CS in Crisis: Is Stanford doing enough to respond to capacity and inclusion challenges?

Feb. 19, 2019, 12:35 a.m.

“STANFORD THREATENS LIMITS ON NUMBER OF CS MAJORS,” the sensationalist all-campus email newsletter The Fountain Hopper (better known as The FoHo) alerted readers in fall 2016, during my freshman year. There wasn’t a lot of substantiating evidence beyond the fear mongering title, and the information was later revealed to be false. Nevertheless, a panic ensued in my freshman dorm, although most of my friends were a ways away from even thinking about declaring.

Undergraduate enrollment in the computer science (CS) department at Stanford has quintupled in the past 10 years. As former CS department chair Alex Aiken once put it in a newsletter, expansion “has varied only between ‘rapid growth’ and ‘very rapid growth.’” This trend is paralleled nationwide. Between 2009 and 2015, the Computing Research Association found that CS degrees awarded at Ph.D.-granting universities increased by almost 300 percent.

Despite tremendous interest in computer science at universities across the country, degree conferral continues to trail labor demand. Data from the Bureau of Labor Statistics indicates that from 1975 to 2015, employment in computing increased twice as fast as the production of degrees in computing and information science. Even if enrollments hit a plateau at Stanford sometime in the next few years, the long-term continued growth of the department is virtually guaranteed.

At Stanford, the dramatic growth of the undergraduate major has generated tensions within a department that preaches inclusion yet faces formidable strain on its limited resources. Today, the CS department is home to just four percent of total University faculty and staff but claims 20 percent of all undergraduate majors. An investigation conducted last year by a visiting committee of non-University computer science professors and industry professionals describes the growth of Stanford’s CS department as “unsustainable” — a word repeated three times throughout the report. Moreover, it highlights that CS faculty are leaving at a historically unprecedented rate: Twice as many faculty have left the CS department in the past decade than in the previous 40 years combined.

With little reason to believe that the number of undergraduates declaring CS will slow in the near future, is Stanford doing enough to ensure that its CS education will continue to be first-rate?

“Pump, not a filter”

At this project’s inception, I was primarily interested in how Stanford’s CS department has achieved diversity within its undergraduates that is by and large unheard of at other universities. In 2015, Reuters reported that CS had become the top major for women at Stanford, and currently, almost 35 percent of declared CS majors at the University are female. In comparison, only 18 percent of bachelor’s degrees awarded in computer science go to women nationwide.

Instructors and faculty within the department readily admit that socioeconomic and racial diversity still needs a great deal of work. But I remained curious about the conditions that made it possible for CS to be more welcoming to women at Stanford than at other institutions. I quickly realized that the best person to direct my line of historical inquiry at was the person who started it all: Eric Roberts.

Upon graduating from Harvard University in 1980 with a Ph.D. in applied mathematics, Roberts set about his job hunt in an unconventional manner. Instead of sending applications to schools with open positions, he wrote directly to the presidents of women’s colleges without established CS departments, determined to bring the field to one of these schools.

In 1980, Roberts became the first full-time CS faculty member and department chair at Wellesley College, a first step in a storied career of increasing diversity in the field through approaching undergraduate education with intentionality. In 1985, Roberts left Wellesley when the department began imposing undergraduate enrollment limitations. He later wrote in a memo that the restrictions made “the relationship between faculty and students adversarial, causing students to become more competitive and, in many cases, angry. Teaching became considerably less enjoyable.”

In 1990, Roberts joined Stanford faculty as the first full-time member responsible for CS undergraduate education. Today, he is widely recognized as the founding figure of the undergraduate CS curriculum at Stanford. As associate chair and director of undergraduate studies in the CS department, Roberts became the principal architect of the highly popular introductory computer science series that draws thousands of students every year from across all seven schools. In that role, Roberts authored seven textbooks now used at high schools and universities worldwide, and he designed a programming sequence that represents a paradigm for CS departments globally today.

At the core of Roberts’ philosophy is a commitment to increasing enrollment numbers through making computer science accessible. Such an approach favors prioritizing the number of women and underrepresented minorities enrolled rather than the percentage. His rationale for establishing a principle of inclusivity rather than selectivity remains as relevant today as it was in the 90s, when Roberts embarked on the task of developing a program from the ground up.

Building numbers indiscriminately inevitably diversifies the pool of potential mentors and peers, increasing the likelihood that members of an underrepresented minority group form a “critical mass” and build a supportive peer community. And given the dire American labor shortage in computing fields, restricting the supply of CS degrees conferred may seem unjustifiable.

