An innovative form of cheating emerges in MOOCs

What if Massive Open Online Courses (MOOCs) aren’t actually courses at all?

Our research teams at Harvard and MIT have shown over and over again that MOOC students look and act nothing like conventional students of either residential universities or online programs.

With a broader age distribution, a more diverse and international student body, wide variation in commitment, and a surprising number of teachers, the MOOC “classroom” looks like no physical classroom on earth.

Now, I and my colleagues, Curtis Northcutt and Ike Chuang of MIT, have discovered a different novel behavior on the MOOC frontier: a new form of cheating, that only MOOCs can enable.

A new cheating technique emerges in MOOCs

Our research has identified a cheating technique in MOOCs where people exploit the ability to create “multiple personalities” online. We call this CAMEO (Copying Answers using Multiple Existences Online).

Here is how it works.

A CAMEO user registers on a MOOC platform not once but twice. This creates two identities. The first identity is a “harvester” that uses a guess-and-check strategy to gather correct answers. The second identity is a “master” that submits the correct answers.

In some courses, people can use CAMEO to obtain a certificate of completion in under an hour.

This kind of cheating is difficult if not impossible in conventional courses.

The prevalence of cheating in MOOCs

Reassuringly, this doesn’t happen too often. Only 1.3% of certificates (1,237 certificates across 69 courses) have been earned by CAMEO.

Unfortunately, the percentage is much higher for “super-certifiers” – those who have earned 20 or more certificates. We found that 25% of these super-certifiers have used CAMEO.

Cheating on MOOCs is very different from conventional cheating.Nicole Abalde, CC BY-ND

Thanks to our algorithm, we are able to tell when achievements in the MOOC space really are too good to be true.

We also found intriguing differences in CAMEO prevalence – both across courses and across countries. For instance, of all certificates earned in computer science courses, only 0.1% were earned by CAMEO cheating.

We believe this is because of the more authentic assessments in these courses, which involve solutions that are difficult to harvest.

Rates in the US were lower (0.4%) than rates in countries like Albania (12%), Indonesia (4%) and China (2%). These rates may indicate that the perceived value of MOOC certificates is higher in these countries than in the US.

The uniqueness of CAMEO

Is 1,237 certificates (1.3%) a significant problem?

We believe it is. CAMEO is not like conventional cheating from friends or online sources. When you can copy from yourself, you can copy endlessly. This can result in the outright falsification of a MOOC certification.

Although the strategy is novel for academic cheating, it reminded us of other strategies, like “multiple account” tricks in open online forums (sockpuppeting, where users reply supportively to their own posts) and open online games (self-collusion, where users gain points by playing against easier versions of themselves).

Why cheat at MOOCs?

Should we shrug at MOOC cheating? Or should we denounce it? I argue, neither.

The “shruggers” assume that MOOC certificates have no value except motivating students to learn. A “shrugger” thinks that cheaters bypass the learning process and thus cheat no one but themselves.

But MOOC certificates have more value than simple motivation. The motivation is internal (edX’s “Challenge yourself… working towards a certificate keeps you motivated"), instrumental ( Coursera’s “Build your professional qualifications”) and external (edX’s “Share it with the world”). Some high school students are using MOOC credentials to enhance their college applications.

The “denouncers” assume that MOOCs exist to replace or devalue conventional higher education credentials. To the “denouncers,” cheating devalues MOOC certificates and suggests that they will never replace or threaten conventional college courses.

But it’s difficult to conclude that MOOCs are doomed on the basis of moderate prevalence of a particular cheating strategy. As we noted, CAMEO rates are fairly low.

How to discourage cheating without discouraging learning

The important question is not how to prevent cheating, but how to do so without paralyzing instructional designers or discouraging well-intentioned learners.

The most obvious solution is to enact virtual or in-person proctoring, as edX is already planning for higher-stakes uses of MOOCs for college credit.

The question that remains to be answered is whether MOOC providers can keep the costs low enough to increase not only learning but certification for massive numbers of students.

Without massive numbers of trusted certificates at low costs, MOOCs risk their central identity as massive open courses. A MOOC should be more than a free online textbook. Certification should mean something.

CAMEO throws down the gauntlet for MOOCs. We hope they will respond.

Andrew Ho receives funding from the Spencer Foundation and the Institute of Education Sciences for his work on assessment.