6

New Flows Through PhDs in Computing

This chapter surveys the expanded—and expanding—landscape of paths into, and out of, advanced computing degrees. Rather than considering pathways to doctoral degrees in computing as a linear sequence, it is important to understand the collection of paths that change and adapt to the unique experiences of students in computing.

Although most doctoral students in computer science earned their undergraduate degree in a computing field, one-quarter of incoming doctoral students earned their undergraduate degree in a non-computing field such as arts and humanities (Wright 2022). Especially as computing becomes increasingly integrated throughout all sectors and disciplines, the variety of paths to and through doctoral degrees in computing will continue to grow and change, much like a braided river, to better appreciate unusual entry points, evolving occupational goals, and opportunities for lifelong continuing education.

TRADITIONAL PATHS INTO COMPUTER SCIENCE PHD PROGRAMS

Inflow: Bachelor’s Degree in Computer Science, Electrical Engineering, or Mathematics

Applicants traditionally enter computer science PhD programs with a bachelor’s degree in computer science (CS), computer engineering, or mathematics. Most doctoral programs in computer science expect applicants to have completed prior coursework in foundational topics in computer science or expect them to take courses after admission to fill any gaps in their preparation. While some institutions specifically list course

expectations, the applicants’ transcript will generally be reviewed for typical foundational courses such as algorithms and data structures, computer organization and architecture, operating systems, theory of computation, linear algebra and discrete mathematics, programming languages and software engineering. Depending on the courses included in the curriculum of these different majors, students may need to take additional courses prior to applying for a PhD program to be competitive for admission or take these courses after being accepted.

For example, students with a background in mathematics may need to take systems courses, and computer engineering students may need to take CS theory courses. It is even possible that some students with a CS major may need additional math courses to be prepared for a PhD program. However, the overlap in preparation for undergraduate students who have completed degrees in these areas is significant enough that students from these disciplines have successfully been admitted to and have completed doctoral degrees in computing with little effort to bridge any gaps in understanding. Challenges and support needed to broaden the participation of students from these majors—by bringing awareness, providing opportunities for research experience, and providing support for putting together a competitive application—are covered in Chapter 3.

Inflow: Master’s Degree Earned Along the Way

Another traditional path into doctoral degrees in computing is the completion of a master’s degree in CS or related fields, such as computer engineering or mathematics. In the late 20th century, advancement to a doctoral degree in CS followed a well-established linear sequence:

• Bachelor’s degree in CS (or related field): Students receive preliminary training in fundamental CS topics, primarily acquired through coursework.
• Master’s degree in CS: The master’s degree was previously viewed as a required stepping stone between undergraduate and doctoral degrees. Students applied for and were granted admission to a master’s program separate from the PhD. To acquire a master’s degree, students completed required coursework and advanced training in graduate-level courses. Unlike the foundational courses completed in undergraduate training, graduate courses encompassed a greater breadth and depth of material, including technical specialization in some cases. In addition to coursework, students were required to complete closely monitored research projects, typically culminating in a master’s thesis, that required a written report and oral defense. It was understood that students who received course-only training in their undergraduate education were not properly prepared for a research-dominated

• doctoral degree. Completion of a master’s degree could, therefore, be used by PhD programs to filter for students who have not done independent research.
• Doctoral degree in CS: Upon completion of a master’s degree and demonstrated competence with research, students could then apply for a PhD in computing. Due to the prior vetting experience, the PhD application and admission process were relatively straightforward.

This seemingly linear sequence began to evolve in the 1990s for two key reasons. First, the notion of direct admission to doctoral programs became the more dominant path for students with CS training. By removing the requirement for a master’s degree, the application to PhD programs became more accessible and more affordable for students interested in research. With the removal of this stepping stone, there was a shift to focus on gaining research experience during completion of a bachelor’s degree. Second, the master’s degree in various subdisciplines of computer science became a widely desired terminal degree and served as a path into gainful employment in computing. Rather than focusing on research as a step toward future doctoral work, the professional master’s degree evolved into a skills acquisition degree in areas including data science, artificial intelligence (AI), robotics, and cybersecurity. The master of science degree curriculum requirements changed to be course-only and, in some cases, students pursuing these degrees could not easily transition into subsequent PhD programs without reapplying. The professional master’s degree and the doctoral degree, therefore, became parallel paths to terminal degrees with different intended skills for students with different career aspirations.

