Harvard Institute of Technology
Disparities between the computer science communities at Harvard and MIT are probably due in large part to the way each was programmed. About 100 years ago, Harvard was programmed, more or less, to neglect Engineering in general, and Computer Science in particular.Brian “Foundation” Shimanuki wins another game. After shaking hands with his opponent, Foundation walks back to his seat with a grin. An enthusiastic crowd cheers on their clear favorite.
It is 9 p.m. in Cambridge. It the last Saturday night in January, and this is the Battlecode finals.
In the game of Battlecode, a red team and a blue team face off. Each team aims to kill the other’s archons—special units that can create other units—while protecting its own archons from attack. Teams employ various basic strategies (defend and gather the resources scattered around the map, attack early and often in the hopes of a quick victory, etc.) and endless minor permutations of those strategies. It’s a complicated, slightlier nerdier sort of chess.
Unlike chess, though, Battlecode, founded in 2001 and run by MIT students, is not actually played by humans. At least, not directly. Instead, two computer programs control every decision made during gameplay.
Teams use the month of prep to strategize, but they also spend it translating and coding that strategy into Java. It’s not the two teams that are competing: It’s their respective programs. By the finals, results are essentially decided. If there’s a problem in your strategy now, it’s too late to change it.
The big twist this year is zombies. Zombies attack any unit, red or blue, within their line of sight, and they get stronger as the game progresses. Foundation’s whole strategy centers around zombies. Instead of creating his own army, he only creates scouts, units that are very fast, but unable to attack. The scouts find zombies and lead them to the enemy.
Foundation ultimately places fourth, but the crowd adores him, and his competitors fear him. On Foundation’s “team wall” on the Battlecode website, an adversary and admirer writes, “You twisted, cruel bastard. 10/10 strategy.”
Though MIT students organize the event, competitors from all over the world can submit code. The final tournament, an hours-long, 16-team affair, features teams from Germany to nearby Harvard.
The Harvard team, “what_thesis,” has far fewer opportunities for such CS-centered gaming on their own campus.
Indeed, when it comes to computer science in Cambridge, there’s little doubt that MIT has the more established culture and curriculum. It boasts more students, more professors, more interest from recruiters, and much higher rankings. Differences between the two schools don’t stop there. Where Harvard’s campus is picturesque, MIT’s is stark. Where Harvard students have “concentrations,” MIT undergrads major in a number—6 for Computer Science, 2 for Mechanical Engineering.
Much like in Battlecode, though, disparities between the computer science communities at Harvard and MIT are probably due in large part to the way each was programmed. About 100 years ago, Harvard was programmed, more or less, to neglect Engineering in general, and Computer Science in particular. To understand that Harvard’s computer science department seems comparatively underdeveloped, look to the man who coded Harvard: Harvard’s 21st president, Charles William Eliot.
The New Education
Maria Stoica ’17 checks her email in the Eliot House dining hall. Though she seems at ease, Stoica is not the sort of student Eliot, Harvard Class of 1853, would have imagined at Harvard. While he certainly opposed allowing women at Harvard, Eliot’s real problem with Stoica would have been her studies. Stoica is a computer scientist, an engineer. Harvard is a liberal arts college. In Eliot’s mind, those things shouldn’t mix.
“The practical spirit and the literary or scholastic spirit are both good, but they are incompatible,” wrote Eliot in an 1869 piece in The Atlantic dubbed “The New Education.” “If commingled, they are spoiled.”
This article, many believe, landed Eliot the Harvard presidency. He penned it as a chemistry professor at the Massachusetts Institute of Technology in Feb. 1869. Months later Eliot assumed the Harvard presidency. Stoica also made the move from MIT to Harvard, transferring between her sophomore and junior years to find a “well-rounded ” culture that she felt MIT lacked. She wanted the kind of “co-mingling” of disciplines that Eliot so excoriated in 1869.
Over the course of his presidency, Eliot tried many times to absorb MIT into the Harvard umbrella, coming closest to success on his last attempt in 1905 before MIT alumni and faculty successfully rallied against the merger.