To increase recruitment, enrollment and retention, the department has targeted each component of undergraduate education. While many other departments in the country treat their introductory sequence as a series of “weeder” courses, Stanford CS has taken the attitude that its elementary courses should be “feeders for the major.”

“As a staff, our commitment is to make the bar high and do everything we can to get students over the bar,” said Cynthia Lee, a lecturer for two core CS classes.

To address swelling class sizes and provide mentors, the department installs what it describes as “stepping-stone role models” — undergraduates, graduate students, lecturers and faculty who offer guidance to students at various points in their University careers. Adjunct classes that provide additional problem-solving practice like CS 107A: “Problem-solving Lab” for CS107 are offered to give students extra help. And bridge programs like the Stanford Summer Engineering Academy introduce computer science and engineering to incoming freshmen to help promote retention in classes and the major.

Building on the ethic of improving accessibility, in 2008, the CS department commenced a redesign of the curriculum to allow students to personalize their education according to their interests. Instead of the former “one size fits all” approach that required students to adhere to a rigid set of courses, an array of tracks was rolled out to allow students to specialize in a diverse set of subfields, including artificial intelligence, biocomputation, graphics and more. Today, students can choose from any of nine tracks, with the additional possibility of individually designing their own track, as opposed to the more restrictive one-track option that existed previously.

Mehran Sahami, the associate chair for education in the CS department, attributed the curriculum revision process to increases in both enrollment and diversity, as well as a rise in the percentage of women in the major. Sahami said that “there’s actually a pretty strong correlation” between the revamped curriculum and the changes in undergraduate diversity in the department.

The principle of inclusion is ingrained in the ethos of Stanford’s computer science department. Instructors and professors religiously repeat the mantra that each course should be a “pump, not a filter.”

“Every method we’ve seen … of constraining the size of the major will harm diversity,” said Philip Levis, associate professor with a joint appointment in computer science and electrical engineering. “Given things like job prospects and the financial lucrativeness of computer science, that’s essentially a form of economic oppression.”

It is not merely coincidental that all faculty members are on the same page about inspiring student interest rather than restricting access. Sahami referenced a CS faculty retreat six to seven years ago, at which participants unanimously voted against a quota for student enrollment in the major.

Since then, Sahami said the department frequently revisits that undivided agreement to double down on its commitment to undergraduate education. Being at Stanford has numbed me to the remarkable nature of Stanford CS’ “big tent” philosophy — at other universities, aggressive weeder courses and even outright rejection from CS programs are common. For comparison, at UC Berkeley, the Electrical Engineering and Computer Sciences (EECS) major is restricted to those who are admitted directly to the program, and students interested in the computer science major must apply and maintain a minimum 3.3 GPA. At Carnegie Mellon, students must either be accepted into the School of Computer Science or transfer in. These schools exercise total control over the size of their undergraduate CS population, with the obvious disadvantage of limiting access and diversity.

I pressed Roberts to explain how Stanford’s CS department has been uniquely able to coalesce around a central organizing principle of accessibility when so many other departments and schools have capitulated to enrollment caps.

He cheekily responded, “I was chairing the department for a long time, and the people I hired had that view!”

The department has stayed true to its commitment to accommodate any student with an interest in the subject, and the size of both undergraduate enrollment and majors is growing. About 95 percent of Stanford undergraduates take some CS course during their time at Stanford; therefore, an overwhelming majority of students graduate with some exposure to the discipline.

Today, CS majors represent approximately 20 percent of the graduating class, and each year, this proportion continues to expand. Gender diversity within the major has followed suit.

“I think relative to places just like Stanford — the Harvards, the Princetons — Stanford is doing much better,” Roberts says.

Given this context, it’s clear why Alex Aiken, then department chair, and Sahami were so amused by the FoHo’s notion that Stanford would cap its CS program. In a follow-up email to the FoHo, they hypothesized that the rumors were spread as part of a clever tactic intended “to get students planning to major in CS to actually declare before the last quarter of their senior year.”

Anyone who knew a thing at all about Stanford CS would know that, as Aiken and Sahami put in their follow-up message, “there is zero chance the department will move to limit the number of CS majors. As in zilch.”

The department is rightfully proud of its steadfast dedication to and success in serving interested students. But in my research, I also heard concerns that if it proceeded without dramatic changes, the CS department would soon collapse under its own weight.

