Special Needs and Concerns of Non-Doctorate-Granting and Less Recognized Institutions
To this point the discussion has focused on the situation of experimental computer scientists and engineers at research-oriented academic institutions with doctoral programs, the Forsythe list schools. Yet institutions without doctoral programs face many of the same problems. Moreover, factors such as size, faculty mix, history, and resource availability can further complicate the matter of career advancement for faculty at schools without doctoral programs. The opportunities and problems of non-doctorate-granting (NPhD) institutions are considered in this chapter.
Chapter 1 argues that the reasons for conducting experimental computer science and engineering research (ECSE) at universities include keeping the faculty on the cutting edge of this fast-moving field, maintaining the vitality of curricula in a technologically sensitive area, and keeping course work exciting by enabling faculty to draw on state-of-the-art examples and exercises from their research domains. These benefits are at least as critical for the NPhD schools, with their greater emphasis on teaching, as for Ph.D.-granting institutions. Accordingly, it is advantageous to use research experience as a means of achieving and maintaining high technical value and relevance.
NPhD schools vary: they include four-year colleges, universities with master's but not Ph.D. programs, and to a lesser extent, Ph.D.-granting institutions that graduate very small numbers of Ph.D. stu-
dents every year. However, they share one crucial characteristic: they do not have a long-established research tradition, and many such institutions have only recently enlarged their missions to include ''research" to any substantial degree. This lack of a research tradition is an example of a problem that Ph.D.-granting institutions do not have, but that manifests itself, for example, in evaluating faculty for promotion at NPhD schools. This and other problems of NPhD schools can be grouped by their common features into three categories: mission, size, and resources. The issues covered here are also generally applicable to universities that have recently established Ph.D. programs.
The NPhD school's goal of having faculty actively engaged in research has several consequences for the faculty, especially for those in ECSE. Chief among these is the substantial teaching load that is characteristic of NPhD institutions. Teaching is a time-consuming activity. When it is done right—and it must be done right, because quality teaching is often the selling point of NPhD schools—there is little time for research. This is a critical problem for ECSE faculty, given the large investment of time and intellectual resources required to create or experiment with an artifact. No school can, in fairness, expect significant research from ECSE faculty in the presence of extensive teaching obligations.
The mission of NPhD schools implies a second serious consequence for ECSE research: the absence of doctoral students. As noted in Chapter 2, Ph.D. graduate students are essential to creating artifacts, the medium in which ECSE research is conducted. By definition, there are no doctoral students at NPhD schools. The alternative is to engage advanced undergraduates and master's degree students in constructing artifacts, but the nature of their participation will differ from that of doctoral students. First, they must be closely supervised, a characteristic that is consonant with the goals of NPhD schools. Second, the requirements of the task must be more completely specified because the student presumably has less background. Third, the magnitude of the task must be modest because the student's available time—both in duration and in hours per day—is typically more modest. Fourth, many master's students are not full-time students. Such considerations limit the artifacts that can be created and the experimentation that can be conducted.
A third aspect of the mission of NPhD schools that affects ECSE research concerns the ability to compete for grant funding. Both federal agencies and industry favor Ph.D.-granting institutions in re-
search funding, and, to a large degree, NPhD schools are ill-prepared to compete for foundation funding because of the lack of administrative support for all aspects of the process. Nevertheless, it seems to be in the interest of both industry and the federal government to provide research support. As noted, the quality of the educational offering is greatly enhanced. Industry should take note of the fact that its laudable policy of donating equipment for teaching at NPhD schools can probably be greatly leveraged by a modest additional offering of state-of-the-art equipment to the faculty for research. A faculty member excited about the research he or she is doing on a company's computer will be a more effective teacher to the students using the company's teaching donations.
Most NPhD schools are smaller than the typical Ph.D.-granting institutions, and so computer science and engineering (CS&E) departments in NPhD schools have fewer faculty. Indeed, to put the NPhD peer group's concerns into perspective with other parts of this report, it may be difficult to identify "experimental" or "theoretical" faculty members at NPhD schools. This can be an asset—certain issues affecting the intellectual environment at Ph.D.-granting institutions are not relevant for NPhD schools—but it can also constrain members of the peer group.
One consequence of small size applicable to faculty who conduct experimental work is that there are few colleagues with whom to collaborate. Given that collaboration is characteristic of much ECSE research, experimental work at NPhD schools must be formulated especially carefully. If an artifact is to be created, it should be smaller in scale and narrower in range because of the smaller number of researchers available to bring expertise to the work.
A much less easily resolved consequence of small size is the lack of a suitable academic mentor. Although there are certainly faculty on campus from whom a new assistant professor can receive guidance about teaching and other aspects of academic life, research guidance and direction may be unavailable. As suggested for Ph.D.-granting institutions with no suitable experimental senior faculty member to act as a mentor, NPhD schools may be able to enlist a faculty member at another school to fill this role. An optimal solution to both the lack-of-collaborator and the lack-of-mentor problems would be to promote extramural collaboration.