Perhaps frustrated by his last failure, Eliot absorbed Harvard’s formerly independent Lawrence Scientific School into the College in 1906. The move marked a complete reversal from Eliot’s position in “The New Education.” In that essay, he advised against putting “scientific schools” in the same organizational wing as the more established fields (the classics, philosophy, etc.) A scientific school within a college, Eliot wrote in 1869, “is the story of the ugly duckling.”
The younger Eliot was right. Harry R. Lewis ’68, professor of Computer Science and former interim Dean of Harvard School of Engineering and Applied Sciences, blames the Lawrence Scientific School-College merger for “effectively put[ting] engineering and applied science in a backwater” for close to a century. With some notable exceptions, Harvard engineering remained an ugly duckling.
In the mid-70’s when MIT (1974), Stanford (1965), and Berkeley (1968) already had Computer Science departments, Harvard remained opposed. In around 1978, Lewis, then a non-tenured assistant professor, proposed a computer science degree at an Engineering faculty meeting; computer science was part of the Applied Mathematics department at that time. According to Lewis’s blog “Bits and Pieces,” Bernard Budiansky, an applied mathematician, opposed the idea, asking, “We’ve never had a major in automotive science, why would we have one in computer science?”
Budiansky’s gripe seems straight out of Eliot’s writings. Eliot was fine with teaching most subjects, but “they should be taught at the university on a higher plane than elsewhere.” Computer science, in Budiansky’s mind, was a practical and technical skill. It wasn’t until 1984 that Harvard awarded its first undergraduate degree in Computer Science, and by that time other programs had almost a decade’s head start. By and large, the early adopters (the MITs, the Stanfords, the Berkeleys) are the top programs today.
In colleges across the country, the debate continues over what computer science actually is. A mélange of poorly defined terms (coding, hacking, computer programming, computer engineering, etc.) are routinely used to describe the field and a bevy of online services (Codeacademy, Code.org, Launch Academy, etc.) advertise a quicker, cheaper route to a job than a four-year degree in computer science.
A Wall Street Journal op-ed published last August titled “Why I’m Not Looking to Hire Computer Science Majors” blamed computer science departments for not adequately preparing their students for the industry. The op-ed criticized the lack of iOS and Android development classes in Ivy League universities like Yale and Harvard. This is a common enough complaint at Harvard, and even at MIT, where students believe the department favors the theoretical over the applied. For Tomas A. Reimers ’17, though, studying computer science is like studying Spanish literary theory: if you’re going to learn it at all, you have to learn it in Spanish.
“There’s… this, ‘You need to learn Spanish’ portion, and that’s what programming is,” says Reimers. “Then there’s proper CS, which is how you think about data structures and algorithms and things like machine learning and visualization.”
Regardless of how you define it, computer science has become immensely popular. Harvard’s Computer Science department climbed from 69 concentrators in the 2007-2008 school year to 348 this year, with an inflection point coming, a chuckling Lewis points out, right around the release of the Facebook-inspired “Social Network.”
A $50 million dollar gift to Harvard Computer Science from former Microsoft CEO Steve A. Ballmer ’77 (with the Ballmer-esque edict to put Harvard “on that list” of top computer science schools alongside MIT, Stanford, and Carnegie Mellon) and a record-breaking, school-renaming $400 million from Business School alumnus and hedge fund manager John A. Paulson have given Harvard the capacity to grow computer science like never before. Throw in an historic $1 billion move to Allston, which Lewis has called “the Promised Land,” and the CS department is looking at a future of sustained growth into uncharted territory.
Harvard’s neighbor down Mass. Ave., though, has been running massive computer science operations for a while. Harvard CS won’t reach the size of MIT’s department any time soon.
Its growth, however, has made the University question the way it has operated since Eliot’s tenure. What is the ideal balance between extracurriculars and scholastic pursuit? What is the proper amount of corporate involvement in a university education? What is the best way to introduce students to a subject? MIT has (at least in part) answered many of these questions, and Harvard must, too, as enrollment and investment in computer science reach a fever pitch.
Here for the Classes?
At MIT, students mostly engage with the computer sciences through their academics. Harvard CS, on the other hand, is as much about what happens outside the classroom as it is about what happens in it, with Harvard students choosing to focus more on the extracurricular side of their CS studies.
None of the real rockstars of Harvard CS, after all, ever graduated from the school. Both Bill Gates and Mark Zuckerberg quit early to commit to their respective start-ups full time.
