Teaching Responsible Computing Playbook

Teaching Responsible Computing Playbook

Working Across Institutions

Authors: Stacy Doore, Sorelle Friedler

While the overwhelming majority of work on incorporating responsible computing into computing curricula happens within the boundaries of a single institution, there are definite benefits in working on incorporating responsible computing across institutions. Most notably, if the team developing a responsible computing intervention wants wide adoption of the material they are developing, developing the material across institutions gives the team a head-start in scaling their material. Further, if the material is developed across different institutions, then the assignment would be tested across different student populations, making it more likely to be adopted by a diverse set of schools. Working across institutions also forces the team to explicitly state any implicit assumptions on background knowledge/student abilities and in the process re-visit these assumptions. Learning from other institutions and their norms could also give ideas on how to make changes in other institutions that can help with incorporating responsible computing. While the rest of the section is focused on working across different institutions some of the benefits of diverse viewpoints and norms can be simulated within one institution by working across disciplines or working with colleagues from the same department who recently joined the department from another institution. For the rest of the section it will be assumed that team members belong to the same (or very similar) departments across institutions.

While working across institutions helps in developing responsible computing material that is more portable across institutions, there are challenges in making such a collaboration work. While not all of them are specific to developing responsible computing material per se, the challenges of working on responsible computing material make some of the challenges more acute than for a more traditional computing material. Next, four such challenges are highlighted.

First, the curriculum could be structured differently in different institutions. In some cases there might not be an equivalent course but even if one focuses on required courses (see the choosing computing courses section for more), differences can crop up in the following ways: (i) a course that is required in one institution might be offered as an elective in another institution, (ii) even if two institutions have the same course (even with the same name) the actual content in the courses could be pretty different. This means that an assignment developed for a course in one institution might not be appropriate for a similar course in another institution.

Second, the baseline background knowledge of students in the same class might differ widely across institutions. Notwithstanding the difference in the contents of technical courses, there can still be some common expectations of the technical background of a typical student in a traditional computing course across institutions. However, for discussing responsible computing, a basic background in social science and humanities can make a huge difference in what degree of difficult conversations one can have in the classroom. While the Gen.Ed. requirements in an UG curriculum could provide such background, whether computing students can be expected to have such background would differ widely in say a small liberal arts school vs. a large engineering school. While student backgrounds are definitely a major factor, these differences also tend to show up in faculty background as well, which can also be a factor in how easily responsible computing can be incorporated into the curriculum.

Third, the amount of control an individual faculty has over changing the course/curriculum content also poses different challenges. Smaller schools where one faculty member essentially teaches all offerings of a given course could find it easier to incorporate responsible computing in the course given the faculty member is onboard. Changing courses in large schools where a large pool of faculty members teach the same course could take more time because more buy-in is needed. On the flip-side in the former case, if that faculty member moves on, then the responsible computing material could be dropped by the next instructor while in the latter case, it would be much harder for an individual faculty member to drop the responsible computing material from a targeted course. The varying degree of control also could determine the nature of assignments/interventions that could be planned.

Finally, the kind and amount of support different institutions can provide to incorporate curriculum changes (including but not limited to responsible computing interventions) can also vary widely. For example, not all schools might have separate centers for teaching and learning. This means e.g. one institution has experts who can help instructors create evaluations based on the latest education research while in another institution the instructor might need to create the course evaluation questions on their own. However, such a disparity can be used to the advantage of the team in that team members from one institution could (indirectly) avail of the resources available in the other institutions.

Key Questions:

  • Who will be the partner institutions? While such partnerships could start at the institution levels (e.g. if multiple schools are part of the same state school system), such partnerships are more likely to start at the faculty level. In other words, the partnerships could start with collaboration between individual faculty members in different schools.
  • Which course(s) will be targeted with the partnership? If the partnership is being started by individual faculty, then targeting common course(s) being taught by the individual faculty members could be one option. Note that these common courses could be taught at different levels (e.g. a required algorithms course vs. an advanced algorithms elective). If there are multiple options, see the choosing computing courses section on how to choose among them.
  • How will the difference in curriculum structure be handled? Once the appropriate courses at each institution have been identified, the actual course content and schedule must be taken into account. If the same material is taught in courses at different levels (e.g. sophomore course in one institution and junior course in another), then the expected technical student background must be taken into account. This might mean different institutions have different versions of the same base assignment.
  • How will the difference in background knowledge of students in social sciences and humanities be handled? Discussion based activities are heavily dependent on students’ background and willingness to engage and societal and ethical issues. Depending on the culture of the institutions, a typical student in two institutions would have differing backgrounds and willingness to engage in US history or societal inequities. This implies that instructors in places where students' background in societal issues is lacking would need to provide more background. Even in institutions where a typical student will have the required background, making the background information explicit is also beneficial to students who are not the norm (e.g. consider an institution with a small(er) international student body).
  • How will differing control of individual faculty members to change curriculum/course content be handled? Differing degree of control could determine the style of responsible computing interventions: e.g. in institutions where the instructor has full control over a course, one could change the topics being covered and have longer projects to incorporate responsible computing while in other places where no single instructor “owns” a course, doing guest lectures with short assignments might be what is feasible. Further, the amount of preparation needed before the classes start could differ (e.g. if a course is administered with a big teaching staff, then the TAs have to be trained in the new material and/or have to test out the assignment before classes start).
  • How will differing institution support be handled? Developing a responsible computing intervention in most cases definitely benefits from support and advice from outside the computing department: ranging from help with creating proper evaluations to having access to colleagues in other departments with relevant expertise in a non-computing discipline. The presence of such expertise in a given school depends on the size and nature of the school. Having dissimilar institutions could make doing these tasks more difficult for some than others.
  • How can each institution partner support each other? While differing institution support and norms can create challenges, the differences can also be used as an opportunity to avail oneself of resources that are present in a partnering institution only.


