acm-header
Sign In

Communications of the ACM

Viewpoint

Ask Not What Your Postdoc Can Do For You . . .


Ask Not What Your Postdoc Can Do for You, illustration

Credit: Shutterstock.com

The number of postdoctoral fellows in computer science (CS) has risen dramatically in recent years. Studies show that U.S. and Canadian Ph.D.'s taking postdoc positions have tripled since 2000,7 and the total number of postdocs in the U.S. rose to new, sustained highs over 2009–2014.6 It is now clear that postdocs are a substantial constituency in research-focused university departments.

The Computing Community Consortium (CCC), whose mission is to promote the vitality of computer science research, observed that postdocs in CS now play a much enhanced part in the conduct of research and education at universities. Postdocs are in training positions yet, anecdotally, most departments pay little heed to the training they receive. Certainly, support for postdocs is not on par with the education of graduate students. CCC asked: what are best practices in supporting the computer science postdoc population? An early document by Jones and Gianchandani3 analyzed such best practices and a 2013 Communications Viewpoint2 raised issues for the CS community more broadly. In the broader context of science and engineering, efforts to improve postdoctoral experience are discussed by Davis.1

Historically, postdocs have been "invisible": they work closely with their faculty advisers and that adviser's group, but have had little standing in the department or university. In essence they have not been "first-class citizens." This situation warrants investigating the best ways to support the career and contribution of postdocs. Concerns to address include the quality of training the postdoc receives, the quality of mentoring, the development of new skills, and the participation of postdocs in the community—both in their university department community and in the research community.

To address these concerns and learn more about the postdoc experience, CCC created a program, with NSF's financial backing, to develop, implement, and institutionalize best practices for supporting postdocs in CS. Competitive awards were made in 2014 to three groups that by design are quite different from one another: a consortium of New York universities (Columbia University, New York University, the City University of New York, and Cornell University, with 63 postdocs participating since the start of the program) led by Shih-Fu Chang and Julia Hirschberg; a consortium of the three research universities in Arizona led by Chitta Baral and Partha Dasgupta at Arizona State University and including the University of Arizona and Northern Arizona University (47 postdocs since the start of the program); and a single CS department program at the University of Washington (50 postdocs since start) led by Brian Curless. Each group includes evaluators to conduct surveys and focus groups, measuring the effect of the programs implemented. We will refer to the three sites by state name or university in this Viewpoint.

The purpose of this Viewpoint is twofold: to raise awareness right now about the need to serve the often neglected community of postdocs in CS departments, and to provide insight into what the New York (NY), Arizona (AZ), and Washington (UW) sites are finding to be the best way to implement best practices for their particular situations. The implementation and evaluation of these practices is ongoing; in a future article, we intend to delve into the details of the implementations and the numerical assessments of their impacts. Though our efforts are a work in progress, we believe readers in CS departments will identify with elements they can implement sooner rather than later.>

Here, we describe the challenges the three awardees decided were highest priorities to address and the rationales for choosing these challenges, and comment on aspects of the implementation of best practices they are pursuing.

Back to Top

Thoughtful Postdoc Evaluation of Career Goals

Postdoc positions in CS are typically short-term, lasting 1–3 years. The next step for a postdoc spans a wide range of options, including a tenure-track, research, or teaching faculty position, an industrial research lab, a startup. Thus, postdocs should use their training time to carefully evaluate career choices and to develop skills that will serve to further their career. To this end, all sites require postdocs to define and review their personal goals for both their current work and future positions. Both NY and AZ require the use of Individual Development Plans (IDPs) in which the postdoc sets concrete goals for development on a schedule. IDPs help frame conversations with postdoc advisers. NY requires the postdoc to discuss goals identified in his/her IDP with his/her advisor but leaves the submitted document confidential to the postdoc. In contrast, UW has created an online "Review of Progress" system similar to an IDP; the form must be filled out every six months, followed by required meetings with the postdoc's adviser to discuss goals and progress toward them. The Review form is tracked electronically to ensure compliance, and to identify problems. NY, AZ and UW have all found the written plans to be very effective. NY adopted an existing IDP, designed within the natural science discipline,1 and has found it to be useful but believes that the scope and content should be further customized to meet the CS field, a task currently being undertaken by the NY team.

