Title: Informed Consent in the Study of Online Communities: A Reflection on the Effects of Computer-Mediated Social Research Author: Elizabeth Reid During 1992 and 1993 I worked towards a Master's thesis offering a cultural study of MUDs, a form of text-based virtual environment system found on the Internet and most often used for entertainment purposes. My research took the form of participant observation: to the intense envy of my friends in the Computer Science department at the University of Melbourne I spent two years playing MUD. The ethical considerations of a study of a living social system were naturally an issue for both myself and my supervisors. I took what I believed were sufficient measures to ensure that my research was done in an ethical and responsible manner. Nevertheless, my research has been a contributing factor adversely affecting at least one of the communities I studied. This essay reflects upon the reasons for those consequences and suggests some lessons to be learned from them. In devising an ethical code to fit the circumstances of the MUD environment I was faced with several problems in deciding what material I could ethically access, what material I could ethically retain copies or records of, and whether and of whom permission should be asked before including material in my thesis. My raw data was drawn from three sources: postings to MUD-related Usenet newsgroups, email sent to me by MUD users, and MUD sessions. Access to the first two categories of information was easily justified. In the case of Usenet articles the author could not reasonably expect to exclude any person from gaining access to his or her words, even if any particular individual were not specifically envisioned as part of the audience. In the case of email anything sent to me was obviously intended to be read by me. The ethics of my access to these two forms of information seemed clear-cut and unproblematic; the ethics of my use of this material for research purposes was less clear. Although Usenet postings may be universally accessible it is doubtful whether each author intends his or her words to be placed in the public domain. In the absence of clearly defined legal or cultural specifications regarding the use of material distributed via Usenet I felt it best to take the more cautious -- and courteous -- path of asking each author's individual permission to include their material in my thesis. In the case of email it seemed clear that since I was the only intended recipient I should seek permission before including such material. Emailed and Usenet-distributed material thus presented me with few ethical dilemmas, and in fact in some lights I may have erred on the side of caution in deciding that I should seek authorial permission before utilising Usenet articles. Deciding how best to collect and treat MUD session material posed more complex problems. Material generated by users of a MUD system is not universally accessible. To gain access to the MUD one must apply to the system's administrators for a character; to gain access to conversations between MUD users one must gain their implied or explicit acceptance as a fellow member of the system. Granted, in the case of most MUDs it is not difficult to gain a character or gain acceptance into the group; nevertheless it does not follow that easy access to material implies that it lies within the public domain. If this was assumed to be so in the case of Usenet I felt I had to more vigilant in my use of the relatively more private and protected material produced on a MUD. Accordingly, before referring to individuals or quoting from conversations and environment descriptions in my thesis I asked for permission from those concerned to do so. I had also to decide whether I was ethically bound to disclose my presence as a researcher to the MUD users I was interacting with. The initial problem I faced in deciding this was the changing nature of the population of any given MUD. It was not practically possible for me to inform all members of a MUD of my research interests without disrupting the normal social flow of each system, since the fluctuating member base meant such announcements would have to be prominently made every few days. This meant that if I were to carry some out my research some measure of deception, or non-disclosure, was inevitable. A related concern was the dramatic difficulty of letting other MUD users know that I was keeping log files of my MUD sessions. My solution was a virtual approximation of a visible tape-recorder: my MUD characters' personal descriptions included mention of a tape-recorder, notebook or other device suitable to the particular milieu of the MUD. Moreover, since the ability to log sessions was an integral feature of the client software commonly used to connect to MUD systems I felt that I was not behaving unethically in logging sessions. Although it may have been possible that some individual MUD users might object to my presence in principle, I felt that since I was not taking any actions unavailable to all MUD users without explicit permission from the users concerned those actions were not in themselves harmful or objectionable. I felt that this pragmatic, perhaps even utilitarian, approach satisfied my obligations to the individuals who made up each MUD userbase. How to act on my ethical obligation to the MUD as a whole and as a community was more difficult to decide. I decided to contact the administrators of each MUD I used, tell them about my research, assure them that I would ask permission of each individual to quote material, and ask for permission to discuss their MUD in my thesis. This decision involved an assumption that a MUD's administrator was in a position to speak for the MUD community as a whole. At the time this seemed a sensible assumption, and one that seemed to be shared by both the administrators and the users. Subsequent events have made me question some the assumptions behind these decisions on how to act, but at the time they appeared sufficient both to myself and to my subjects. My initial comfort with my chosen ethical guidelines was reinforced by responses from my subjects. Not once was I refused permission by either MUD players or administrators to include them in my work. My initial concern that many people might not allow me permission to quote or mention them in my thesis turned out to be unfounded. On the contrary, in the vast majority of cases people seemed not only willing to participate but actively to delight in the prospect. In all cases I promised to make sure my subjects were able to read the finished product, but only rarely