Even those who don’t go so far as to drop out, though, look for ways besides academia to pursue computer science. Between working as a teaching fellow, joining an organization like the Harvard Computer Society, or working for fun on side-projects or start-ups, Reimers says computer science “allows you to focus all of your energy on it.”
“If you want to dump your life into computer science, [the subject] very much encourages that,” says Reimers. “In fact, it rewards it.”
But then, Harvard students are notoriously committed to life outside of the classroom. The proportion of students who self-reported spending more than 10 hours a week on extracurriculars rose from 27 to 43 percent between 2010 and 2013, according to data from the Senior Survey.
According to Divya Shanmugam, an MIT junior in Computer Science, life is the classroom. “I would say that classes take up a majority of your time, and then if you manage everything well you can do extracurriculars,” says Shanmugam. “I’m starting to quit my extracurriculars and say, ‘I am here just for the classes.’”
Shanmugam was drawn to MIT as a high school senior in part because of the “starkness” of the campus and the “no frills education” that it seemed to promise. This ascetic ideal of pure academics is foreign to MIT Computer Science professor David R. Karger ’89, who took music classes and sang in a choir during his undergraduate years at Harvard.
“[Some] people take nine courses in a semester and ace them all ... It’s very hard in that environment to hold back and say, ‘I want to have time,’” says Karger. “I can’t imagine having time to do that with the kind of course loads that MIT students face.”
MIT also boasts a conventionally stronger department, mostly due to its size. When it comes to computer science departments, size matters. Harvard CS, according to new Harvard hire and former MIT professor Madhu Sudan, has always had a reputation for strong individual professors (including two winners of the Turing Award—a sort of Nobel Prize for computing) but a weak overall department. It lacked “critical mass.”
Since 2007, Harvard CS faculty has grown from 19 professors to 32. Harvard has 12 electrical engineering professors. MIT’s Electrical Engineering & Computer Science department, meanwhile, has 165 professors. There are more MIT EECS professors in the prestigious National Academy of Engineering than there are total Harvard CS professors. Unsurprisingly, MIT also offers many more computer science courses.
While Stoica, who has taken courses at both schools, says the workloads are comparable, Roger Zou ’17, former president of the Harvard Computer Society, says he got the sense from summer colleagues at Google that MIT was “more rigorous for sure.”
“MIT has a lot more courses that are on the same level of rigor as the Operating Systems Course [CS161] as well as, of course, a greater selection of courses,” says Zou. “They’re more famous, they’re better taught. But that’s just because, you know, [they’re] MIT.”
If MIT CS classes are hard, Harvard’s are no walk in the park, either. Zou himself determined as much when he “scraped” the Q Guide and wrote up the results in a Medium post. According to Zou’s analysis (which includes a number of caveats, including some duplicates in his dataset that could potentially skew results), CS classes have the heaviest workload at 10.7 hours per week, about five hours more than the average workload.
Hana C. Kim ’16 says that the Q guide, whose current highest rating option for workload is “greater than 14” hours per week, should include more demarcations over 14 hours.
“I’m on the longer side of the spectrum, but some of those classes will take 20, 24, 30 hours [a week]. So you want to know the gradient up there too, you know?” says Kim. “I expect all my CS classes to take way greater than 10 hours.”
Some courses, like CS161, are notoriously grueling. On the first day of class, Professor James Mickens, taking a leaf out of the U.S. Marine Corps’ book, flashed a powerpoint slide that read, “Pain is weakness leaving the body.”
At MIT, 6.172 (another CS course) is similarly fear-inducing. A “performance engineering” class, 6.172 teaches students how to take basic programs and make them as fast as possible.
“[6.172] can end up being a 40-hour-a-week class because they don’t really give you a benchmark for how fast [you can] make this,” says Shanmugam. “It’s just like, ‘Go forth, make it fast.’”
Karger says he worries that the workload of computer science, especially at MIT, is too demanding.
“[When I was an undergraduate], I didn’t feel like I was in a permanent state of exhaustion from trying to manage this massive workload. I do feel like that’s the state at MIT,” Karger says. “I’m pretty sure that MIT students come out having studied a lot more stuff than Harvard students. It’s not clear to me that that’s at all important to their futures.”