☐ Identify partner institutions

☐ Identify target course(s)

☐ Identify alignment and ways to handle different curriculum structure in different institutions

☐ Identify ways to handle different student backgrounds in social science and humanities

☐ Identify ways to handle different levels of control of individual instructors to change curriculum/course content.

☐ Identify ways to handle the different institutional support and norm.

☐ Identify ways in which each institution partner can support each other in the implementation of the curriculum.


Brown University, Haverford College and University of Utah

At Brown University, Haverford College and University of Utah we created responsible computing assignments/projects in data structure and algorithms courses. The three schools are of varying sizes: from Haverford being a small private school to Utah being a large state school.

Checklist walkthrough (Here, we show how some of the checklist items in the previous section of this document could be used. The checklist did not exist when this partnership was created, so this walkthrough is meant only for illustrative purposes.)

  • Identify partner institutions
    • We met at a workshop on teaching responsible computing topics and began discussing the lack of such topics in the data structures and algorithms curriculum. We decided to work together to start addressing this shortcoming.
  • Identify target course(s)
    • We made the decision to target theoretical courses since at the time this collaboration started we were not aware of any work on incorporating responsible computing in theory courses, and our own research and teaching interests lie in these areas. We ended up focusing on data structure and algorithms courses.
  • Identify ways to handle different curriculum structure in different institutions
    • At Haverford, the data structures course is a traditional CS-2 course and Algorithms is a 300-level course and both are required (but there are no algorithms electives). Both these courses were targeted for the responsible computing assignment.
    • At Brown, the data structures is also in a traditional CS-2 course but the course covers more material than a traditional data structures course (e.g. it covers Dijkstra’s algorithm for path finding). Only this course was targeted for the responsible computing assignment.
    • Utah had the same course structure as Haverford but also had algorithms electives. All three levels were targeted for the responsible computing assignment.
    • The mis-match in topics covered in the data structures courses at Brown and Haverford made some assignments developed at Brown not suitable for Haverford (since they were advanced for the data structures course but not advanced enough for the algorithms course). Some such assignments were modified or moved to the upper level course at Haverford, while some assignments would have required too many changes to be appropriate at the other institution.
  • Identify ways to handle different student backgrounds in social science and humanities
    • At both Brown and Haverford, students have a broad liberal arts background and hence, even a typical computer science student could be assumed to have a background in US history and structural inequities and in general are willing to engage in discussions related to societal issues.
    • By contrast computer science students in Utah typically do not have the background to meaningfully engage in topics related to societal issues.
    • The above meant that assignments created at Brown or Haverford but taught at Utah needed additional attention paid to giving societal background material explicitly and structuring discussion sections so that students could engage.
  • Identify ways to handle different levels of control of individual instructors to change curriculum/course content.
    • At Haverford, a single instructor always teaches the algorithms course every year, so it was easy to make the necessary changes to incorporate responsible computing assignments. Different instructors teach the data structures course, so developing clear and reusable materials was more important for that course.
    • At Brown, while there was a single instructor and there was department buyin, the data structures course has a big teaching staff, which meant the bulk of the work had to be done before the semester started.
    • At Utah, there were multiple instructors and there was not broad department level buyin. It turned out to be easier to incorporate guest lectures (accompanied by short assignments) into the various courses.
  • Identify ways in which each institution partner can support each other in the implementation of the curriculum.
    • At Brown, we developed some survey questions using the best practices from the field, which we were then also able to use at Haverford.

Bowdoin College, Colby College, and U of Maine

At Bowdoin, Colby, and U. Maine, the responsible computing modules were first created at Bowdoin College. After a project team member was hired at nearby Colby College, the development and pilot work continued at both institutions. The next stage will be using the team’s in-state network to expand the module pilot at the University of Maine.

Checklist walkthrough (Here, we show how some of the checklist items in the previous section of this document could be used. The checklist did not exist when this partnership was created, so this walkthrough is meant only for illustrative purposes.)