did I have to offer to do so --more often the people I wished to quote would take the initiative in asking me to let them know how to get hold of a copy of the thesis once it was completed. I was also subject to the phenomenon of people who on learning the nature of my research set about to deliberately manufacture quotable quotes, something which strengthened my conviction that non-disclosure of my research activity was both justified and necessary (I will refrain from a detailed discussion of the sub-phenomenon of MUD users who would kindly offer to give me any experience in the art of tinysex which I might deem useful to my research -- a novel pick-up line if ever I heard one). In requesting permission to quote them I gave my subjects the option of being referred to anonymously. Initially I had included in my requests the assurance that all identifying information would be excluded from the thesis. However, I later altered this to offering the option of anonymity since many of my subjects requested that I cite their real names, email addresses or MUD character names. In several cases I was given permission to quote only on the condition that I cite the subjects' email address and real name. With hindsight I have become ambivalent about this, though at the time it seemed the more ethical path to accede to the wishes of my subjects in this respect. My subjects' enthusiasm about the prospect of being quoted initially seemed to me an adequate and indeed extremely clear indication of informed consent. In retrospect I think that assumption should be qualified. In at least one case my research has had a negative effect on its subjects. During 1993 I had several online conversation with the administrator of a MUD designed as a meeting place and support centre for survivors of sexual abuse. I met this administrator, who has since asked me to refer to her by the pseudonym 'Jenny', on an unrelated MUD. On learning the nature of my research she offered to tell me about an incident on her MUD. A user of JennyMUSH, as she asked me to refer to the MUD, had staged an elaborate and protracted attack on other players, using textual forms to mimic sexual abuse. The victims of this attack had turned the tables on the intruder, and taken violent and cathartic simulated revenge. This incident was fascinating to me on two levels. Firstly, it demonstrated the deeply visceral nature of MUD simulations of violence and power. Secondly, the social and administrative changes seen on the MUD in the aftermath of this incident lent themselves to a discussion of hierarchies of control in online environments. I very much wanted to include a discussion of JennyMUSH in my thesis, and Jenny herself seemed equally keen to have the incident written about. The MUD's users seemed to share this enthusiasm. Jenny announced my interest on the MUD's bulletin board. Some discussion ensued, but none of it indicated a reluctance to take part in my study. The players who participated in the discussion seemed to welcome the chance to talk about this violent chapter in the MUD's history. Four users sent me logs of the incident in question; seven others sent me highly personal reflections upon it. For the first time since I had begun my research project I found myself discomforted by the eagerness with which MUD users seemed to embrace the opportunity to be a research subject. Where I had been amused and occasionally embarrassed by the revelations of tinysex enthusiasts, I was now disturbed by the degree of self-revelation being shown to me by the users of JennyMUSH. I had been asked not to reveal the location of the MUD and to refer to it and it's users by pseudonyms, but these were the only restrictions placed upon me. Had I chosen to I could, with the full consent of those involved, have published extracts from email and MUD session logs which revealed deeply personal information about these people's lives and experiences. I chose not to do so, and though I did write about the MUD I did not quote from the logs or email sent to me by the users. It was not an intellectual discussion of research ethics which decided me against this. It was a purely personal decision, based on my own sense of discomfort about the possibly prurient nature of publishing such details as were sent to me. In retrospect I am very glad that I acted as I did. After making my thesis available on the Internet I received -- and still do occasionally receive -- many items of email from people interested in JennyMUSH. Most of these requests came from people who expressed an interest in joining the MUD for personal reasons, but a significant number came from a variety of social scientists who wished to conduct further studies of the MUD. Initially I forwarded each of these requests to Jenny. After three months she asked me not to forward any more, but instead to tell all enquirers that the MUD was now closed to new members and that existing members were not interested in being studied. The MUD had reached a crisis point. Where a feeling of safety and privacy had reigned there now existed distrust and wariness. Users were connecting to the MUD with declining frequency and the social networks and alliances that had flourished on it showed signs of strain. In order to protect and consolidate this small virtual society Jenny had decided to batten down the hatches and increase security on the MUD. While I do not believe that participation in my research was the sole factor in this threat of social disintegration, it had contributed to it. The violent incident which had sparked my interest had itself damaged many of the threads of trust and support that held the MUD together, and had been the catalyst for a tightening of security. Nevertheless, Jenny and I both came to the conclusion that this uncomfortable stage in the MUD's development may have been quickly overcome had it not been for the decision to invite further public scrutiny and personal exposure through participation in my research project. In the wake of this news about JennyMUSH I began to question my decisions on how to measure informed consent among online subjects. In particular I began to doubt the wisdom of taking enthusiasm for my project to indicate both knowledge and acceptance of the risks which participation in it might entail. The often commented upon disinhibiting effect of computer-mediated communication could conceivably lead people to agreeing to or even insisting on a kind of public exposure by which they may eventually be harmed. The allure of self-revelation can be strong in any circumstances. The popularity of confessional talk shows and literature