“They don’t know what they don’t know”

“What do you think —”

She was clearly exasperated. She paused, collected herself and let out a laugh. “What do you think the student to teacher ratio is for computer science?” Monica Lam, who has been a professor in the CS department since 1988, implored me to do a calculation and to let the numbers speak for themselves.

This past fall, the average CS class contained approximately 120 enrolled students. But even this number is generous because it includes smaller introductory seminars and graduate level courses, many of which require the instructor’s consent or an application to join. The student-to-faculty ratio within the department is approximately 12 to one. Excluding emeritus, courtesy, adjunct and visiting faculty, the ratio is closer to 20 to one. By contrast, the University advertises a four-to-one ratio school-wide.

What is unusual in CS is that class sizes for higher-level courses do not dwindle in size for upperclassmen who have completed the core. For example, this past fall, 1,040 students were enrolled on the first day of class for CS 229: “Machine Learning,” a course with a roster of more Teaching Assistants (TAs) than most humanities seminars have students.

Recorded lectures, long lines at office hours and automated grading have become staples of the undergraduate computer science experience. Dixee Kimball ’18, a senior majoring in CS, said that he “mostly [doesn’t] go to lecture” and instead learns from online slides and recorded lectures — a fairly common practice among students, he added.

Office hours are frequently crowded, especially in the days before problem sets are due and prior to major exams.

“I’ve seen people show up at office hours and wait for two hours to get the answer to one question,” said Amy Liu ’17, who majored in CS and was a TA for a core CS class. “That’s kind of crazy to me.”

“It’s clear that they grade assignments in a very scaled fashion,” Kimball said, referencing the widely used online grading platform Gradescope, which allows students to submit assignments online and receive digital feedback from instructors on their work.

He acknowledged that the staff has done a “reasonable job” given the quantity of grading they must do but expressed that “for the sake of learning, it could have been better to have more personalized feedback.”

Beyond the classroom, even the undergraduate advising program — deliberately put in place to foster personal relationships between CS faculty and declared students — has become ineffective given the sheer number of students it must accommodate.

“I don’t actually know the name of my advisor,” Kimball responded when I asked whether he had ever made use of faculty advising in the department. “Certainly quite a few of my friends either don’t meet with their advisors or have met their advisor [just once],” he added.

From a faculty perspective, lecturer Cynthia Lee echoed Kimball’s insights. “Advising is the one thing where the scaling issue is just much more difficult to tackle,” she said.

Lee indicated that she currently has around 50 advisees and said that it would be impossible for her to serve as “a one-on-one tour guide for life and for the major and career for that many people.”

The uniformly large class sizes in computer science suggest that students may complete the major without ever engaging with instructors and professors in a smaller, more intimate setting — an essential component of the individualized learning experience that Stanford prides itself on. When I talked to Sahami about large class sizes, he expressed the difficulty in balancing the desire for individualized attention with accessibility to the major.

At the beginning of the 2017-18 academic year, enrollment in CS courses surged by an additional 10 percent. The very fact that students enthusiastically pour into CS classes each year despite decreasing individual attention from faculty indicates that the department, at least to some extent, has found innovative ways to scale up.

The section leading program — in which undergraduate instructors are paid to host small seminars for introductory CS classes 106A and 106B — was established by Roberts before the recent upswing in enrollment and has evolved into an important tool for managing swelling class sizes in the introductory series.

Julia Daniel ’17, formerly the head TA of CS 106A and currently a TA for CS 103, identified the culture of students sharing knowledge with peers as a key advantage of how the CS department has responded to large classes.

“[The undergraduate section leader model] is super scalable because you’ll generally have a pretty solid proportion of students who want to take on that role, relative to students who want to take the class,” she explained. “That can scale to any particular size you want. That works even in classes with over a thousand students.”

These scaling strategies are inventive and have alleviated resource concerns. But they do not compensate for the lost benefit of directly interacting with experts in the field. As Roberts puts it, “One of the things that you like to do in a university is give students the chance to interact with the people who are … really changing the field.”

When I asked Lam about what she saw as consequences of the growth in enrollment and large class sizes, she told me, “You should ask the students. I think that sometimes they don’t know what they don’t know.”

She hesitated, conceding that she had just contradicted herself. “There are people who think that, ‘Oh, it’s fine — I can handle these large classes,’” Lam continued, “but they have not seen what it means if the classes are smaller.”

