Symposium Description Etiquette for Human-Computer Work [The following is the initial proposal for this symposium. It represents an introduction to the intent and mindset of the organizing committee and, hence, a preface to the papers accepted in this volume.] As computers become more and more integrated into all aspects of our daily lives, as they become smarter and more capable, and as we begin to allow them to take on autonomous or semi-autonomous control of critical aspects of our lives and society, it becomes more and more important to define roles for both the human and the computer part of that relationship that each side can live with. The techniques and technology for building such systems are, we claim, already developed-at least to a first order approximation. What's missing is a clear, workable notion of how we will live with such systems. The rules that govern such relationships are, we claim, etiquette rules. By 'etiquette', we mean the defined roles and acceptable behaviors and interaction moves of each participant in a common 'social' setting-that is, one that involves more than one intelligent agent. Etiquette rules create an informal contract between participants in a social interaction, allowing expectations to be formed and used about the behavior of other parties, and defining what counts as good behavior. Each of us encounters and uses a host of different etiquettes in our daily lives. Etiquette in human-human relations is not just about which fork to use, it is also about who speaks first when ordering a meal, how long a pause in speech or action must be before an onlooker concludes the actor has missed a cue, whether or not one can swear in a poker game vs. a church, who's in charge of deploying the landing gear in flight, etc. We establish etiquettes to smooth and make more efficient and comfortable our interaction with others in virtually all social settings, yet etiquette may be may be more or less effective for the goals of those who use it-it frequently facilitates communications, but can sometimes obstruct them. Etiquette expectations may be violated (in which case, the etiquette that one party is using is not in keeping with that expected by others)--sometimes to potent and useful effect, but more frequently resulting in confusion and/or frustration. Concerns about human-computer etiquette are not exactly synonymous with User Interface generation-since we're saying precious little about the 'look' part of look and feel, and etiquette, we argue, can and does exist in a wide range of interface implementations (R2D2 had a more effective work etiquette than HAL, though HAL won out on the natural language front, for example). Etiquette concerns are not synonymous with human-centered design approaches, which are a much broader category of which etiquette would be a subset. Etiquette is, or should be, an aspect of the design of mixed initiative systems, automation, adaptive automation and adaptive information management systems, etc.-the part that has to do with, very specifically, how human and automation need to, and want to, interact for effective and acceptable relationships. So, methods of generating correct levels of trust enters into it, human performance and user acceptance do too, as does everything we've learned about adaptive system design and architectures. Reeves and Nass [1] have shown us that media will tend to be regarded with the same set of expectations and assumptions that we use to interact with human actors. This is, perhaps, more true to the degree that the media exhibit dynamic, adaptive, autonomous action. And yet, there are any number of styles of interaction (i.e., etiquettes) we use with other humans-some of them are pleasant, some annoying ... and some are more conducive to effective work than others. While we may want a computer to occupy any of these roles or support any of these styles of interaction in various contexts, the goal of effective, safe and reliable work is certainly one of the most important. Thus, as we move toward more complex, adaptive, intelligent and ubiquitous computing systems, the question of how to create 'etiquettes' that will support productive work becomes more and more important. The focus of this symposium will be on etiquette perspective in human-computer work-how to generate etiquette and embody it in systems, how to measure it in human interaction with computers, what types are effective or ineffective in what domains, etc. As such, this symposium has the potential to bridge very divergent research populations who are interested in (mostly) the same thing-people from the Human Factors and automation community (e.g., Parasuraman, Scerbo, Hancock, Woods, Smith, Sartre, Endsley, Shalin, etc.) who are interested in building real world systems in which adaptive automation or information plays a role and yet human control and safety are at the forefront, and people from the computer science world who have traditionally built desktop systems (like Maybury, Nass, Cassell, Puerta, Hammond, Maes, Shneiderman, Szekley, Langley, Lieberman, Jameson, Horvitz, etc.) where adaptiveness and aiding are equally important but ease of use, attractiveness, and even entertainment value are higher and the costs of error are lower. We will welcome inputs from traditional interface designers, those engaged in intelligent and adaptive user interface design, those engaged in research on personified interfaces or embodied agents. Researchers in Natural Language, Computer Supported Cooperative Work, Ubiquitous Computing and Machine Learning may provide important insights and/or capabilities. Psychologists and Sociologists who have insights as to the development of trust, effective team work and methods for studying and manipulating etiquette in social settings will be very welcome as well. Symposium Structure This symposium will seek to further define the notion of etiquette for human-computer relationships, to identify its similarities and differences to etiquette in human-human relationships, and to provide a variety of examples of etiquettes that do and don't contribute to human-computer work. Since we expect and hope to attract researchers from a variety of divergent backgrounds, we will begin the symposium with a set of invited, short summaries of the various contributing fields by specified field practitioners (either the organizing committee members or their designees). We will attempt to stimulate thinking about human-computer etiquette by an invited presentation from an 'etiquette expert'-either a social psychologist or sociologist or, perhaps, an etiquette maven. Substantial time will be devoted to the presentation of prior or current work from the various disciplines with an emphasis on its relevance to the notion of etiquette. At least one panel/discussion session will focus on defining 'etiquette' and arguing for and against this perspective on human-computer interface and interaction design, one will involve brainstorming about etiquette failures in current human-computer or human-automation systems and creative thinking about potential alternatives or fixes, and one will involve practitioners from various disciplines describing the approach they might take (and the issues and concerns they might have) in designing an appropriate etiquette for a human-computer interaction for a challenge problem. Organizing Committee Christopher A. Miller (chair) Smart Information Flow Technologies 1272 Raymond Ave. St. Paul, MN 55108 Phone: 612-716-4015 Fax: 651-645-0246 cmiller@SIFTech.com Justine Cassell MIT Media Lab E15-315 20 Ames Street Cambridge, MA 02139 justine@media.mit.edu Phone: 617-253-4899 Fax: 617-253-6215 http://www.media.mit.edu/~justine Clifford Nass Center for the Study of Language and Information Stanford University McClatchy Hall Stanford, CA 94305-2050 Phone: 650-723-5499 Fax: 650-725-2472 nass@leland.stanford.edu Raja Parasuraman Cognitive Science Laboratory The Catholic University of America Washington DC 20064 Phone: 202-319-5825 Fax: 202-319-4456 parasuraman@cua.edu