faculty at Georgia Institute of Technology, where our Online-MS in CS (OMS CS) was praised by President Barack Obama for its innovative accessibility and low cost.

In the last six years, we have also come to understand who is taking MOOCs. We now know that MOOC students tend to be older than traditional college students, have above-average wealth, and are well educated. MOOCs do not serve the masses. They do not serve to replace traditional education, but to augment it. They do not “ democratize education” as many had hoped.

Recent innovations in online learning are proving to have a “rich get richer” effect—those already likely to succeed benefit, and those left behind are left at an increased disadvantage. We argue that our current technology is further undermining educational equity. Computer science departments have an ethical mandate to do better.

Rawlsian Justice

Is it OK to work to help the already advantaged? Certainly, the latest smartphone and the latest luxury car are unapologetically created for the privileged. Our current education system is regrettably not so different from a luxury car—for a price, deluxe experiences are available that advantage the children of the rich and help reproduce their privilege. However, we have higher aspirations. We hope education can serve as a leveler, helping everyone to reach their full potential.

Some difference in privilege is necessary and even desirable to create a thriv-ing culture. How much privilege is OK? What are our obligations to work toward greater equity, for a society that aspires to be just? These profound questions were most eloquently addressed by the philosopher John Rawls. Rawls was an ethicist who argued that for a just society, “social and economic inequali-ties are to be arranged so that they are both to the greatest benefit for the least advantaged and attached to offices and positions open to all under conditions of fair equality of opportunity.”6 Rawls called this “the difference principle.”

Most undergraduate computer science majors are taught about Rawlsian Justice. ABET-accredited programs must include a course in computing and society, which in-

We used to think
MOOCs were going
to change higher
education and
would democratize
education.

cludes ethical frameworks. This definition of justice is ours. It is the one we teach our own students.

Using Evidence to Tell Us If We Are
Reaching the Least Advantaged

We used to think MOOCs were going to change higher education and would democratize education. In 2012, a reasonable person might have seen development of MOOCs as a way to bridge social and economic inequi-ties. By creating MOOCs, CS departments could reasonably claim they were using their privilege to provide great benefit to the least-advantaged members of society.

Today, we have evidence MOOCs do not work like that.

People who take MOOCs already have access to education and tend to be wealthy. Over 60% of MOOC participants already have undergraduate degrees.1 People who take MOOCs tend to be wealthy. A 2015 paper5 reports, “[MOOC] registrants on average live in neighborhoods with median incomes approximately .45 standard deviations higher than the U.S. population.”

Analyses of Georgia Tech’s OMS CS shows the students who apply to the program are demographically different from those who take a face-to-face MS CS program.4 The average OMS CS applicant is a 34-year-old mid-career American, while the average in-person applicant is a 24-year old non-American. MOOCs reach a population that would be unlikely to get a master’s degree in another way. The program is transformative for mid-career professionals—people who are already successful, but aspire to more in their careers. As information technology is increasingly becoming critical to every

aspect of our society, OMS CS is playing a global role in preparing us for the future. MOOCs are well worth offering, but the population being served tends not to be the least advantaged.

We now know that MOOCs as we have used them so far violate Rawls’ Difference Principle—we are further advantaging the already advantaged. We have an ethical mandate to do better.

We Have to Check
If We Are Doing Better

How do we reach the least-advantaged students? Around the U.S., we can see CS departments trying a lot of ways to provide access to CS education. They are offering summer camps, putting their undergraduates into high school and elementary classrooms to help teach CS, or creating “road shows” to demonstrate computer science to elementary or secondary school students who may not know what computer science is. Some of these work. Many do not.

Often, providing computing educational opportunities to “everyone” operationally means only the most-ad-vantaged students actually get access. Free and open summer camps are often filled first by the most-privileged students who tend to hear about the camps and fill them before less-privi-leged students get a chance.

It is challenging to figure out how to make free and open resources available to less-advantaged students. For example, our colleague Betsy DiSalvo found that many free CS learning resources are never discovered by disadvantaged families simply because the families do not know the right terms to search for.2 The for-profit companies are better at tailoring their websites so their resources are the first to appear for the terms that (for example) immigrant families use when searching for learning resources.

Researchers are still working to understand why MOOCs fail students from less-advantaged backgrounds. Access is part of the problem. An experiment offering Udacity MOOCs to San Jose State University students was ended early because the online students had disappointing performance compared to the face-to-face students. Part of the problem there was that the online students did not always have access to broadband Internet when at