Back to Top

Quality Mentoring

Guidance from experienced mentors is critical to help postdocs stay on track to achieve their goals. Interestingly, each site has a different, complementary approach to encouraging this mentoring. UW emphasizes getting the most out of adviser mentoring using the postdoc's plan to track and ensure that progress and career goals are discussed regularly with the research adviser. AZ has established non-adviser "Champions," who are faculty members not directly advising a given postdoc, but who are available to discuss IDPs, postdocs' concerns, career advice, and so forth. NY provides a mentor pool besides the postdoc's advisor, comprised of other CS faculty members in the consortium, professionals in the local industry ecosystem, in addition to individual career counseling within their program.

Back to Top

Acquiring Skills

Postdoctoral appointments provide both an opportunity to improve postdocs' research portfolios and to develop skills for their future careers; these skills will enable them to be more effective on day one of their next jobs. Each program promotes skill development for postdocs. Though in most cases the skill-development programs are aimed at academic careers (a common goal for many postdocs), a number of these skills are transferrable to industry careers. NY has a particularly strong program of frequently offered workshops and panels to develop leadership skills, academic writing, grant writing, teaching, public speaking, and entrepreneurship. They have also offered an extended teaching course comprised of 10 sessions. Leveraging the breadth of their consortium, they are able to utilize resources such as Cornell's leadership program and Columbia's writing programs. In two years, NY hosted an extraordinary 66 events, many of which are recorded and available online for review by members of the NY program. AZ offers panels, workshops, and mini-conferences that overlap with NY's described above, but also include topics such diversity and ethics. UW offers competitive awards that fund undergraduates to work on the postdoc's independent research. The award process offers a kind of academic, on-the-job training. In this program, postdocs define an independent project, write a short NSF-style proposal to fund an undergraduate to perform the research, receive critical reviews from a committee of faculty and past postdoc awardees, receive funding (if accepted) to perform the research, recruit an undergraduate, and mentor the undergraduate through the project. All of this is done independently of the postdoc's faculty adviser.

Back to Top

Effective Networking

A key part of landing the ideal job and being effective in that job is making oneself known to others in the field—networking. Postdocs benefit from attending conferences and talking to other researchers in academia and industry. To support this goal, each site offers travel awards that supplement any travel (for example, for conference presentations) already supported by advisers. NY additionally offers an annual community networking event with postdoc presenters and guest speakers; these events rotate among the institutions. NY and AZ host well-attended industrial networking events. AZ emphasizes social events among postdocs and industry exposure workshops to encourage in-person interaction with industry leaders.

Back to Top

Belonging

Each site observed that, prior to beginning these projects, a postdoc was hired by a faculty member and vanished into a lab, interacting with few others in the department, before quietly leaving. Now, all three sites sponsor activities to enhance the sense of belonging with the postdoc community and the department community. This "belonging" is important for postdoc satisfaction, breaking down barriers to foster out-of-area interactions, and ultimately for career development. AZ has instituted weekly research presentations by faculty and invited speakers, and has a social time preceding events; it also holds monthly informal lunches with postdocs and faculty to discuss future events and build a greater sense of community. UW has established per postdoc on-boarding orientations, email announcement of each postdoc arrival, department Web-page announcement of any major postdoc accomplishments, monthly postdoc-only lunches, an annual lunch with the department chair, inclusion in all email pertaining to researchers, and encouragement to attend the frequent social activities offered by the department. UW faculty also discuss each postdoc in annual review-of progress meetings, more broadly increasing faculty awareness of both the postdoc community in the department and the individual postdocs. NY hosts quarterly orientations for new postdocs, hosts community building events during National Postdoc Appreciation Week and community excursions (for example, a guided museum exhibit focused on computer science history in NY), in addition to the annual postdoc symposium rotated over consortium campuses for postdocs to present their research to other postdocs and faculty.

Back to Top

Institutional Support

Implementing and sustaining postdoc best practices requires a commitment from faculty and staff and support from the university as a whole; each of the sites funds personnel to run their programs. NY has a full-time staff member devoted to the effort, and AZ and UW each have a half-time staff member. NY's requirement has been greater due to the larger size of its multi-institutional effort and the greater organization and coordination required by the many workshops and other events it runs. UW and AZ also devote individual faculty time to engaging with the postdocs. Having faculty and staff support has been essential to developing sound programs to nurture postdocs appropriately, but such support is often difficult to fund outside of the programs described here, which are currently (but not indefinitely) supported by CCC. At AZ, support and participation from the graduate college and the Science Foundation of Arizona has helped in many events and has shown a path toward sustainability beyond the grant period. At the same time, taking inspiration from their CS program, AZ State University's graduate college is developing a general program for all its postdocs.