NEW INFLOWS

In addition to bachelor’s degrees in CS, there has been a recent rise in a set of computing-related training opportunities that either do not yield a traditional CS–labeled degree or do so with a non-traditional structure. This section is divided into two parts to evaluate direct paths—students with training in at least a subset of traditional, core CS topics–—and indirect paths—students with different science, technology, engineering, and mathematics (STEM) skills that can feed later into advanced training in computing.

Direct Paths: CS+X and Related Programs

In early 2004, the concept and label of “CS+X” were introduced by Alfred Spector in a lecture discussing the growing importance of computer science research and collaboration across all disciplines. Spector argued that “the impact of information based technologies will continue to grow—probably at an accelerating rate. In nearly every segment

of society, there are both quality and production improvements because of increased use of automation and digital communication” (Spector 2012). His work inspired a set of academic experiments focused on CS+X degree programs for a range of partnered X departments at several different universities. Three notable examples that will be discussed in detail include Stanford University, the University of Illinois Urbana-Champaign (UIUC), and Northeastern University.

At Stanford University, the CS+X Joint Major Program (JMP) was approved by the Stanford Faculty Senate in 2014 and began as an experimental pilot program in the fall quarter of the same year (Ruff 2016). X partners were from the School of Humanities and Sciences and the Stanford Computer Science Department was itself located in the School of Engineering. The first two CS+X majors approved were CS+English and CS+Music. However, these experiments proved to be unsuccessful, and the program was quietly terminated in 2019 having apparently fallen victim to a common drawback of joint majors—including too many unit requirements on each side of the joint degree. The announcement of the program’s retirement further elaborates:

The experimental CS+X JMP aimed to reduce the total unit requirement for each major. However, over the course of the pilot program, both students and faculty advisers agreed that the unit requirements for the joint major were burdensome. Students who dropped a CS+X JMP said there were too many units required and, in some cases, it prevented them from pursuing other academic opportunities, such as studying abroad. (Leighton 2019)

At UIUC, the CS+X program launched in 2013 and, in contrast to the Stanford experience, has seen robust growth. Launched initially as four partner X programs in the College of Liberal Arts and Sciences—Anthropology, Astronomy, Chemistry, and Linguistics—the program as of the publishing of this report consists of 14 blended degrees highlighted in Table 6-1.

The UIUC CS+X program architecture includes the following three key features:

• “X” ownership: Each CS+X degree is owned by the Department of X at UIUC, and the degrees themselves are granted by the Department of X. Budget models are highly variable across different universities, but in general, there are significant financial advantages for the department where the major is housed. At UIUC, these advantages accrue to each X unit.
• Careful units management: The Stanford CS+X JMP experiment appears to have floundered because students found the requirement to complete the CS and the X course requirements burdensome. At UIUC, each degree is roughly 40 percent CS, 40 percent X, and 20 percent general education

• requirements. As noted by leadership at UIUC CS (Rutenbar 2019), socializing the idea that each CS+X was a critical subset of the CS curriculum, and not all of the CS curriculum, was key.
• Common CS core: There is a reasonable argument that, in an ideal form, each CS+X degree would have a custom, bespoke CS curricular component, with courses specially designed to synchronize with the target X application area. Unfortunately, this makes scaling out the number of X partner departments rather challenging. The design strategy was a (mostly) common CS core, for all the X units. This made expansion much more tractable at the whole-campus scale. This also means that, for the purposes of the committee’s analysis, for most CS+X students at UIUC, their preparation is more uniform.

TABLE 6-1 University of Illinois Urbana-Champaign CS+X Degree Programs

SOURCE: Data from University of Illinois Urbana-Champaign, Blended CS Degrees, https://cs.illinois.edu/academics/undergraduate/degree-program-options/cs-x-degree-programs, accessed June 1, 2024.

Northeastern University has yet another take on this idea, with their undergraduate Combined Majors program. Their target X partners range widely, including the sciences, engineering, business, art, media, and design, and the humanities and social sciences. There are three broad categories of combined bachelor’s degrees: CS and partner X, data science and partner X, and cybersecurity and partner X.