She added, “Especially in computer science, there are a lot of projects … a lot of creativity. We can help, but not when the sizes are so big.”

Her final assessment? The department “needs drastic changes.”

“Collapse in morale”

The second time I met with Eric Roberts, he sat down and rummaged around in his backpack for half a minute before fishing out a thick stack of paper. Each department is visited once every 10 years by outside consultants — typically, professors from other schools and experts in the field — who evaluate the state of affairs and circulate a report internally.

“I can’t let you have this,” he warned me.

I imagined that he brought it along to martial physical evidence to lend weight to his words.

“It uses the word ‘unsustainable’ three times in there,” he said, deftly pointing to various lines in the report. “Everybody wants to think, ‘Well, we can probably weather this storm.’ What universities hate is bad news, so everybody has a can-do attitude, which is fine, except that it makes it much harder if things reach a point where there has to be retrenchment, because then people realize that” — Roberts mimicked a gasp — “‘It didn’t work as we had hoped.’ And the collapse of morale is much more serious.”

What collapse in morale? I asked.

“It’s within the faculty,” he clarified, before comparing the phenomenon at length with the well-documented occurrence of doctor burnout.

Many CS departments at peer institutions are battling similar problems. While bachelor’s degrees in computer science have quadrupled since 2006 nationwide, the number of tenure-line faculty has flatlined. The premium of working in industry poses a persistent threat to the supply of candidates for academic positions. While more than seven percent of undergraduates eventually pursue a Ph.D. degree in every other STEM discipline, only one to two percent of undergraduates who major in computer science pursue a Ph.D. From this vanishingly small slice of students, only about 30 percent accept positions in academia.

Judgment day is just around the corner for the CS department, Roberts warned. While the shortage of faculty is acute for CS departments across the country, Stanford’s retention rate is especially bad. The 2017 Visiting Committee report that Roberts came equipped with found that twice as many CS faculty have left Stanford in the past decade than in the previous 40 years.

There are several reasonable hypotheses for why faculty are leaving that have little to do with soaring enrollments. The affordability crisis in the Valley may be pushing faculty to areas of the country where the cost of living is less exorbitant. And for those who stay, the lure of industry — the opportunity to do cutting-edge research while depositing paychecks that are double or triple a university salary — is proximate and seductive.

Additionally, according to CS professor Philip Levis, “the School of Engineering actually has some pretty strict rules” about taking leave and maintaining industry ties, making it difficult for faculty to balance one foot in nearby tech companies and the other in teaching.

Finally, Levis explained that the CS department at Stanford is relatively more egalitarian with its pay, with at most a “factor of two difference” in salary between its faculty members. “If you look at a place like UC Berkeley, you can see factors of three or four difference,” he said. While maintaining comparatively even pay may foster a healthy work environment, it also may make Stanford less competitive than peer institutions.

Roberts refused to ascribe a singular cause to why faculty have steadily trickled out of the department in the last 10 years but stressed that the statistic alone should raise red flags about how Stanford is faring in the competition for top-notch faculty. He told me that although he cannot speak for others, overload — “too many students for too few faculty” — was the reason he went into “retirement.” He currently lectures at Reed College in Portland, Oregon, where he teaches an introductory CS course to approximately 50 students.

Lam concurred with Roberts. “We are so overworked,” she said. “We tried everything we could. I think that you will find our department to be extremely reasonable in terms of what we are doing. Everybody is working so hard to try to keep this up.”

John Mitchell, a CS professor and the newly appointed department chair, acknowledged that faculty are overworked but highlighted that most are overjoyed with the flood of student interest.

“Everybody who teaches in computer science really just feels deluged and is working extremely long hours trying to keep up with the student interest,” Mitchell said. “People suffer from lack of sleep, but they feel they’re doing something worthwhile by teaching students.”

Despite this effort from the faculty, Lam believes that students are still underserved and that the only solution is to dramatically increase faculty hiring.

Roberts underlined that Stanford’s CS department fares particularly poorly in the search for tenure candidates. He mentioned that “a lot of Turing Award winners used to be at Stanford but weren’t when they got the Turing Award… A few of those were people who had not gotten tenure at Stanford.”

“There is a sense in which we’re seeing harbingers of a decline,” Roberts said.

However, there are those within the department who do not see eye-to-eye with this assessment. When I asked Levis, who has served on the hiring committee, how the department has endeavored to match increasing student enrollment with new hires, he thwarted the assumption in my question.

