Posted on October 8, 2020

Radicalized Antiracism on Campus—as Seen from the Computer Lab

Stuart Reges, Quillette, September 29, 2020

The campus battle over what I’ve previously called the equity agenda has recently shifted almost completely from a focus on gender to a focus on race. This has been accompanied by a series of surreal spectacles at the University of Washington in Seattle, where I teach. In the aftermath of the George Floyd protests, student activists have made new demands upon the school’s administration, while scathingly denouncing anyone they perceive as dissenters.


The background to this is a petition containing seven demands put forward by the university’s Black Student Union, including a call to remove a statue of George Washington that’s been on campus since 1909. Protestors have installed “resistance art” at the base of the statue, painted it red, plastered it with posters, and left messages in chalk. At first, university staff attempted to clean the statue and remove the art, but eventually they just allowed it to accumulate. Very little came of the protest other than informing Cauce that they consider her a traitor for refusing to immediately submit to the petition’s demands.

As at many other institutions in the United States, the public focus of administrators has turned to the idea of antiracism. The highest-ranking diversity officers from our three campuses sent a joint email in late May with the title “Antiracism work is all of our work.” The authors explained that “We are united and unequivocal that antiracism must be at the core of all we do if we are to dismantle the destructive and oppressive effects of white privilege and systemic racism, which is the cornerstone of all U.S. social institutions, including our criminal justice system.” It was unclear whether they were merely pandering to the protestors, or whether this message signaled a serious institutional commitment to substantively overhaul existing policies. Either way, it seemed prudent to educate myself on the topic, and to consider whether I should make changes in my own courses at the university’s Paul G. Allen School of Computer Science and Engineering.

I began by reading Ibram X. Kendi’s 2019 bestseller, How to Be an Antiracist, since this is the book so many people are talking about, and since my university’s antiracist messaging seems consistent with Kendi’s broad denunciations of American society.


Kendi mentions that he does not like to use phrases like “systemic racism” because he considers them vague. He prefers to zero in on the set of identifiably racist policies that, as he defines them, “produces or sustains racial inequity between racial groups.” But he then proceeds to define these terms in a way that indicts pretty much the entirety of American society. By Kendi’s analysis, in fact, I am an unrepentant racist who has perpetuated racial inequity in every policy decision I have made in my career. So is everyone I know. I was hoping to find some middle ground where Kendi and I might meet. But I couldn’t.

* * *

Where racial inequities are concerned, there’s plenty of data for Kendi to consider in the field of computing. The Computing Research Association reports that blacks receiving computer-science PhDs, working in technical roles at prominent companies, and teaching computer science at universities represent less than two percent of personnel in these areas. The data show that the pipeline that produces professional computer scientists is producing very few black computer scientists.

I am most familiar with the portion of this pipeline that involves introductory computer-science courses for undergraduates, sometimes described as CS1 and CS2. I have also been heavily involved in the Advanced Placement exam, known as AP/CS A, which is a College Board CS1 course similar to the one I teach at UW. I pulled data from the College Board to produce the following chart, showing the average score obtained on the AP/CS A exam, broken down according to self-reported racial identity. The exam is scored on a five-point scale, which can be thought of as roughly equivalent to the grades of A, B, C, D, and F. (Anything below three is considered a failing grade.)

During the 23-year period covered by the available data, black students have scored, on average, about a point below whites and Asians. Scores have been going up over time for all groups, but the gap has persisted. In 1997, the gap between black students and the white/Asian average was 1.09. It has since grown to 1.26, as of 2019. All of this is publicly available information.

By Kendi’s definition, AP/CS A is a racist test because it produces unequal outcomes. That makes me a racist because I regularly participate in the annual exam reading. In fact, I was the chief reader when the AP/CS A exam was introduced, which means that I decided how it should be graded and how raw scores should be converted into AP scores.

Is it possible that the course is the problem? I have been teaching a similar course for over 30 years, and have heard a litany of complaints, the most of common which can be paraphrased as follows:

  • The focus on computer programming is too narrow, and will fail to motivate women and underrepresented minorities.
  • Having students work individually on programming problems gives the wrong impression of computer science, and fails to expose students to the high degree of collaboration required in the field.
  • Focusing on just programming does not allow students to explore the broad social implications of computing, including how different communities have been impacted by computers.

You might imagine that a different course, one that addressed these concerns, might help overcome the racial performance gap. But why just imagine? A group of computer-science educators has spent years trying to develop exactly such a course. The result is known as AP Computer Science Principles. In this course, programming is just one of several major topics. Other big ideas include creative development, and studying the impact of computing on society. Students work on projects in groups on topics that interest them.

Unfortunately, the results have been underwhelming. The percentage of black students taking the exam went up, but their average relative performance did not. We have just three years of exam data, but in each of those years, blacks scored, on average, a full point below the average for white and Asian students. In 2019, for instance, the scores averaged 2.3 for black students, 3.26 for white students, and 3.47 for Asian students.

The best minds in our field—and the ones eager to remedy the type of inequities that Kendi identifies—have failed to produce a course and exam that can close the performance gap. So what should we do instead? Both AP CS exams test mastery of programming skills and computer-science concepts. Students who fail to master these concepts will have difficulty creating or analyzing computer software. What is the alternative?

In some circles, there is a taboo against describing this kind of data—even though, as noted above, it is public information. But how can you solve a problem like racial inequality without measuring its effects? Kendi himself would seem to agree with this principle, at least in its broad form, or he would not have cited home-ownership data and similar statistics in his book.

But Kendi also writes that, “the idea of an achievement gap means there is a disparity in academic performance between groups of students; implicit in this idea is that academic achievement as measured by statistical instruments like test scores and dropout rates is the only form of academic ‘achievement.’” In other words, perhaps the black students are just as accomplished as their white and Asian counterparts, but the exam is misfiring as a measurement tool. In this case, the only real alternative would be to abandon testing altogether. And I worry that proponents of antiracist policies will soon be urging administrators to take this step.

Kendi might consider the elimination of tests—or, at least, this kind of test—as a win for black students, because they will no longer be collectively disadvantaged by what he sees as a racist exam. But in fact, we would be harming individual students, by denying them the opportunity to demonstrate their level of mastery. That includes the black students who receive high scores on these exams, and who can go on to pursue educational and professional opportunities on that basis.

On the other side of the country, these ideas are playing out in the debate about how Thomas Jefferson High School for Science and Technology in Alexandria, Virginia should admit students. TJ, as it is affectionately known, has consistently been ranked the top high school in the nation by US News & World Report. It has the most impressive high-school computer-science program that I know of. I have worked with TJ computer-science faculty for over 30 years. Admission to TJ is highly competitive, and is based mostly on the results of an anonymously scored exam. But as a Quillette author recently discussed, the Fairfax County School Board is proposing to replace the exam with a lottery system, and its justification for doing so is very much in line with Kendi’s argument. The perceived problem, as a recent news article reported, is that TJ’s “student population has remained mostly white and Asian.”

The TJ story is playing out in many other places. A technology magnet school in Seattle switched to a lottery system for admissions four years ago. Hundreds of colleges and universities dropped their SAT/ACT admissions requirement because of COVID-19, and are considering making the change permanent in order to address diversity concerns. Magnet schools in New York City are considering dropping their entrance exams. Several departments at my university are considering dropping the GRE as an admission requirement to graduate programs, under the same rationale.

The result will be that the competence-sorting function that once was the domain of admissions officials will now be kicked down the road to professors, who, in turn, will be pressured to maintain a racially balanced grade curve. Inevitably, employment recruiters will have to take on the sorting role, perhaps by administering the same kind of basic tests that schools are now shunning.