The list of combined majors is extremely broad, as the following (abridged) examples demonstrate:

• Northeastern CS Combined Majors: Biology, Business Administration, Design, Economics, English, History, Journalism, Linguistics, Mathematics, Neuroscience, Philosophy, Physics, Political Science
• Northeastern Data Science Combined Majors: Biology, Business Administration, Chemistry, Criminal Justice, Economics, Health Science, Journalism, Linguistics, Neuroscience, Philosophy, Physics, Psychology
• Northeastern Cybersecurity Combined Majors: Business Administration, Criminal Justice, Economics

The Northeastern program is less homogenous in its curricular architecture but still reuses core CS and data science classes as the core of the computing side of each degree structure, alongside partner X discipline specific courses and university-wide general education requirements. For each combined major, there are also other required courses and sets of electives which are chosen depending on X. For example, a required course for CS+Design is Human–Computer Interaction, whereas a required course for CS+English is Natural Language Processing. A key requirement for approval of new combined degrees is that they must contain 9–10 courses from each discipline, 1–2 integrative courses, and a capstone. The motivation is that a student should be able to work or attend graduate school in either discipline as well as at the juncture.

The relevance of this overall CS+partner degree landscape is that these new programs create a new population of bachelor’s-level students with a subset of traditional CS bachelor’s degree training. That subset varies by institution and to some extent, by the focus of the partnered discipline, however, in some cases the course load is equivalent to the requirements for each individual major. This is a potentially positive development for the overall goal of increased flow into computing PhD programs.

The availability of student trajectory and outcomes data across the blended/combined CS+X style degrees vary widely with the relative age of the programs, including approximately 10 years for Illinois CS+X degrees and more than 20 years for Northeastern combined degrees. To further understand how long-term career trajectories and educational plans for these students relate to doctoral degrees in computing, the committee received preliminary data from leadership for both programs to answer the

question of whether students go into the workforce or to graduate school after completion of their degree and which discipline students pursued.¹

Northeastern has been graduating combined majors for 20 years and has tracked the outcomes of these programs. The students go to graduate school in each of the disciplines and often do research at the juncture. For industry positions, they most often either go to a pure-tech firm or to a company that works at the juncture (e.g., a bio-CS major might work at Google or at Pfizer). Northeastern has also tracked co-op placement (internships) and finds there is no difference in placement between within-discipline and combined majors.

At UIUC, for CS+X programs that have graduated at least 20 students so far, a rough analysis suggests about 5–20 percent go on to a graduate degree, with almost all of those going (fresh out) for a master’s degree and almost all going for a graduate degree go in a CS-related area. About one in five go straight to a PhD, and all of those go to a CS PhD (with the additional caveat that these are still small numbers). Anecdotally, it seems that only the STEM CS+X programs send some students to graduate school in something other than CS, and then it is X-related. Compared to CS more generally, these CS+X programs send a higher percentage of students to graduate school overall.

Direct Paths: Statistics and Data Science

In 2016, the Computing Research Association’s (CRA’s) Committee on Data Science published an article detailing the manners in which data science might serve as a basis for future computing research (Getoor et al. 2016). In this work, the committee describes the fundamental principles of data science as “a novel mix of mathematical and statistical modeling, computational thinking and methods, data representation and management, and domain expertise.” Data science and statistics knowledge provide a unique and necessary background for the study and understanding of computing. This section offers insight into the growing investment in the combination of data science and computing, as well as examples of research which encourage students of data science or statistical background to engage in computing research.

Data science education is a growing space encouraged by an increasingly data-driven world. As it matures and becomes more robust, it is expected to allow for a variety of pathways for undergraduate students to explore (NASEM 2018b). The University of California, Berkeley’s College of Computing, Data Science, and Society, as well as Cornell University’s Department of Statistics and Data Science housed within its College of Computing and Information Science, are just a few examples of institutions acknowledging the interconnected nature of data science and computing. New York University offers a

___________________

¹ Data from conversations with Carla Brodley at Northeastern University and Nancy Amato and Else Gunter at the University of Illinois Urbana-Champaign, for their respectively combo/CS+X degrees.

certificate program in computational social science as a means of equipping individuals interested in conducting data science work focused on the study of society with mathematical and machine learning (ML) foundations. Statistics as a field, separate but complementary to data science, has become more focused on numerical and computational techniques. This focus allows statistics students to similarly engage in interdisciplinary work utilizing data to analyze, solve problems, and investigate phenomena.