“There’s this question of should faculty appointments scale with enrollments?” Levis claimed that from the University’s broader perspective, “If your majors increase four-fold, you don’t get a four-fold increase in faculty.”

Instead, Levis argued that decisions about how many faculty should be hired should come from “a deep intellectual and disciplinary standpoint” — an analysis of the ways in which a field of knowledge is evolving from a theoretical and practical vantage point, he said. He labeled the call for increased faculty based primarily upon enrollment numbers a “mythical man-month,” referencing a collection of essays from the nineties that highlighted the paradox that larger teams can sometimes be less effective than small ones in software project management.

Beyond differential faculty perspectives on the issue, national organizations have also taken stances in the debate over whether a method of gradual change is sufficient to address enrollment surges. In 2017, the National Academy of Sciences released a report titled, “Assessing and Responding to the Growth of Computer Science Undergraduate Enrollments” that issued a less equivocal verdict. According to the report, academic institutions should be “proactive” in response to increased CS enrollment.

“Taking incremental actions to get through the next year or semester are unlikely to produce the best outcomes for the institution and have in the past been associated with negative outcomes such as decreased participation of women undergraduates in computing,” the report reads.

In line with these so-called “incremental actions,” Jennifer Widom, the current dean of the School of Engineering, said that “there has been some net growth [in CS faculty hiring] but it is slow,” attributing the sluggish growth to “the extremely high standards of faculty hiring in the CS department.” 

“Over the past five years the overall size of the CS faculty has grown to 40 full-time-equivalent (FTE) tenure-line faculty to 48 FTE,” Widom wrote in a follow-up email to The Daily. “Several lecturers have also been hired.”

But this approach is consistent with a philosophy of slow-moving improvement on the status quo, which has done little to mitigate the chronic departmental bloat.

Despite the ostensible disparity in outlook between professors who side with the status quo and others who decry it as untenable, the report highlights common ground that is so plainly visible it almost feels redundant to point it out: Computing will play an “overwhelmingly important role … in the future of society and the University.”

In our conversations, faculty returned to the motif that they were passionate about teaching students and doing cutting-edge research. However, as Roberts alluded to, Stanford cannot sit back and assume that it is the only institution capable of providing this type of job satisfaction. The findings of the Visiting Committee report are as unambiguous as they come: Short of serious reorganization and reform, Stanford CS will fall down the ladder of top-ranking CS programs.

But concerns within the department may not alone be reason enough for enacting permanent, structural change — after all, Stanford maintains it is a liberal arts university. If Levis is right to advise against rash decisions on faculty expansion without due consideration of where CS sits in the University and how this position has transformed in the past decade, then this is precisely the conversation that we must now have.

College of Computing?

“I think we need to do something big,” Roberts said.

In October, MIT announced a $1 billion commitment to found the Stephen A. Schwarzman College of Computing, which will open in September 2019, to foster interdisciplinary opportunities and to accommodate escalating enrollments. Currently, each of MIT’s academic departments belongs to one of five schools. The College will operate at a distinct administrative level from the schools as an “institute-wide entity.”

Most notably, the College will add 50 new faculty positions — increasing, by fiat, the size of MIT’s total faculty by five percent — and will initiate the construction of a new building. MIT’s recent announcement revives an ongoing discussion in Stanford’s department over whether establishing a new administrative structure, like the College of Computing, might be the answer to its growing pains.

There is significant room for improvising the implementation of such a plan at Stanford. The central feature unifying various “College of Computing” plans is that CS would be elevated from the departmental level. But there is considerable latitude in executing this prototype at Stanford: This new administrative structure could entrust leadership of the “College” to a dean, putting CS on equal footing as Engineering, Humanities & Sciences and Medicine; or it could establish a “College” operating at a level separate from both departments and schools, similar to the College of Computing model unveiled at MIT. Most likely, these changes would be accompanied by numerous faculty appointments and the construction of a state-of-the-art building.

Roberts tells me that he has been pushing the department to consider some variation of a “College of Computing” proposal for the past 15 years. He claims that there is “overwhelming support” within the department for founding a “College of Computing,” recalling a straw poll at a faculty retreat around six years ago that showed that two thirds of faculty expressed support for the formation of a school.

Yet Roberts delivers this information with a tone of defeat, resigned to the reality that this kind of transformational change is a long shot. “Because a small group in the leadership is philosophically opposed to it — dead-set against it — it hasn’t happened,” he explains.

