Bots, Agents, and Other Human Subjects: Ethical Research Online

Elizabeth A. Buchanan, Ph.D. and Scott Dexter, Ph.D.


This paper investigates an emergent divide in research ethics discourses surrounding the concept of “human” subjects in emergent forms of computer science and Internet–based research. Using a cross-disciplinary approach, the authors seek to present novel ways of thinking through and solving applied ethics challenges facing researches in computer and information sciences.

The history of research ethics is based in and around biomedical and behavioral models, and subsequently expanded to include social and humanities-based models of research. Research ethics, in general, are codified in national legislations and in particular, in disciplinary norms. Sometimes such extant regulations and these disciplinary norms are out of sync, as in, for example, the Oral History Association, which successfully argued to be excluded from the purview of formal regulatory ethics boards in the United States and elsewhere. However, ethics boards across the world are conforming to stricter utilitarian models (Buchanan, 2010), often risking individual rights and justice in their practices. This movement may be the result of a number of factors, most recognizably, a more legalistic and litigious environment for researchers and institutions. But, we argue that this movement towards a stricter utilitarianism is also the result of emergent forms of research which minimize the “human” in research—this movement is characteristic of a “research ethics 2.0,” (Buchanan, 2009).

Part of the challenge faced in refining ethics standards to properly account for research conducted on or within a network is that the “raw material” of research tends to be viewed by the researcher as “data objects” rather than “human subjects”. In some projects, say, an effort to develop new network protocols for optimal real-time delivery of video, the data being studied is probably not sensibly construed as being produced by a human (though even here, IP addresses of participating computers may be recorded, and may be linked to humans – is this an issue of ethical concern?) Other projects may focus on segments of a network which are commonly viewed as “social”; such research may focus, for example, on how such “spaces” are structured (eg the topology of social networks); or it may focus on the nature of the trans- and interactions which arise. In all these cases, data which may be connected to a human subject may be easily obtainable and/or necessary for the conduct of the research. Or, in bot research, the evidence of a “human” subject is minimal at best, as more CS research distances the “subject” from the researcher. Instead, a bot or agent is seeking or scraping “data” and risk seems minimal. Thus, an ethics board will look at the benefits of the research more liberally, if at all, and often conclude that the research will be advantageous to more people than it could possibly hurt. This stance undermines the concept of the human in digital and virtual realms, minimizing the extent to which such automated research can affect an individual’s autonomy, privacy, consent, and basic rights.

The emergence of research ethics 2.0 challenges the long-standing process of research, questioning what Forte (2004) has described as scientific takers and native givers. Within the discourse of research ethics 2.0, the accepted principles of human subjects research are interrogated. Such pressing questions as listed below must be discussed within disciplinary specificity but also with the goal of cross-disciplinary best practices:

  • What are public spaces online and what rights do researchers and researched have in such spaces?
  • How is confidentiality, if anonymity is no longer an option, assured in such venues as MUDs, MMRPGs, and other digital worlds?
  • Are “agents” humans?
  • Can a bot research another bot ethically?
  • How–and should–informed consent be obtained (Lawson, 2004; Peden & Flashinski, 2004);
  • Is deception online a norm or a harm?
  • What are harms in an online environment?

And, ultimately, what are the ethical obligations of researchers conducting CS or Internet-enabled research and how do they fit into or diverge from extant human subjects models? Are alternative ethics review models possible, especially in light of emergent models of research, and how should they be constituted

By examining specific cases of CS and Internet-based research, this paper will affect a broad impact on applied ethics, which cross disciplinary boundaries; the real-world practices of researchers from a variety of disciplines; and the practices and policies of ethics boards seeking to ensure human subjects protections in novel environments and research contexts.