  • Identify partner institutions
    • Stage 1 partners were Bowdoin College and Colby College. Stage 2 partners will include the University of Maine and Bates College.
  • Identify target course(s)
    • At Bowdoin College and Colby College, we identified several early level CS courses (e.g., CS1, CS2, etc.) to develop and pilot computing ethics modules to share across institutions. Introductory course module topics include: CS representation and bias, gaming technologies, entertainment and media, facial recognition, and voting technologies.
  • Identify ways to handle different curriculum structure in different institutions
    • The introductory concepts taught across the institutions in Stage 1 and Stage 2 are not dramatically different, however, the size differences between the classes in the small, liberal arts context (20-30 students) vs. the state university (100+ students) means re-thinking how, when, and where the modules will need to be adapted for the larger classes. For example, the smaller schools provide more flexibility in the scheduling of the module readings, discussions, and activities. The larger university implementation will need to incorporate assistance from the lab instructors and learning assistants to lead and moderate online discussions.
    • One important difference during Stage 1 was that due to different approaches to dealing with COVID, Bowdoin was restricted to online only classes and Colby was able to return to campus to teach face to face. This meant that the pilot modules needed to be converted to an online format for Bowdoin classes and all materials and activities were adapted to make them work for a synchronous meeting schedule through a project website. The Colby teaching format allowed for more hands-on activities and group role playing due to the face to face format but the activities needed to be adapted for smaller group sizes and be broken up into different meeting spaces.
  • Identify ways to handle different student backgrounds in social science and humanities
    • Because both Stage 1 schools were small, private colleges with CS programs of similar size, student demographics, and a strong focus on liberal arts and interdisciplinary CS pathways, we had few challenges working across institutions except for faculty feeling comfortable making space in the busy early level syllabi for embedding the module readings and discussions.
    • During Stage 1, there were a significant number of students who were CS and liberal arts domain double majors (CS/Sociology, CS/Economics, CS/Psychology, etc.) who were interested in the responsible CS ethics curriculum development and served as students researchers. These students were passionate about how discussions of ethics in CS might enhance their own classes and had many ideas for new module topics and narratives. They created and helped to pilot modules, curated narratives, created project dissemination materials such as conference papers and posters, and enrolled in Independent Studies with project faculty to devote more time to researching a particular topic related to their own interests.
    • We are anticipating that the Stage 2 adaptation to the university setting will require more online student engagement efforts and small group moderators to promote discussion and activity completion. Because the module topics should be fairly familiar to most CS students, including first generation and non-traditional students, we hope the module content will still be engaging and thought provoking.
    • There may also be a wider divide between students in the Stage 2 context due to larger diversity of social and political viewpoints, hence the plan for active discussion moderators.
  • Identify ways to handle different levels of control of individual instructors to change curriculum/course content.
    • While smaller schools may have an advantage of a high level of flexibility and control over the course content and placement of the modules, when working with faculty teaching the same courses within our own institutions, we have experienced reluctance to embed the modules as a part of technical concept discussions or found some faculty were in favor of pushing them to the end of the semester only as an ‘add-on’.
    • During the Stage 2 adaptation, we will benefit from the support of a lead faculty member being able to dictate when and where the modules will be taught through the lab instructors and learning assistants.
  • Identify ways to handle the different institutional support and norm.
    • As we are working in Stage 2 to adapt and pilot the module materials in larger introductory CS classes, having already adapted the materials for Bowdoin’s online implementation, it should be easier to accommodate UMaine’s larger size classes and asynchronous structure needed for small and large group discussions in each of the modules.
    • The new website format may also eventually allow students to talk with one another across institutions about the narratives and share how their thinking may have evolved over the course of the module or the class. Facilitating student conversations across institutions and very different CS programs is a secondary goal for the next phase of this project.
    • Sustainability of the collaboration beyond the Stage 2 funding is already in place, with an NSF Collaborative Research IUSE submission across the institutions to form faculty learning communities around many of the ethics module topics and support a cross-institutional training program for undergraduate learning assistants to lead module discussions in classroom and co-curricular CS dept. settings.
  • Identify ways in which each institution partner can support each other in the implementation of the curriculum.
    • We have relied on our institutional partners to help with implementation at each institution. For example, Bowdoin partners have a deeper level of expertise in narrative structures, how to find/store new narratives for the repository, and ethical frameworks. Our Colby and UMaine partners have more expertise in connections to emerging applications and their ethical impact for individuals and communities. We have guest lectured in person or remotely on these topics to provide expertise and coverage of specific aspects of the modules. In Stage 2, we are making videos of some of this content that can be used with larger classes and thinking about ways to offer remote panels that feature faculty from all of the institutions so that students can tune in to participate in a live discussion of the topics in a more interactive way.


Related Pages

Authors and Contributors

Image of Stacy Doore

Stacy Doore (author)

Image of Sorelle Friedler

Sorelle Friedler (author)