“I think there’s been a failure in the competence of the leadership of the department,” Roberts continues, in reference to faculty retention and strategic planning.

In 2015, Jennifer Widom, the outgoing CS department chair, was succeeded by her husband, CS professor Alex Aiken, when she was promoted to the post of Dean of the School of Engineering. Aiken served as department chair from 2015 to 2018.

“They’re both opposed to [a College of Computing], and that’s enough,” Roberts says. “I’m 100 percent convinced it’s the right thing to do, and I’m 100 percent convinced that Stanford will never do it.”

When I reached out to Widom for comment, she replied, “MIT’s ‘[College]’ is very different from what was being discussed here within CS but not so different from some [long-range planning] initiatives.”

Although Roberts claims Widom and Aiken both opposed the “College of Computing” idea during the straw poll from several years ago, Widom denied involvement in “any group discussions about a [College] of CS at Stanford.”

“I haven’t been to a CS faculty retreat in many years, and I’ve not been involved in any group discussions about a School of CS at Stanford,” Widom wrote to Roberts in a heated back-and-forth email chain, which included me. “You must be mixing me up with someone else!”

Roberts aptly refuted these claims, stating that although he had never discussed the idea of a Stanford School of CS with Widom individually, Widom has “been part of the discussion at several faculty retreats over the years.”

“It would be wonderful for Stanford if your position has changed on this issue, because Stanford is, unfortunately, standing remarkably still while other universities are zooming ahead,” he continued, adding his belief that MIT’s proposed College of Computing “dwarfs anything at Stanford.”

Aiken did not respond to The Daily’s request for comment on this matter.

“The Schwarzman College of Computing: if you talk to anyone at MIT, they’ll tell you they don’t know what it is yet,” Widom told me when I asked her whether Stanford intended to make any similar moves. “We’re taking the approach where we are going to figure out what our structure is going to be, and then we’ll go ahead and announce. The scale of what we will announce is probably relatively similar to the Schwarzman in the long run.”

But when I followed up to ask if she meant that Stanford intended to follow MIT’s lead, she curtly replied, “No, I didn’t say that.”

In particular, Widom directed me to research Human-Centered Artificial Intelligence (HAI), which was soft-launched this past October by Fei-Fei Li and John Etchemendy. Its announcement brims with optimism about the prospect of convening the finest AI researchers in one place, and its website claims an impressive interdisciplinary army of 66 academics as its “team.”

In an email to The Daily, Li wrote that there would be both “top down” undergraduate involvement in the HAI initiative — classes, research and “intellectual events led by faculty” — as well as “bottom up” student-initiated activities. Etchemendy added that “HAI is very different from other efforts, both in the breadth of its focus (including both technical and humanistic concerns), and in the scale of its ambitions.”

HAI, which was also given a nod by President Marc Tessier-Lavigne in his address to the Academic Council on the long-range planning process, represents an effort to keep Stanford relevant in a fast-changing field. But the HAI announcement at Stanford makes vague and sweeping statements about “guiding the future of [artificial intelligence]” with no actionable claims, in stark contrast with the bold and straightforward “College of Computing” announcement made at MIT. Few details are publicly available about how much funding has been allocated for this initiative and what kinds of projects will be supported by it. And at least overtly, it does not appear to address the capacity crisis in the department. The speculative, optimistic take is that Widom’s remarks and the HAI announcement are nebulous, for the time being, in the run-up to the disclosure of more concrete plans. Or they may simply represent attempts to defer on providing a detailed blueprint for the future.

Roberts argued that in particular, Stanford is falling behind in the zero-sum game of faculty hiring. “All these places that are building schools … are going to be extraordinarily attractive to the people who are already at institutions like Stanford,” Roberts said.

He was clearly frustrated that others in the department do not take the threat that institutions like MIT pose to Stanford’s CS department seriously. In describing the attitude of some within the department, he said, “It’s this hubris that won’t quit.”

He imitated an exaggerated version of their posture — “It’s Stanford! Everybody comes to Stanford! Nobody would turn Stanford down!” — of course, despite the fact that faculty are leaving at record rates.

By this point in our conversation, I was convinced that the leadership in Stanford CS had turned a blind eye to the capacity crisis that has gradually reshaped the student experience and that has become painfully acute for faculty. But a new administrative structure should not merely be a half-baked instrument to temporarily address the current resource strain. From a disciplinary point of view, it should reflect a philosophical stance on how knowledge corresponds to the contemporary world.