According to Shanmugam, while it’s relatively common for Harvard students to take CS classes at MIT, MIT students generally cross-register with Harvard only for humanities or language classes, like the Tamil class Shanmugan takes at Harvard.
A Conversion Experience
If there’s one Harvard course MIT students know, though, it’s CS50.
Zou came to Harvard intending to concentrate in Government, but freshman fall he “took CS50, and everything changed.”
He spent his freshman summer programming his own iOS app, his sophomore summer working at Youtube, and plans to work for Square this coming summer. His aluminum Square water bottle and laptop, which bears two Youtube decals, sit on the table in front of him.
“I took CS50, and everything changed,” is a common enough line for Harvard CS concentrators, but at MIT, the norm, anecdotally at least, is to have some background knowledge of computer science before college.
CS50, Lewis says, “creates a lot of converts.”
“You just have to give the credit to Professor Malan here, but Professor Malan is himself a convert to computer science,” says Lewis. “He was a Gov major until he took CS50 and it changed his life, and he believes that it will change your life, so he’s got the evangelical spirit.”
Though CS50 is a common enough starting point for CS majors, it has little in common with the rest of the department. Professor Greg Morrissett, who from 2009 through 2015 taught CS51, the course after CS50, used to begin his lectures with a short, clipped parody of Malan’s signature line: “This is not CS50.” The opener was always met with enthusiastic applause.
According to Morrissett, the remark was meant to poke some fun at the branding efforts of the “CS50 machine,” but also to let students know that the course would have less support (fewer teaching fellows and video walkthroughs) and be more challenging than Malan’s well-funded, meticulously choreographed operations.
Morrissett, among others, feels that CS50 is not particularly representative of the other CS courses in the department, though he says the difference benefits students.
“The faculty and school decided that it was important to invest heavily into CS50 in order to open up the exciting world of computing to as many Harvard students as we possibly could,” Morrissett, who left Harvard for Cornell at the end of last year, wrote in an email. “So in this sense, no, CS50 doesn’t give students a good feel for what will happen with subsequent courses—they’re not going to get David Malan and they’re not going to get this amazing support staff and organization.”
CS50 is unique in other ways as well. In addition to a sizeable Harvard undergraduate enrollment, CS50 was last semester’s most-enrolled course at Yale and also boasts an extensive online audience via the online education non-profit edX. CS50’s intellectual property was also the subject of an apparent tussle between Malan and Harvard for the course’s trademark.
All this programming doesn’t come cheap. CS50 is the only Harvard course with corporate sponsors, a practice that has drawn criticism from Harvard undergraduates and MIT professors alike. Karger says CS50’s sponsorships make him “queasy.”
“The course [should not be] a job fair. I really get that sense from CS50. It’s funny, you can see the energy, the enthusiasm that CS50 creates. It’s not clear that we need to have energy and enthusiasm created,” says Karger. “There’re all these people pouring into computer science; I would almost rather they get an introductory course that really tells them what computer science is.”
Lewis, however, sees no problem with corporate sponsorship.
“I think it’s great any time anybody wants to make an investment in our program,” says Lewis. “So I don’t think it’s a problem as long as they’re not setting an agenda for what we’re actually teaching.”
Lewis sees corporate funding as a necessary means to expanding access to education.“Bernie Sanders says education should be free to everyone, all the time,” he says. “Great, but the way the world works, we actually need money to keep our programs running.”
MIT CS Professor Srini Devadas, the current chair of the EECS Curriculum Committee, is similarly uncomfortable with corporate sponsorship of an undergraduate class.
“I fundamentally dislike that idea. I hate the idea of mixing sponsorship into undergraduate classes. I mean, I’m a little surprised honestly,” says Devadas. “If someone tried to do that here, I mean, I wouldn’t get violent, but I’m getting a little worked up here. I don’t think we would do that.”
Malan declined multiple times to respond to these criticisms in person or on the phone, later sending an email statement which read “CS50 is indeed a large and, in turn, expensive operation, and we gladly accept support from industry and alumni to provide students with all the more resources and opportunities.”
Devadas acknowledges that a number of MIT wintersession activities as well as upper-level research classes (known as SuperUROPs) do have corporate sponsorships (though MIT’s Dean of Undergraduate Education Dennis Freeman says that the university “is very careful to distance the companies from the students”). While Devadas says that he teaches MIT’s introductory programming class in Python in part because it’s “a language that could get people a summer job,” he still has reservations about CS50’s sponsorships.
