Students Explore Network Maps
Professor Edenberg speaks to the class.
Georgetown’s Responsible CS team used its second engagement in Professor Ray Essick’s Advanced Programming course to explore network maps in ethical terms. When a computer scientist designs a network map, what ethical considerations should she be attuned to? How can she evaluate how ‘good’ a network map is from an ethical vantage point? Asking such questions can help students understand how highly technical choices can have ramifications for millions of people’s interests in privacy, unbiased hiring decisions, and other core values in a digital economy. Professor Essick and Professor Elizabeth Edenberg, Senior Ethicist at Ethics Lab, co-led the session.
We began with a seemingly simple scenario with three actors: a job applicant, a company advertising a job, and an online recruiting site. Working in pairs, students first drew a basic UML sequence diagram for the use case of submitting a resume to an employer via the online recruiting site. Of particular interest were the communications between each actor: what transactions occurred, and what data was shared during each transaction. We then challenged the class to reflect on the interests and concerns of the three actors. Students identified a range of interests, including the applicant’s wish to have her application prioritized, transparency about which requirements are absolutely essential to apply for a job, and an applicant’s interest in ensuring that her profile is not “filtered out automatically” from viewing the ad by the recruiting site. The class also grappled with added complexities, including how a recruiting system might rank-order multiple candidates, how its recommendations could modulate based on an applicant’s professional network, and whether various actors within a company (for example, interviewers and HR) should be privy to distinct sets of data. With these ethical considerations in mind, students built out their network maps to reflect the ethical interests of the actors they decided to prioritize.
The Problem of Bias
Students were especially interested in the problem of bias, asking questions about various ways in which bias can manifest in the hiring process and how such risks might be mitigated. Building on these questions, the second half of the class challenged students to consider who has responsibility to ensure that hiring is done fairly and what network map adjustments could be made (and by whom) to ensure that only the right information gets through to ensure impartial hiring decisions. One student suggested that––while applicants need to ensure all their information is accurate––applicants should have the fewest responsibilities because they have the least power. Another proposed that the employer should first assess skills and fit, leaving review of resumes until the end; a third student raised worries about this approach if applicant volumes are high. Yet another student said that resumes should be maximally formalistic and standardized to limit risk of bias based on personal idiosyncracies and ethnic, racial, or geographic identifiers. Returning to their network maps, students identified where ethical responsibility should be focused and introduced new technical requirements to account for ethical concerns.
As in our first session in Advanced Programming, the class discussed the normative valence of technical terms, focusing on what counts as a ‘good’ or ‘bug-free’ network map in ethical terms. Echoing a theme from our first engagement, Professor Essick reiterated why ethics should come in at the earliest stages of design. “If we wait until the API is set and then realize there’s a bad design decision, we can break things and rebuild, but it’s much harder and institutional interests get a lot stronger once designs are set in place.” Bringing in ethics at the ground level encourages “thoughtful decisions that minimize, rather than eliminate, problems when stakes are low and do what is right and not just profitable.”
This project is funded by a Mozilla Responsible Computer Science Challenge grant and is part of the university’s Initiative on Tech & Society, which is training the next generation of leaders to understand technology’s complex impact on society and develop innovative solutions at the intersection of ethics, policy and governance.