jeudi 7 mars 2013

The secret ingredient of MOOC unveiled...

The meteoric entrance of MOOC (Massive Open Online Courses) was certainly the dominant event in the world of higher education in 2012, at least in the United States, where the New York Times described 2012 as the « The Year of the MOOC » [1]. It even mentions the « MOOC Mania » [2], [3], [4].

We will try to understand why we are witnessing not a fad but rather the beginning of an industrial revolution in education.

That said, we can read that MOOCs will finally give little change in current education. These articles highlight its shortcomings and warns against too much enthusiasm [5], [6], [7].

Many convince themselves quibbling on the high dropout rate among users of MOOCs, in the order of 90% [8], [9], or on their simplistic pedagogy, either ensure that the MOOCs are downright ineffective or even dangerous for weak or poorly supported students [10]. In all these youthful errors, they see evidence that the MOOCs are not here to stay or they are just a fad that once past the excitement of novelty leave behind them very low adoption.

Moving quickly on criticism based on the high dropout rate, since it takes only a few clicks to be registered for a course on a MOOC platform. Therefore, the MOOCs attract a large number of lurkers and many enroll in courses for which they do not have the prerequisites. It is in the nature of the MOOCs to give a chance to the greatest number of people to sign up and try.

For the rest, many of the criticisms come from the eternal « resistance to change » or from the imperfect knowledge of the possibilities of Big Data and Machine Learning technologies.

The MOOCs represent a true revolution in progress for four reasons: firstly, their low cost, secondly, their convenience since they are particularly well suited to the digital natives, thirdly, their intrinsic qualities and fourthly, above all their almost unlimited potential to improve and evolve..

Cost advantage

The economic question is at the heart of the MOOC revolution, new technologies challenge the very economy of education by their low cost [11], since free is a hard to beat prices!

Practical advantage

Beyond the mediatic's hype, the first MOOCs are eminently practical as best suited to the needs of the digital natives who are mobile and constantly connected.

Anyone, at anytime and anywhere in the world will have access to courses, insofar he or she has a device connected to the Internet with a browser and enough bandwidth to watch a video online [12].

The vision of small groups of 20 people in a dynamic face to face interaction with their professor is far from the reality of most students. Instead, this romantic vision gives way to that of a large auditorium of 200 places with a teaching assistant a little bit lost on a platform that gives a three hours long lecture to students who half-listen, surf the Web, tweet, write text SMS or chat. Also, very few students, the ones more extroverted and the more motivated, are engaged in a two-way discussion with the instructor or raise their hands to ask questions [12].

In addition, the lack of attention and the typical multitasking behavior of digital natives require that we chopped up the lecture in short clips of 10 to 20 minutes duration, which is a common practice for the first generation of MOOC.

We see the same adequacy of MOOCs for continuing education which is the natural pool of users for the first generation of MOOCs [13].

Another use of MOOCs is called the « Flipped Classrooms » where the students have access to online content and perform exercises before going to class. Then, the real classroom becomes the place where to discuss difficulties and to go deeper into certain subjects [14].

Quality advantage

That said the first generation of MOOCs is already a pragmatic answer able to meet some of the needs of people learning online. Just enough, not too less, not too much. Good engineering practice!

This is the famous pragmatism of the Silicon Valley entrepreneurs, the same one which gave birth to the iPhone that everybody wants today but the first Apple computer in its plywood casing was built in the garage of Steve Jobs parents.

The quality of teachers, contents and pedagogy of the current MOOCs is often better than for traditional courses. Let's be honest, in many ways short video clips accompanied by exercises which are automatically evaluated to check the understanding of students in an active learning strategy described as « Mastery Learning » [15] represent a significant improvement over many courses currently taught in auditorium.

The capacity to improve itself

MOOCs have a huge potential for further development in the future and in several directions, including management, monitoring and certification of competencies. This is also a direction in which I am personally involved.

Indeed, the first generation of MOOCs is still primitive compared to more sophisticated systems such as intelligent tutoring systems and other computer-assisted instruction systems which are far remained confined to laboratories.

