The participants of #NumericalMOOC will have noticed that we made only one video for the course. I thought that maybe I would do a handful more. But in the end I didn’t and I don’t think it matters too much.
Why didn’t we have more video? The short answer is budget and time: making good-quality videos is expensive & making simple yet effective educational videos is time consuming, if not necessarily costly. #NumericalMOOC was created on-the-fly, with little budget. But here’s my point: expensive, high-production-value videos are not necessary to achieve a quality learning experience.
The fixation with videos in MOOCs, online courses and blended learning is worrisome. At the edX Global Forum (November 2014), it was often mentioned that producing a MOOC is a high-cost operation, with an estimated average expense of $100,000 per course. This is probably a somewhat overindulgent price for appearance, rather than substance. There is no evidence justifying the “production value” from a learning perspective. In fact, as far back as 1971, Donald Bligh concluded that “there is not much difference in the effectiveness of methods to present information.”  In this sense, a video—however nicely produced—is not better than a lecture.
A personal anecdote
I recently decided that I needed to brush up on my knowledge of statistics. As a student of mechanical engineering, I took an undergraduate “Probability and Statistics” course years ago, but I don’t remember that much. With an interest in education research, I now feel I should be in command of one particular concept: p-values.
So, I registered for a MOOC: “Statistics in Medicine,” by Kristin Sainani, on Stanford Online. The content of this course is excellent and Sainani is brilliant: she uses many examples from the medical literature that make the subject come alive. Of course, I could hardly have the time to complete a traditional MOOC (with deadlines and a set end date)—last Fall, I was designing and teaching my own MOOC!—and the best I hoped for was to watch the lectures dealing with the concepts I wanted to learn.
I watched the first few lecture videos, to get a baseline, then cherry-picked the videos dealing with p-values. It was perfect to watch the vids on my iPad while on the treadmill, getting some needed exercise at the same time. The concepts were clear: I could follow the explanations easily and the examples put things in context and helped me understand the importance of knowing statistics!
But two weeks later … I couldn’t remember how sample size affects significance, why statistical significance did not imply clinical significance, and how confidence intervals are related to p-values. I had to watch the videos one more time, then everything was clear again.
But a month later … You know where this is going.
Now, consider this: I have a PhD in Aeronautics from Caltech—I’ve proved that I am a “good student.” Yet, without manipulating the new concepts through writing things down, making summaries, diagrams, working through examples and so on … I just forgot.
Intro video for Practical Numerical Methods with Python
Watching videos is not better (or worse) than sitting through lectures
Videos are nice, they can get you exposed to a new concept for the first time in an agreeable way, but they do not produce learning, on their own. Students need to engage with the concepts in various ways, interact with ideas and problems, work through a process of “digestion” of the learning material.
Granted, the general statement above refers to a particular kind of video: the expository “lecture video,” usually a narrated Power Point and sometimes a “talking head” video. These are no better (or worse) than traditional lectures. They are simply convenient because there is no need to travel and sit in the same room with other students and instructor. And because you can rewind and watch again.
Despite their popularity in MOOCs and flipped classrooms, “lecture videos” have the same pitfalls as regular lectures: they provide a false sense of clarity and are utterly forgettable.
Educational-video champion Derek Muller (Veritasium) claims, for example, that typical physics videos do nothing to clear students’ misconceptions—unless these misconceptions are tackled head on, creating a sense of confusion. But most educational videos don’t do that: they try instead to be “clear” and avoid confusion.
Unfortunately, most of the video I see in MOOCs is not very effective, despite the so-called production value. I question the rationale behind spending large sums in making documentary style videos and placing emphasis on the amount of time students spend watching.
What are videos good for?
Institutions like the UK’s Open University have for decades been researching the best uses of video ; among several others: to show complex (or expensive, or dangerous) experiments; to illustrate ideas using slow-motion, fast camera or animation; to substitute for a field visit; to demonstrate techniques or mechanical skills.
Screencasts using on-screen annotations (a.k.a. digital inking) are certainly useful for things like homework-solution review and exam preparation. But students tend to use them strategically, sometimes viewing a video or a particular portion multiple times.
With more time, I would certainly like to add a handful of screencasts to #NumericalMOOC to support some theory, explain tricky concepts or demonstrate a technique. But that would be icing and not the cake. Quality learning is happening without them, because we combine learning pathways, instructional scaffolding, interactive computing with our IPython Notebooks, and independent student work.
“Lecture videos are a central feature in the student learning experience in nearly all MOOCs,” advises the MITx Office of Digital Learning.  I think this advise is misguided. Their recommendations to make videos better are sound (keep them short, informal, etc.), but the overall emphasis is too much on the instruction, and too little on the student—which is where learning really happens.
The problem with making videos “central” to the student experience is that it comes at the expense of higher-order learning activities. More worrying is that students will spend almost all their time watching videos, as if that could magically elicit learning, without the hard work.
Videos can be one device for building a MOOC or a small online or blended course, but not generally the most important one. We need to acknowledge the limitations of video and place emphasis on authentic learning and not just “engagement” (time watching, # of clicks). 
I’ve been making screencasts for ten years and my videos on computational fluid dynamics have amassed more than 300,000 views. Yet for #NumericalMOOC, we designed the central learning experience around a set of IPython Notebooks, and meaningful yet achievable mini-projects for students. I guarantee learning results to any student that fully engages with these! “Dazzling presentations do not necessarily result in learning.” [Bligh, 2000; p. 121] Donald A. Bligh, “What’s the use of Lectures?” Jossey-Bass Publishers: San Francisco, 2000 (first US edition of the revised 1971 work)  Derek Muller, “Khan Academy and the Effectiveness of Science Videos.” http://youtu.be/eVtCO84MDj8  Bates, A. (2005) Technology, e-Learning and Distance Education London/New York: Routledge. See: http://www.tonybates.ca/2012/03/10/pedagogical-roles-for-video-in-online-learning/  K. R. Green, T. Pinder-Grover, J. Mirecki Milunchick (2012) “Impact of screencast technology: Connecting the perception of usefulness and the reality of performance,” Journal of Engineering Education, Vol. 101, No. 4, pp. 717–737.  MITx: Video Engagement in MOOCs, http://odl.mit.edu/mitx-video-engagement-in-moocs/  P. J. Guo, J. Kim & R. Rubin (2014), How Video Production Affects Student Engagement: An Empirical Study of MOOC Videos
(c) 2015 Lorena A. Barba, released under Creative Commons Attribution License CC-BY 4.0