Rice University distinguishes its Data Science Initiative into two categories: methods development and applications. Both categories offer a multitude of interdisciplinary computational pursuits which might allow for tangible examples of research being conducted throughout the country. Applications of data science include computational analysis of medical technologies, geographic information systems, public health systems, social and public policy, space phenomena, and software design. A background in any application might lead to PhD level work where scholars scrape large amounts of data from a given platform to study and visualize social movements, public sentiment, gender politics, climate change, cybersecurity, urban development, or platform affordances. The methods development program at Rice University highlights how advances in data science methods directly support ongoing studies of CS fields, including ML, computer vision, and natural language processing.

Direct Paths: Online Computer Science Bachelor’s Degrees

Especially in an increasingly virtual world, opportunities for online degrees in computing have become abundant, with the Massive Open Online Course (MOOC) concept appearing around 2011. At first, there were a few large, informal experiments, such as the Stanford AI MOOC, which surprised the world when it rapidly recruited more than 50,000 participants (Markoff 2011). This was followed rapidly by several start-ups, including Coursera, Udacity, and EdX, that attempted to commercialize these ideas (The Economist 2013). Now, there are many computer science courses available across these and other global online course platforms, and recently online offerings have pivoted toward complete degrees on these platforms, initially master’s degrees, but now also bachelor’s degrees. Discussion of the master’s degree landscape will be continued below.

Offerings in this space began with individual courses on individual technical topics, and then progressed to bundles of related courses, labeled as sequences or specializations, with emphasis on named credentials that could be useful for employment. Now, full bachelor’s degree programs are available through these online course platforms in computing and other related spaces. For example, Coursera offers two complete bachelor’s degrees in CS at the Birla Institute of Technology and Science and the University of London, as well as one in data science and AI at the Indian Institute of Technology Guwahati. Similarly, EdX offers two accredited bachelor’s programs in data

science and business analytics at the University of London, London School of Economics, and CS at Simmons University.

When considering pathways to doctoral degrees in computing, additional questions arise regarding the preparedness of students who obtain their degree through a completely online platform. In particular, opportunities for closely mentored undergraduate research experiences appear to be significantly different as students do not have access to professors within their department for summer research and must seek out opportunities outside of their program. However, unlike other STEM undergraduates, students completing online bachelor’s degrees are uniquely reachable for the purposes of marketing follow-on degree opportunities in computing. For example, all major MOOC platforms offer mechanisms to place opportunities for career advancement and employment in front of students with particular technical training, in some cases limiting opportunities to students exhibiting particular success in their courses. If efforts were made to reach these students about doctoral degrees in computing, it would be beneficial to understand student motivation for pursuing the online degrees, whether purely vocational or for career advancement, and trajectories for students completing these programs.

Indirect Paths: The Computer Science Master’s Revisited

As discussed previously, the master’s degree in CS has changed over recent decades, evolving away from a step on the way to a doctoral degree and emerging as a highly coveted terminal degree for evidence of vocationally viable computing skills. With this change came the emergence of large, online master’s degree opportunities in computer science. The Georgia Institute of Technology (Georgia Tech) Online Master of Science in Computer Science (OMSCS) is perhaps the best known example. Launched in 2014, in partnership with the Udacity MOOC platform, the degree was offered at both global scales and at an unusually small price point (~$7,000, significantly less expensive than many top-ranked peers). The program started with 500 admitted students but grew to 4,000 enrolled in 2016 (Carey 2016). A Harvard University study concluded that the program was, in effect, creating a new market for graduate education, as students who were not admitted to the program did not apply to any other graduate computer science degree (Goodman et al. 2016). This study suggests that “access to this online option substantially increases overall enrollment in formal education, expanding the pool of students rather than substituting for existing educational options.” Today, Georgia Tech OMSCS enrolls more than 13,000 students online. As of spring 2024, the program has graduated 11,952 students over its 10 years. Another example is the first MOOC master’s of science (MS) in data science on the Coursera platform, introduced by the University of Illinois in 2016 (Bollinger 2016), which later evolved to a full MS in CS degree (with a data science focus track). Today, this degree enrolls roughly 2,000 online students and

comprises the largest graduate degree program in the rather large Grainger College of Engineering at Illinois.