Levis, for one, offered a more cautionary attitude towards announcing a College of Computing. What is best for one department may not be best for the University at large, he warned.

“I think you want to look at — when we look at ourselves as a University — would it make sense, given where this field sits, for it to be broken into a separate school, sitting at the table of the [executive committee] of the University, with the deans?” Levis said.

Levis highlighted that schools like MIT and Carnegie Mellon University are altogether “different beasts” as historically technology-focused institutions and that the models they offer therefore may not be appropriate for a place like Stanford, which also emphasizes the liberal arts. In response to whether Stanford should establish a College of Computing, he paused to carefully word his answer and said, “There could be a time when that is a good idea, and that time might not even be that far off, but I don’t think that is something suitable today.”

When I asked Levis what conditions might make a College of Computing the right direction for Stanford to move in, he responded, “If the computer science department were sufficiently larger.”

Earlier in our conversation, Levis said that increases in faculty should be predicated on an evaluation of the relationship of the field of computing with the rest of the University. If anything, all signs indicate that the time has come for the CS department to seriously reevaluate its position in the University.

Although the capacity crisis may by itself be an unsatisfactory justification for considering the College of Computing model at Stanford, it is also increasingly clear that both skyrocketing student interest and faculty brain drain are symptomatic of broader challenges Stanford faces in shaping the evolving relationship between CS and the rest of the University and industry.

Last year, in the spring issue of this magazine, Opinions writer Anna Sofia-Lesiv ’20 registered her shock at the fact that only two people in CS 181: “Computers, Ethics, and Public Policy,” a class of 150, raised their hands when asked by professor Keith Weinstein, “How many of you think you will write software that will affect people’s safety?”

But such questions are, for all intents and purposes, a shout into the void when posed in a class of 150 (or this winter, what will be a class of almost 300). “Liberal arts education” describes not a subset of academic disciplines but rather a process — an understanding that is missing among those who are smugly content with a model of undergraduate education that does not provide a forum for students to meaningfully consider intent, application and hybridity in their chosen discipline.

Several people I spoke with expressed sympathy with anxiety from administrators and those outside the CS department that establishing a “College” or “School” would further accelerate current enrollment trends, sucking the life out of every other discipline in the University. But such a view subtly fortifies the ever-heightening wall protecting a center of “liberal arts” intellectualism and thought from encroaching technologism, as if these two cannot be fused in productive ways. Students have, for years, imposed, joked about and struggled with the “fuzzy-techie” divide. Now it is the administration’s turn to grapple with how to respond to the growth of the CS department in a way that breaks, traverses and transcends the belabored dichotomy.

Commonly held beliefs among students, faculty and administration alike that CS education is inherently less demanding on individual attention reinforces the loathed opposition between STEM and the liberal arts — that the former consists of rote problem-solving, while the latter asks profound questions of the world. I don’t claim to know what an alternative to the current divide might look like. But universities that are engaged daily in the classification, categorization and structuring of human knowledge have the right tools at their disposal to ignite these conversations.

The historically unprecedented loss of faculty — at a time when the field is gaining import with each passing day — only exacerbates existing pressure on limited resources in the department. It presents two possibilities: It can either trigger needed internal introspection, or reflection can altogether be sidestepped, making long-term strategic planning about the CS department’s coordinates and course increasingly impossible.

When I interviewed John Mitchell, he was just a week shy of stepping into his new role as department chair. He underscored that the department is a “collaborative, consensus-based organization,” as if to advise me that his provisional opinions are not royal decree.

I asked him for his thoughts on MIT’s College of Computing.

He sung its praises. “I think the MIT structure is really very creative,” Mitchell said, “and it’s visionary, it’s forward-looking, and the way they’ve thought about this is a tremendous example for everyone.”

Mitchell lauded MIT for acknowledging that computing is increasingly embedded in all fields. He continued, “What we should do here is acknowledge the same global trends and academic and intellectual trends, and then try to figure out how to respond to that in a way that reflects the nature of Stanford.” He ended his response on the remark that “we would be poor academics if we didn’t ask deep, hard questions about what’s going on in the world.”


Contact Jasmine Liu at jliu98 ‘at’ stanford.edu.

Jasmine Liu is a senior staff writer and writes for Opinions and Arts & Life.

Login or create an account