This is only the first step. The invisible part of MOOCs is the massive collection of data on the behavior of students. We are talking about Big Data and the results will be used to improve the next generation of MOOCs [16]. We recognize there a common practice of Web 2.0 « the Google like process » of exploiting the data of its millions of users to improve the results of its search engine. Here, user data is a goldmine. There is no free lunch. When the use is free is that the user is the product.

Also, precisely because of their massive nature, the ability of MOOCs to improve and evolve is almost unlimited. In fact, it is hard to believe, but it is quite possible to offer quality education like individual tutoring to a large number of students using Big Data technologies.

In order to improve its course, it is not easy for a teacher to find the sources of confusion and less effective pedagogical approaches from the small samples of data collected with a class of 20 students. At the contrary, a MOOC with its thousands of students can use statistical methods to detect problems and improve teaching. You can also use machine learning to discover situations (or patterns) where students have common problems, in order to present evidence or explanations to help them. Compared to human, a computer is always patient, he never gets angry and he is always ready to resume its explanations, making it ideal teaching tutor. Thus, we will therefore see directly emerging from data fine grained improvements in order to customize education for each student in a way that was simply not possible, because we had neither the time nor the means to do [12]. This is the promise from MOOCs of second and third generation.

One of the biggest blockers toward a wider adoption of MOOCs today is that they do not issue diplomas neither course credits [17]. For this, we should ensure that students are doing their work themselves. One can even imagine situations where students will pay others (outsourcing) to do their homework or exams. But with time, things will evolve and we'll find a way to ensure that students did not cheat and deserved the credits or diplomas that could potentially be awarded to them [18].

In the short term, the best way to avoid cheating is to perform controlled examination (onsite exams). There are test centers where after being duly identified, the student is evaluated in a controlled environment (isolated post, surveillance camera, no internet access, sometimes even rf-shielded room or Faraday cage which prevent using radio frequency waves to communicate). Thus Udacity, one of the three top MOOC providers has signed an agreement with Pearson VUE, which operates some 4,000 test centers in 170 countries [19]. It has even been announced for soon, online surveillance systems based on the analysis of student behavior and eventually their eye movements (eye tracking) from images taken by a small webcam [20]

Obviously, there is a great demand for tools to detect plagiarism. For homeworks, there are technologies to detect fraud based on text analysis, lexicometry, stylistics, statistical analysis and comparison with texts harvested from the Internet and specifically indexed for this purpose. This is a very active R&D subject that is expanding at a remarkable rate.

Again, due to the accumulation of data about students, using statistical analysis and Big Data technologies we will be able to improve plagiarism detection tools and ensure that students do themselves their work.

The two founders of Coursera Mr Ng and Ms Koller are experts in Machine Learning and Big Data and both professors at Stanford the same for Sebastian Thrun the guy behind the Google car who is started Udacity a year ago. Without a doubt, the « big heads » in Machine Learning behind Udacity and Coursera will not stop there. They will use all the data collected about students to build the second generation of MOOCs, then the third...

A disruptive innovation

The ability of MOOCs, to be constantly improved, is the signature of a disruptive innovation. The idea of ​​disruptive innovation has been first stated by Professor Clayton Christensen of the Harvard Business School in his book « The Innovators Dilemma » [21].

Perhaps the word « disruptive » was a bit hackneyed about MOOCs. But the word « disruptive » has a specific meaning when it comes to « disruptive innovation » and MOOCs exhibit the essential characteristics of a « disruptive innovation ».

According to Prof. Chistensen's definition, « A disruptive innovation is an innovation that first established itself in simple or "low end" applications, then moves inexorably towards the "high end", to eventually dislodge the established competitors » [22]. A major advantage of disruptive innovation is to pass relatively unnoticed, as judged somewhat menacing in its infancy by established competitors, until it is too late [23], [24], [25]. Precisely, the first generation of MOOCs is a pragmatic and cheap answer already able to meet many of the needs of people who learn online. It's good engineering! Just enough, not too less, not too much. So nothing to worry...

So we saw the potential for evolution and continuous improvement of MOOCs mainly due to Big Data, the secret ingredient of MOOCs.