“You have to be very careful when you do these things,” Devadas says. “Let’s not send the wrong message at the freshman level. Why don’t we just teach them how to program?”
Corporate Culture
Battlecode—which, with a course number of 6.147, counts as a half-course for MIT undergraduates—is one of the sponsored Wintersession activities Devadas mentions.
Indeed, after each round of the Battlecode final tournament, a different corporate sponsor makes their way up to the Kresge stage. Oracle (“Integrated Cloud Services”) touts its stellar work-life balance. Cruise (“Join the Driverless Revolution”) paints itself as a great modernizer, pioneering the self-driving car (somehow, though, the AV hook-up to their promotional video is bungled; the Battlecode directors gamely take the blame). Bloomberg’s presenter starts off with an enthusiastic, though tepidly-received, “Good evening Battlecode-icans!” Their pitches, though, all more or less boil down to the same message: Working here is a lot like Battlecode. We solve cool problems. We’re hiring.
While only a select few company representatives get to speak during the event, all sponsors are allowed tables at a mini-career fair of sorts in the foyer an hour before doors open. Students, munching on Battlecode-supplied pizza, engage in small talk with the recruiters.
A recruiter from Facebook tells a student that “WhatsApp has been notoriously efficient.” Recruiters ask how people feel about staying local versus moving west, and every table has some sort of goodie to give out. T-shirts, water bottles, key-chains, umbrellas all branded with the appropriate company logo are handed out like pamphlets at a political rally. Uber shows up without merchandise, just a few boxes of cookies to lure students.
While all sponsors are eager to talk to MIT computer scientists, some are less eager to talk to press. Apple declines to chat for reasons of “confidentiality,” and DE Shaw is similarly tight-lipped. A representative from Oracle, one of Battlecode’s platinum sponsors, though, talks about the company’s “huge affinity for MIT and the MIT community” and mentions that all the Course 6 students at Battlecode are a draw. PDT Partners also cites the huge concentration of computer science majors at MIT as a reason to recruit there.
Most sponsors stick around until Monday to go to the XFair, a tech-focused career fair held in MIT’s gym which drew 111 tech sponsors this year. Xfair is one of two popular student-run career fairs on campus. The other, the MIT Fall Career Fair, is by far the larger of the two.
The fall fair routinely attracts over 300 companies, most of whom are desperate for software engineers. So large is the demand for Course 6 majors that some students have taken to calling the fair “Course 6 Appreciation Day.”
Freeman, MIT’s dean of undergraduate education, worries that incoming freshmen might be driven to study computer science largely because “they see all the job offers [at the fall fair] are for the people in CS.”
Sophie Goemans, a junior majoring in 7 (Biology), says that every MIT student, “at some point or another, claims that they’re going to do CS at MIT.”
“You come here and you look at the CS’s and you’re like, ‘Oh my god, these people are literally going to be Forbes Fortune 500. They’re gonna be in that crowd,’”Goemans says.
While Karger says the financial security of a career in tech is certainly a draw, the opportunity to work in an exciting field also attracts students.
“[It’s] not just money, but [also] a good career,” he says. “That doesn’t only mean money, it means doing fun stuff, having an impact on the world. Computing is a great opportunity to do all of that. You don’t have to be a starving artist. You don’t have to go work in boring banking. It’s really exciting and really lucrative at the same time.”
While tech recruiting at Harvard is not nearly as developed as it is at MIT, SEAS and the Office of Career Services are working to close the gap. Keith Karasek, the SEAS Director of Experiential and Career Development, works to facilitate recruiting for SEAS majors in a school that he says has traditionally focused more on finance and consulting recruiting.
The OCS’s two main CS-related career fairs are the Big Data and Analytics Fair in the fall (with 77 participating companies in 2015) and the Start-up Career Fair second semester. While both fairs are much more limited in scope than MIT’s, SEAS can also send a select number of its students to participate in MIT’s on-campus interview program, which is more tech-heavy than Harvard’s.