testifies to both the exhibitionistic and voyeuristic streaks that lie in many psyches. In online environments where consequences to the actual lives of participants can be hidden behind the illusion of a virtual/actual dichotomy this tendency toward uninhibited behaviour can both make the social researcher's job seemingly easier and thereby place an added burden of responsibility on his or her shoulders. I do not think that researchers should necessarily always carry the onus of responsibility to second-guess their subjects' reasons for consent. Such an expectation of paternalistic empathy would be unreasonable. Nevertheless, where particular circumstances may influence a subject's decision to consent to involvement it is doubly important for researchers to inform subjects of the consequences of such agreement. In a similar vein, the researcher needs to be doubly watchful for tendencies to objectify or depersonalise his or her subjects. That same disinhibition which may cause subjects to flock towards the chance to be immortalised in research can also cause the researcher of online communities to distance him or herself from the humanity of subjects and to accept at convenient face-value their willingness to reveal themselves to an unknown audience. For instance, I now believe that I should have insisted on anonymising all quotations, even if this meant that in some cases I would be refused permission to quote. Particularly in cases where revelations of a personal nature were made I should have taken pains to outline to my subjects the possible future consequences of such material being published, and told them that anonymisation was one of the precautions I would take as part of a policy of responsible research. In all social research involving human subjects there is the potential for harm to come to those subjects. Guidelines for ethical behaviour can never eliminate that potential; all they can do is attempt to minimise the risk. It would be disingenuous of researchers to pretend that such risks was non-existent. We must face our moral and ethical dilemmas rather than side-stepping them. As researchers we can never be sure that our actions won't have adverse consequences for our subjects. Try as we might to act with consideration and responsibility we can never eliminate all risk. No matter what measures we may take to protect our subjects from the onslaught of outside interest the experience of scrutiny inherent in being involved in a research project may itself be damaging. Quite aside from the moral consideration of tampering with the very system one is studying, at some point we must all ask ourselves what kinds of risk are acceptable, and to what extent the importance of knowledge does or does not outweigh the potential personal discomfort of the subject. In choosing to conduct social research we have all presumably decided that the value of knowledge is at least sometimes greater than the value of personal safety. Ethical considerations are best weighed in an atmosphere of personal honesty and no matter how ethical and considerate we may feel ourselves to be as individuals, as researchers we must admit and negotiate the potential consequences of our desire to pursue and create knowledge. To deny the existence of ethical pitfalls in social research is irresponsible; to decide that since they exist they are unavoidable is equally so. If it is a genuine interest in and wish to understand our subjects which drives our research, then our skill as researchers will only suffer if by denying or glossing over the potential for harm to our human subjects we distance ourselves from either the consequences of the research process or the needs of our subjects. While keeping in mind that adverse consequences can result despite even the best and most well-meaning of efforts, we must not allow this to excuse us from the obligation to make those efforts. In the particular circumstance of studying online interactions we must take especial care to guard against not only any increased tendency to objectify our subjects, but also the tendency for our subjects to underestimate the potential consequences of consent. The criteria for informed consent which may be sufficient in face-to-face research environments are not necessarily enough in a medium in which subjective experience is easily objectified, and information is easily devalued. It is not simply a question of researchers doing their best to respect subjects' sensibilities and wishes -- on the Internet, where perceptions of interpersonal affect and effects may be obscured by the nature of the medium itself, we must be very careful to ensure that our subjects are aware of the potential consequences of agreeing to participate in our studies. In my own case I don't believe that I necessarily acted badly. I certainly didn't act with disregard for the effects on JennyMUSH's population. Mostly importantly, I decided against taking advantage of my privileged position as a computer-mediated researcher in order to benefit from the tendencies toward self-revelation amongst my subjects. I can be thankful that by doing this I averted what might have been a more damaging outcome for JennyMUSH. Nevertheless, I believe that to some extent I allowed my own sense of computer-mediated detachment from the personal reality of the people behind the MUD characters and communities I was studying to influence some of my decisions on how much self-revelation on the part of those people was acceptable. As it is, the MUD survived the crisis -- had it not, I doubt that I would have been given permission to mention it once more in this article. It has, however, been changed by the experience. Jenny no longer advertises the MUD in various mailing lists and Usenet newsgroups. New members are now attracted only through discreet word of mouth, and must be sponsored and vouched for by an existing member. Internal constraints on the building and programming abilities of members have been instituted. Members are no longer able to use the command which allows messages to be sent simultaneously to all others logged on at any given time. These measures have made the MUD safer and less open. As Jenny herself wrote in the only direct quote I included in my thesis, events have forced the members of the system to trade "freedom to" for "freedom from" -- a trade-off I regret having been a factor in necessitating. I hope in future that I and other researchers will take care to consider the especial factors involved in undertaking social research of online communities in order to minimise the potential harm which may come to the human subjects involved.