I could be wrong, but MOOCs have the potential to be a disruptive innovation. This would have huge implications!

In a futuristic and somewhat utopian video, but which has the merit to shake the belief, the EPIC 2020 group says education in the world will dramatically change over the next decade [26].

Assuming that this will not be really the case, we can trust that MOOCs will minimally force higher education institutions to invest in distance learning.

[1] Pappano, L. (2012, November 2). The Year of the MOOC - Massive Open Online Courses Are Multiplying at a Rapid Pace. The New York Times.
[2] Davidson, C. N. (2012, October 1). MOOC Mania. reblogged from the Chronicle of Higher Ed. Blog,
[3] Vardi, M. Y. (2012). Will MOOCs destroy academia? Communications of the ACM, 55(11), 5–5. doi:10.1145/2366316.2366317
[4] Martin, F. G. (2012). Will massive open online courses change how we teach? Communications of the ACM, 55(8), 26. doi:10.1145/2240236.2240246
[5] Boullier, D. (2013, February 20). Mooc : la standardisation ou l’innovation ? «
[6] Greatrix, P. (2012, October 8). MOOCS: 12 Reasons for universities not to panic. Registrarism. Blog.
[7] Greatrix, P. (2013, February 15). Why MOOCs won’t kill universities. Registrarism, Blog
[8] Balch, T. (2013, January 27). MOOC Student Demographics. the augmented trader. Blog.
[9] Guzdial, M. (2012, April 20). Udacity’s CS101: Who are you talking to? Computing Education Blog. Blog.
[10] Rosenthal, A. (2013, February 18). The Trouble With Online College. The New York Times.
[11] Guillaud, H. (2012, October 17). L’innovation éducative : une question économique ?
[12] Schmidt, D. C. (2013, January 7). Episode 191: Massively Open Online Courses, Software Engineering Radio episode 191,
[13] Webley, K. (n.d.). MOOC Brigade: Who Is Taking Massive Open Online Courses, And Why? Time.
[14] Hopkins, C. (n.d.). Future U: fear and loathing in academia | Ars Technica.
[15] Mastery learning. (2013, February 24). In Wikipedia, the free encyclopedia.
[16] Daphne Koller: What we’re learning from online education | Video on (2012).
[17] Anders, G. (2012, June 5). How Would You Like A Graduate Degree For $100? - Forbes. Forbes.
[18] Bates, T. (2012, October 29). MOOCs move into credit-based higher education. online learning and distance education resources.
[19] Udacity. (2012, June 1). Udacity Blog: Udacity in partnership with Pearson VUE announces testing centers. Blog.
[20] Eisenberg, A. (2013, March 2). Keeping an Eye on Online Test-Takers - New Technologies Aim to Foil Online Course Cheating. The New York Times.
The companies: ProctorU and Software Secure
[21] Christensen, C. M. (1997). The Innovator’s Dilemma: When New Technologies Cause Great Firms to Fail. Harvard Business Press.
[22] Bass, R. (2012, March 1). Disrupting Ourselves: The Problem of Learning in Higher Education (EDUCAUSE Review) | EDUCAUSE. Educause. Educause Review Magazine, Volume47, Number 2, March/April 2012.
[23] Vidéo d'une présentation du professeur Clayton Christensen sur l'impact l'innovation de rupture que représente les CLOM devant un comité du sénat de l'État de Utah « Higher Education Appropriations Subcommittee » aux États-Unis. Harvard professor Clayton Christensen, disruptive innovation and higher education. (2012).
[24] Schubarth, C. (2013, February 13). Disruption guru Christensen: Why Apple, Tesla, VCs, academia may die. Silicon Valley Business Journal. newspaper.
[25] Michael Horn, & Christensen, C. M. (2013, February 20). Beyond the Buzz, Where Are MOOCs Really Going? | Wired Opinion | Wired Opinion. Magazine.
[26] Bill Sams, Marshall, N., Kelvin, M., & Hanlin, M. (2012, June 4). EPIC 2020 | Higher Education Reform. EPIC 2020 | Higher Education Reform.