One relevant aspect of these online programs is the novel demographics of the student population; many were not in computing related fields but were coming into this data science opportunity from a broad set of disparate backgrounds. The idea that computing is an attractive field in which to pursue additional training, and that a master’s degree in computing is an accessible strategy to do so, is another unique component of this landscape. For example, the Align MS in CS at the Khoury College of Computer Science at Northeastern University has structured a master’s program specifically to address this market, promising to provide students the foundational knowledge needed to successfully study alongside direct entry graduate students. A specially designed two-semester on-ramp program allows these students access to the materials of a traditional CS master’s degree. Align has been in operation for 10 years; to date it has graduated 1,340 students and currently has 2,174 enrolled. The program claims that 70 percent of Align MSCS graduates become software developers as their first post-graduation job, and within 3 years, 35 percent of these Align alums have been promoted to senior roles. Northeastern started the MS-Pathways Consortium to help other universities create these types of programs, and as of 2023, there are 23 other universities offering bridge programs to the MS in CS for non-computing majors. In addition, seven NSF Scholarship for Service CyberCorps schools are working to create such bridges to their MS in cybersecurity with more in the pipeline.

Indirect Paths: Broader Bachelor Degrees

In the most recent growth of computing, increased demand to enroll in upper-division computing courses from non-majors has reflected the integration of computing throughout all sectors and disciplines. The end result is that computing departments and schools not only host thriving majors but also have become major service units to their institutions. As such, students with broader degrees in STEM disciplines are now exposed to a broader range of computing topics during their undergraduate training. The resulting undergraduate computing landscape is both broad and heterogeneous in nature, with the introduction of several new pathways into computing doctoral degrees.

Most natural sciences and engineering disciplines have a well-defined applied computational subdiscipline. For example, computational astronomy, computational physics, computational chemistry, and computer engineering are all well established in the STEM landscape. Notably, each of these disciplines is typically housed in the same school as the parent discipline and not in the computing department. Especially in recent years, there has been a significant increase in the use of ML and AI methods across these fields, necessitating more advanced computing training for students in these

subdisciplines. Unlike the physical sciences, computational biological sciences have historically been separate from primary biology departments, leading to a landscape that is exceptionally varied. For example, at Carnegie Mellon University, Computational Biology is a department in the School of Computer Science, at the University of Pittsburgh, Computational Biology is a degree program jointly owned by the Department of Biological Sciences and the Department of Computer Science. The University of Pittsburgh also has an independent graduate-only Department of Computational and Systems Biology in the School of Medicine.

Outside of STEM fields, a broad range of AI-enabled analytical tools have created a set of computationally enabled subdisciplines in the humanities and social sciences. The National Endowment for the Humanities now has a separate Office of Digital Humanities to support researchers in these humanities fields and the National Science Foundation has a Directorate for Social, Behavioral and Economic Sciences that similarly funds research in the social sciences. Students in these degrees will, again, see a subset of courses in a traditional computing degree, customized to pursuits in these disciplines.

Finally, as described in more detail in the following section, information science schools (I-Schools) have become increasingly popular. In addition to offering a parallel PhD path, alongside a more traditional computing PhD, many I-Schools have started offering a bachelor’s degree. These degrees often have a large subset of a traditional computing degree, but the disciplinary breadth of many I-Schools resembles the disciplinary breadth of the X partners in CS+X programs, as discussed above. I-Schools recruit faculty not only from computing and other STEM fields but also from the humanities and social sciences. While some I-School undergraduates will be indistinguishable from computing majors, many students will have a more heterogeneous preparation.

The broader commonalities across these undergraduate programs are as follows: (1) populations of undergraduate students with some training in computing; (2) significant disciplinary breadth in terms of their “non-computing” experiences; and (3) an increasing likelihood of exposure to some ML or AI methods, of relevance in their discipline. Obviously, the disciplinary breadth here can be quite significant. When considering pathways into doctoral degrees in computing, there is additional information needed to understand the preparedness of these students and likelihood to pursue a computing relevant graduate degree.

NEW OUTFLOWS

In addition to doctoral degrees in computing, as discussed throughout this report, there are new pathways available for students interested in pursuing doctoral studies in

computing related fields. For many discussions, these degrees are considered together with traditional CS degrees; however, doctoral programs for applied computational science and information science possess different requirements and touch different areas of study that require additional understanding, which is explored in the following section.

