Politics, Ethics and Network Analysis

Regular readers of this blog are likely also subscribers to the SOCNET listserv, the preeminent distribution list for all things related to networks and network analysis. For those unfamiliar with SOCNET, the exchanges are most often related to methodological questions, conference and paper call announcements, and job postings. Occasionally, however, one post will spur a much more lively discussion among list members, and over the weekend the list witnessed exactly that.As many readers may have been following the exchange in real-time, I will only briefly summarize the proceedings. Last week a post-doctoral position with the Common Operational Research Environment (CORE) Lab at the Naval Postgraduate School was posted. The post described the job as the following:

CORE research projects support field operations in the areas of counter-terrorism, counter-insurgency, and stability/reconstruction. The research entails data coding, data entry into ACCESS and/or EXCEL, and data analysis using software packages such as UCINET, R, and Pajek etc. An educational and/or work background in geo-spatial analysis is also highly desirable.

This was followed by two posts that quickly plunged the list into a wild exchange that oscillated between political diatribe and genuine thoughts on the ethical application of network analysis techniques. Full disclosure, I interpreted most of the messages as more diatribe then thoughtful insight, and added my own request to move the political discussion off the listserv, to which I was quickly chided by Barry Wellman for attempting to stifle a productive discussion of ethics; wherein Barry reminded all that he was—in fact—a founder of the listserv INSNA (International Network of Social Network Analysis) in his signature line. Of course, as several people pointed out, there was never any real contestation to the discussion of ethics. Couching this discussion in accusations of “solicitation for people to abet in the murder of innocent civilians;” however, is not only counter-productive, but provides further evidence to the wildly misinformed position of many academics as to how the military attempts to apply findings from the social sciences.

As the discussion seems to have died down, unfortunately, I am left wholly unsatisfied. While the discussants were completely preoccupied with deciding whether a debate of ethics was relevant the the listserv (again, to which no one was actually protesting, and moreover was never a concern in any of the original posts), no discussion of ethics ever actually occurred. As such, there are two areas I would like discuss with regard to ethics, and then open the discussion to all of you.

Data sources – Social network analysis is fundamentally about understanding the structure and dynamics of human relationships, and as such, many data sources are inherently private. In the age of social networking services; however, people are freely adding otherwise private information about their relationships into the public domain. Very recently, the so-called ‘Gaydar’ project at MIT has created a stir on the ethics of making statistical inference about the sexuality of people on Facebook based on who their friends are. While the statistical power of the students’ findings is quite limited; however, it is not unreasonable to imagine a more robust version of this research being developed in the future. What then would stop someone with such a tool from using it to various decision on the basis of inferred sexuality from Facebook ties? Though never discussed, it would not be unreasonable to raise similar ethical questions about the work of Christakis and Fowler on obesity. What, then, are the ethical boundaries of data? It seems logical that for network analyses conducted from data gathered in a lab or through a survey the usual IRB requirements suffice; however, to what extent should these ethical boundaries be extended to public data? As many corporations, most notably Netflix, use these statistical techniques to generate revenue, should there be concerns as to the ethics of creating profit from network data?

Military/law enforcement application – As much of the SOCNET discussion was initiated by members’ concern that network analysis was going to be used to somehow cause harm to innocent people this is clearly an area that must be addressed. A critical starting point for this discussion is the importance of perspective. As the SOCNET exchanges highlighted, people have very different notions of what is and is not a “bad guy,” and thus use this to motivate any discussion of ethics. What is more important, however, is that most people can agree that bad guys do exist, whose stated (rather than interpreted) intent is to harm innocent people, and to what extent it is possible these people should be stopped. Network analysis is an imperfect tool, and when its application is motivated by political rather than practical goals it can certainly provide dangerous conclusions. As with any statistical analysis, the results in a network analysis can be chopped and altered to provide a desired result, but of course this is incredibly irresponsible. What is equally irresponsible, however, is to assume that the military or law enforcement are always guilty of this.

Having had first-hand experience in this area, I can attest to the fact that great care is taken to ensure exactly the opposite is true, and any tactical decisions never proceed from the findings of a single analysis. The ethical concern here, in my view, is the intent of the research. That is, if some research is intended to find exactly X, and any conclusion not X will be ignored, then this is clearly a gross ethical violation. On the other hand, if the military and law enforcement agencies are using network analysis (as I have seen them) to better understand the structure and dynamics of adversarial groups with a stated intent to harm innocent people, then the ethical concerns on the nature of the data collection, which becomes a matter of the courts—a different conversation entirely.

Finally, I would like to suggest that if there are any readers of this blog that fall into the category of “founder” or “organizer,” for SOCNET or INSNA I think a panel on ethics at the 2010 Sunbelt conference may be the most productive was to facilitate this discussion further. Again, I welcome your thoughts on these, or any other points related to politics, ethics and network analysis.

Other’s thoughts:

• SOCNET is the new CRAIGSLIST? – Steve Borgatti
• Thanks to Ann for the reference below, which is indeed an excellent reference on this subject. “Ethics and defense agency funding: some considerations,” Rebecca Goolsby. Social Networks, Vol. 27, 2.

Photo: The Boston Globe

Automatically Generated Related posts:

1. On the Disutility of Network Analysis and the Top 5 Problems
2. Measuring Network Effects with No Network Analysis
3. How to: Use Python and Social Network Analysis to Find New Twitter Friends
4. Network Analysis for International Relations
5. Network Analysis in the Social Sciences