Much of the recruiting and negotiation advice at Harvard exists outside the classroom. Stoica found her tech internships at Groupon last summer and Goldman Sachs this summer through a Grace Hopper Conference, a series of conferences for women in computing. Email lists, both formal—like Women in Computer Science and the Harvard Computer Society’s—and informal, are also tools to find jobs or internships and negotiate pay offers.
Tech talks, also frequently publicized over email lists, are another avenue for students and recruiters to find one another. While CS51 does not have a formal sponsorship arrangement with quantitative trading firm Jane Street (which works in OCaml, the obscure programming language taught in CS51), representatives of the company come to Cambridge every year to deliver a lecture and host a tech talk, distributing pizza and free t-shirts on the trip. Other companies, like Facebook and Dropbox, have delivered tech talks on campus too.
The tech industry’s penchant for free, branded merchandise has permeated MIT as well, as one quick walk around campus confirms.
“If you see someone wearing an MIT t-shirt, chances are they’re a tourist, and if they’re wearing a Dropbox t-shirt, chances are they’re an MIT student,” says Goemans.
Engineering Allston
“Open. Connected. Active. Transparent. Livable. Sunlit. Social. Flexible,” read the first words of a Jan. 2013 document detailing ideas for SEAS’s new Allston campus. “These words keep coming up when SEAS faculty talk about the exciting possibilities of a new Harvard campus in Allston.”
The same words also do a decent job of describing Course 6’s home at MIT, the Stata Center. At lunchtime on weekdays, it’s rare to find an empty table in the building’s lobby, and lines extend out of the café and coffee shop. The hum of conversation and the click-clacking of keyboards bounce off the high, irregularly shaped ceiling.
In contrast, Maxwell-Dworkin, Harvard CS’s home, is decidedly cramped and unsociable. During shopping week, it was not uncommon for classes to spill out of their assigned classrooms, ultimately abandoning them altogether in favor of empty common spaces.
“We’ve set up a few tables in the ground floor at Maxwell-Dworkin. It’s not great, not very inviting. No food. No coffee,” says Lewis. “It’s not as homey as I wish it would be and I think it’ll feel homey over [in Allston], and that will be good.”
While MIT’s community spaces may be “homier,” both schools are extremely pressed for office space. They face similar problems: While student interest is growing at a steady clip, faculty hires are limited by both availability and physical space. who’s available and whether there’s physically room for them.
Devadas says that Course 6 is “in a serious space crunch,” a problem which started shortly after it moved into the Stata Center in 2004. Before Course 6 can make new hires, it has to submit a “space plan” to the Dean of the School of Engineering, proving that the new hires will have a place to work.
While there is certainly an end in sight for Harvard’s space concerns, not everyone shares Lewis’ optimism about the Allston campus.
“I think it will be terrible,” says Reimers, a resident of Pforzheimer House. “I’m curious to see how Harvard does it. I have not been impressed with how they’ve done distance before... To me, it seems like a high-level bureaucratic decision being passed down with some regard, but not as much as should be given, to how it will affect the students.”
Reimers worries that shuttle schedules will not be able to accommodate the irregular hours of college students. “You’ll probably find CS students in classrooms at like 4 a.m. And if you don’t have a shuttle running, then that’s a problem,” he says.
While much remains uncertain about the Allston move, some things are for sure: There will be an engineering campus in Allston, and the computer science department will move there.
In a way, this isn’t too different from Eliot’s plan. When he was attempting to absorb MIT in 1905, his plan was to move it across the river to Allston, on roughly the same land that Harvard Athletics occupies now and that the SEAS campus will occupy in 2020. For Eliot, the Allston campus was a chance for Harvard to separate the engineers from the scholars. He promoted the very “socially isolating” split that Reimers fears.
But this time around, the goal isn’t a rigidly divided Harvard Institute of Technology. Rather, as Harvard’s current president wrote in a letter to the editor of The New York Times, the College’s professed aim is to mould “scientifically sophisticated humanists and humanistically grounded scientists and engineers.”
Graduate School of Education Professor Howard E. Gardner ’65 agrees, though he’s still wary of “a slippery slope toward becoming, in effect, a school of professional or pre-professional training.”
In an email, Gardner described a national study of higher education he’s conducting, writing “A student at that school said ‘I am getting a liberal arts education AND an engineering degree.’ I hope that is what Harvard undergraduates will aspire to.”