Information Schools

Many information schools have historically grown out of library and library science schools. In recent years, the curriculum at these schools have become more computationally oriented, employing applied computing techniques with increasing frequency in faculty and student research and coursework. Simultaneously, the curriculum at these schools have retained a focus on exploring and understanding the human experience, with researchers trying to understand the social impact of the events or data being examined. As a consequence of these two high-level curriculum characteristics, information schools train their students to use a variety of both qualitative and quantitative methods in their approach to research.

Areas of Study and Faculty Expertise

Unlike computing departments where the focus of study is on computers and the creation of new computing algorithms and techniques, information schools have a much less focused purview as they are studying all aspects of information; while computers may be used to facilitate that study, they are not necessarily the focus of the research being undertaken. As such, these schools cover an exceedingly diverse set of areas, with corresponding heterogeneous and interdisciplinary expertise in their faculty members. For example, departments may include the study of fields that may also appear in CS departments, including human–computer interaction, ML, AI, security, privacy, and visualization. Computing-adjacent fields, such as algorithmic fairness, data science, computational social science, health informatics, bioinformatics, information economics, computational linguistics, and data analytics, may also be areas of study. However, much further afield topics also exist in information schools, including library science, health and well-being, indigenous knowledge, information and society, digital youth, sociotechnical and information systems, ethics and policy, and information justice.

Given the breadth of research fields included in information schools, faculty expertise covers a wide range of topics. Some faculty members have doctorates in computing focused areas, including CS, human–computer interaction, ML, and information security. Others have expertise in other STEM fields or health-related fields, such as medicine, public health, or nursing. Because information schools also explore research related to humanities, faculty members also have backgrounds in traditional humanities fields including history, sociology, psychology, and economics as well as having more applied or

professional degrees related to law, business and business management, and library science. Finally, some faculty members come from backgrounds in political science or public policy, exploring topics related to technology in society and ethics and technology.

Computer Science Competency Requirements for Admission

Although it is anticipated that students will acquire additional training or knowledge while in the PhD program, there is an undefined core set of knowledge expected upon entry that can later be augmented through coursework related to a student’s specific subfield. Creating a strict set of course or skill competency requirements that would be applicable across information schools is challenging due to the highly interdisciplinary research opportunities within these schools. Consequently, admissions to information schools are highly individualized, with applicants highlighting which specific field of research they want to pursue and faculty members assessing applicants interested in their field to determine if they have the requisite skills and training. The level of computing knowledge and competence, therefore, depends on what level of computing technical skill is needed for research in the applicant’s desired subfield. For example, a student who wants to work on ML may be expected to have skills and knowledge equivalent to having earned a bachelor of science degree in CS, but they may still be required to take advanced ML courses during their PhD. Whereas a student who wants to work in health informatics may be expected to have a background in public health or nursing but not have advanced computing skills; they may be required to take computing classes to acquire the basic computing skills needed for research in health informatics.

What does appear to be common across the application materials for information schools is the applicant having earned at least a bachelor’s degree, which some schools prefer to be in the area the applicant wishes to pursue for their PhD; a clear idea of what field of research or area they want to do research on and an explanation of why that field interests the applicant, with some schools requiring specific faculty members to be indicated in the application; and evidence of the promise to be capable of research, which in many cases is prior research experience.

Post-Graduation Career Paths

Students earning their PhDs from information schools pursue a wider set of opportunities post-graduation than their computer science counterparts. Many graduates become faculty members in academic institutions, often in a diverse set of departments as there are a limited number of information schools and because the wide range of student research areas make them qualified to teach and do research in a variety of departments. Similarly, a significant fraction acquires positions in industry at technology companies as researchers, research and development engineers, and data scientists. Others choose to engage in public interest roles in government agencies, nongovernmental organizations,

nonprofits, and university-based research centers, while some join think tanks or become policy makers, economists, or directors of health science organizations.

Computational-X

Individuals with computing knowledge also now have the opportunity to obtain a PhD in computational science fields, such as computational biology or bioinformatics, computational chemistry, and computational physics. These programs vary considerably with respect to the expectations for applicants’ knowledge and proficiency of computer science and computing, the expertise of the faculty members involved in the programs, and even the departments in which these programs are situated.

