online artificial intelligence course as an early MOOC example Stanford AI course completion: about 20,000 completed homeworks - Roberts argues many students did finish substantial portions of the course Completion rate outside top 100 colleges: 50% or less - Roberts says many institutions below the top tier see very low graduation rates Jobs requiring formal education/licensure in the U.S.: fewer than 5% after World War II; about one-third now - Kling uses this to argue credentialing has become much more important over time High school/college test timing example: 18 months after class - Roberts references evidence that economics course effects often disappear after this period Swarthmore Honors Seminar: about half the students - Kling describes a program where outside examiners set and grade exams independent of the professor Recommended study practice: write ideas in your own words every night - Roberts and Kling discuss self-testing as a learning method

Pivotal Quotes: "Teaching equals feedback." — Arnold Kling: Kling’s central thesis about what makes learning effective "What would you do if there were just one student in the room?" — Arnold Kling: Kling’s guiding principle for designing better educational experiences "The world’s nicest holding pen." — Russ Roberts: Roberts describing college as a place for exploration and transition to adulthood

Implications: The biggest education gains will likely come from personalized, feedback-rich systems and cheaper alternative credentials, not from simply digitizing lectures. Colleges may face pressure to justify tuition by proving real learning, not just signaling.