Back to Top

Conclusion ...

Our efforts are a work in progress, and we are continuing to evaluate the impact of our respective implementations of best practices. After completing their efforts, the three sites will collectively recommend a suite of effective strategies for training and nurturing CS postdocs, to ensure their success; these recommendations will appear in a follow-on article that provides greater depth about the programs, discussion of pros and cons of practices tried, and measurements of the impacts of the efforts. We do not expect a one-size-fits-all implementation of best practices. Each CS department has different strengths, challenges, and resources that will define the steps that they can realistically take. But it is clear to us that by taking a few actions, the support for postdocs can be enriched greatly. This observation is reinforced by the findings of a recent study by the National Academies of Sciences, Engineering and Medicine.5 Because postdocs now play a major role in the research activity of many departments, the quality of the work of those postdocs is a material determinant of the quality of the research of a department. Postdocs are not just another group to train; they are taking major responsibility in running labs, mentoring graduate and undergraduate students and even teaching.

Back to Top

... and a question

We conclude with an important question: How sustainable are these efforts? At a minimum, what we do to support graduate students can often support our postdocs as well. Both are at a somewhat similar stages in their careers. But such programs require dedicated staff, and such staff are expensive. Research awards and travel grants also have costs.

NY and AZ began their program by teaming with offices that support postdocs across the university. Indeed, much of the skill development support could be shared across postdocs in many science and engineering disciplines, yet still be of high quality for computer science postdocs. Individual development plans might likewise be common across "like" disciplines and their administration—to a great extent—supported by a school or the university, rather than by the department's lesser resources.

However, while some faculty advising and postdoc evaluation could be interdisciplinary, much of it needs to be specific to computer science. For example, computer science departments can take advantage of the vibrant research labs in industry to assemble a diverse mentor pool for CS postdocs, and topics covered in the IDP can be customized to meet the unique needs of the CS field. Creation of a faculty "postdoc committee" assignment is straightforward, but the faculty member(s) must be proactive in engaging the postdocs for the assignment to be of any value. Daily interaction with others in the department—students and faculty and postdocs—is also necessary for the postdoc to have a quality training experience.

Determining an effective division of labor and costs between the department and the larger university organizations will be key to creating sustainable programs for supporting a critical resource, our CS postdocs.

We urge all universities, departments, and faculty to consider: "Ask not what your postdoc can do for you, but what you can do for your postdoc ... to grow and advance toward a successful career."

Back to Top

References

1. Davis, G. Improving the postdoctoral experience: An empirical approach. Science and Engineering Careers in the United States: An Analysis of Markets and Employment. R.B. Freeman and D.L. Goroff, Eds., University of Chicago Press. 2009.

2. Jones, A. The explosive growth of postdocs in computer science. Commun. ACM 56, 2 (Feb. 2013), 37–39.

3. Jones, A. and Gianchandani. E. Computer Science Postdocs–Best Practices. 2012; http://bit.ly/2mInIFH

4. myIDP; http://bit.ly/2j0zlCB

5. National Research Council. The Postdoctoral Experience Revisited. 2014.

6. NSF NCSES data; http://bit.ly/2j0m790

7. The Taulbee Survey; http://bit.ly/2ANbDRP

Back to Top

Authors

Chitta Baral ([email protected]) is a Professor of Computer Science at Arizona State University.

Shih-Fu Chang ([email protected]) is Senior Executive Vice Dean, School of Engineering and Applied Science, and Professor in the Electrical Engineering Department and the Computer Science Department at Columbia University.

Brian Curless ([email protected]) is a Professor in the Paul G. Allen School of Computer Science & Engineering at the University of Washington.

Partha Dasgupta ([email protected]) is an Associate Professor of Computer Science at Arizona State University.

Julia Hirschberg ([email protected]) is Percy K. and Vida L. W. Hudson Professor of Computer Science and Chair of the Computer Science Department at Columbia University.

Anita Jones ([email protected]) is University Professor Emerita at the University of Virginia.


Copyright held by authors.
Request permission to (re)publish from the owner/author

The Digital Library is published by the Association for Computing Machinery. Copyright © 2018 ACM, Inc.


 

No entries found