Areas of Study and Faculty Expertise

Computational biology and bioinformatics PhD programs are frequently located in their own departments, separate from traditional biology PhD programs. In particular, these computational PhD programs may be situated in medical schools or situated in computer science schools, while a traditional biology department may be housed in the school of sciences at the same university. In these computational biology departments, faculty expertise may cover a wide range of scientific fields as well as CS fields. For example, faculty members’ doctorates may have been earned in science fields, including neuroscience, physics, chemistry, systems biology, pathology, cellular and molecular medicine, biochemistry and molecular biology, and biological sciences, or in computing or mathematical fields, including computational biology, CS, bioinformatics, biostatistics, electrical and computer engineering, ML, and applied mathematics.

Conversely, computational chemistry and computational physics PhD programs are frequently housed in traditional chemistry and physics departments, with computational chemistry and computational physics being a specific specialization students can pursue within the PhD program. In these departments, faculty members tend to have more doctorates in a less wide-ranging set of fields. For example, faculty members in some chemistry departments earned doctorates in physical or theoretical chemistry, physics and biophysics, chemical engineering, theoretical and computational quantum chemistry, or statistical physics. Similarly, physics faculty members in departments with computational physics specializations have doctorates in fields that are representative of traditional physics fields, with no specializations in computing. Faculty members with doctorates in computing-related fields are uncommon.

Computer Science Competency Requirements for Admission

Unsurprisingly, the expected computational proficiency and knowledge vary across the different science fields, with computational biology and biostatistics PhD programs requiring greater knowledge and skill than the other computational sciences.

Computational biology and bioinformatics programs may expect students to have advanced preparation in multiple fields, including biology, CS, statistics, or mathematics. Students in these PhD programs frequently have undergraduate training in data structures, algorithms, and ML. Departments vary with respect to the level of computing skill required; in some departments, students with mathematical or statistics academic preparation instead of CS may be expected to still have a basic understanding of programming languages frequently used for processing data (e.g., Python or R), while other departments expect significant computer programming experience equivalent to introductory programming and data structures courses.

In contrast, chemistry and physics PhD programs do not typically specify any computing proficiency requirements for applicants, likely because computational chemistry and computational physics are simply considered specialty research areas with the larger PhD program for that science; requiring all applicants to have computational skills when they are not intending to pursue computational research is not necessary.

Post-Graduation Career Paths

Across all of the computational science fields, PhD graduates pursue postdoctoral and faculty positions. Some continue on to medical school or, in computational biology, become medical residents. Additionally, many become industrial research scientists or data scientists, take jobs at start-ups, and pursue careers in government and nonprofit research laboratories or other government agencies.

RECOMMENDATIONS

There is a rising set of heterogeneous paths that could feed into PhDs in computing. At the bachelor’s level, many of these rely on subsets of the traditional CS bachelor of science for their own computing-related requirements but different degrees, such as different adjacent X disciplines or CS+X styled blended degrees, incorporate different subsets of a traditional computing degree. There is a rising landscape of adjacent degrees at the bachelor’s and doctoral levels in the applied computational science arena as well as evidence of AI being integrated into some of these adjacent STEM areas. At the same time, master’s degrees in computing have largely pivoted toward terminal, professional degrees with focus on upskilling and credentialing for workforce development. Yet, because these programs appear to be widespread and available at very large scale in online MOOC-style formats, they represent potentially large populations of students with a set of potentially PhD-ready skills.

Although the breadth of this landscape of programs and degrees that overlap with traditional computing degrees is increasing, data to track career trajectories for all of these new and adjacent inflows is lacking. In many cases, schools are independently tracking their student outcomes, but there is not yet a national scale survey instrument for these programs. As such, it is still difficult to understand quantitatively how these new inflows might be feeding PhD programs in computing.

Recommendation 10: The National Science Foundation and the Computing Research Association should collect data on the trajectories and outcomes for students with undergraduate degrees in fields other than computer science and engineering—including information sciences, data science, CS+X interdisciplinary degrees in external departments, computational science, and computational physical and biological sciences.

Recommendation 10-1: To assist applicants transitioning from majors other than computer science and engineering, the National Science Foundation and the Computing Research Association should survey and track admissions requirements for doctoral degrees in computing, including course requirements, research experience expectations, minimum test scores, and other qualifications for admission.

Recommendation 10-2: To better understand the impact of new pathways from fields other than computer science and engineering into doctoral programs in computing, the National Science Foundation and the Computing Research Association should coordinate to collect data pertaining to the specific bachelor’s and master’s degrees received prior to admission into these doctoral programs.