Finally, small size brings the reality of inadequate staff support. Although this is partly a resource problem (see below), it has another
implication. Because the staff inadequacies at NPhD schools are so severe, faculty are often called on to perform system administration and maintenance tasks for departments and teaching facilities that are performed by technical staff at Ph.D.-granting institutions. These tasks take valuable time that could be applied to research. NPhD school administrations must provide staff support if they expect meaningful research from their faculty. Staff support is as beneficial as a reduced teaching load in terms of allowing time for research. It also represents a more economic allocation of resources.
Budgets are tight at NPhD schools and resources are scarce, and it is unrealistic to set a goal of faculty research without providing the means to achieve it. Critical to conducting ECSE research, as noted throughout this report, is access to the Internet. The ECSE community is electronic, and faculty cannot realistically participate without being, literally, plugged in. The importance of Internet access is underscored by the fact that it is key to solving the collaboration and mentoring problems discussed above. Similarly, hardware, software, and laboratory space are necessary to support experimental research. Although successful researchers can perhaps be expected eventually to provide for much of their own equipment and software needs, researchers at NPhD schools need seed funding to create laboratory facilities and sustained support for maintenance, staff, and so on.
It is obvious that if ECSE relies almost exclusively on conferences as the means of rapid dissemination of information, it is essential for ECSE researchers to attend conferences. NPhD schools have especially limited travel budgets, which constrains the opportunity for experimentalists to travel. Nevertheless, if faculty members are to be effectively engaged in experimental research, they must make their accomplishments known, learn of the latest advances in their field, see demonstrations of proof-of-existence artifacts, and so on, at workshops and conferences.
MODELS FOR CONDUCTING ECSE RESEARCH AT NPHD SCHOOLS
The smaller scale of ECSE research projects feasible at NPhD schools mitigates some of the problems noted elsewhere in this report (e.g., the need for extensive infrastructure); nevertheless, many problems remain. The key question is how to organize ECSE research in the NPhD context to maximize success.
As already noted, the size and complexity of artifacts are constrained by the resources that can be invested in them—faculty time, student time, funding, equipment, infrastructure, and so on. Thus the first model for research at an NPhD school is not to build artifacts at all, but rather to concentrate on experimenting with existing artifacts (e.g., software that has been updated or hardware located at other sites and accessed remotely) or verifying the experimental claims of others. Enormous savings are realized by not having to create the original artifact.
Because the tradition in ECSE is to share, it is not difficult to acquire access to artifacts; research software can be copied by using FTPs, and the remote log-in capabilities of the Internet can be used to obtain access to research hardware located off-site. Once remote capabilities are available, experiments in the proof-of-performance research style are possible. Of course, the proper equipment will be needed to run them, but in many cases a scientific workstation suffices. Additionally, there may be a natural "research constituency" available to those using this approach. That is, other scientists, possibly including the artifact's creators, may be working in the area and share an interest in the results, thus forming a valuable technical peer group. Although such work might not be publishable as a full article or conference presentation, it might be appropriate for publication as a "technical correspondence."
A second model for research by NPhD school faculty is to collaborate with faculty at Ph.D.-granting institutions. Such collaboration could take the form of contributing to the creation of, or experimentation with, a large artifact. The mechanism for initiating such collaboration is the personal relationship of the researchers involved. To launch the research properly may require the NPhD faculty member to visit the project, for example, during the summer. At project meetings the (often ill-specified) research strategy for creating the artifact can be established. Then, by using the Internet, telephone, fax, and other forms of communication, the project collaborators keep in contact during the academic year. Such arrangements are decidedly in the interest of all participants.
Computer science departments at NPhD schools should encourage faculty interested in experimentation to assess realistically how such research can be conducted under the prevailing conditions. At
Ph.D.-granting institutions, experimentalists should be encouraged to pursue collaboration with faculty at NPhD schools when there is an intellectual basis for it. Federal agencies can promote such collaboration in several ways, including encouraging and simplifying grant subcontracting. By exploiting the unique features of ECSE, greater research diversity and higher-quality education can be achieved.
As noted above, research has only recently become a goal of NPhD schools, and the research tradition may not be strongly established. Thus, when evaluating a faculty member for promotion, senior faculty and administrators at NPhD schools, lacking decades of experience in assessing research and applying newly established policies, may be even more inclined to rely on raw "paper counts" than their peers at Ph.D.-granting institutions. This, of course, greatly jeopardizes ECSE faculty, who may have a modest publication list. Guidelines described elsewhere in this report should be applied when assessing ECSE faculty.
Realism must be applied in evaluating the accomplishments of junior ECSE faculty for tenure at NPhD schools: expectations must be in accord with circumstances. As already discussed, time is the most critical component of a junior faculty member's career. It must be possible to begin research quickly, there must be time to conduct the experimental work, and there will be a delay in the time between the artifact's or experiment's completion and the time the impact of the work is perceived. Resources—equipment, Internet connections, software, and so on—as well as students are critical to success. When assessing the faculty member's record and when requesting letters of evaluation, the constraints and circumstances that affected the faculty member's probationary period must be made explicit.