Dear Colleagues: Please forgive this sketchy draft. I look forward to your comments and reactions which no doubt will contribute to development and improvement. Sincerely, HFN TWO CONCEPTIONS OF TRUST ONLINE Helen Nissenbaum Department of Culture and Communication New York University Paper prepared for the conference on Trust and Honesty Budapest Collegium December 12-13, 2002 Introduction This paper reflects on the subject of trust online and on trust in general to the extent the more limited domain yields insights into the general. As a mere fragment, the online context serves only as a partial microcosm, yet, as an instance of a particular kind of phenomenon – a natural phenomenon which has evolved on top of a deliberately Draft-2 3/8/16 1 constructed artifact -- it brings into relief (in more tractable form, perhaps) compelling dimensions of trust. Two, in particular, will be the focus of my paper: a) a duality in core conceptions of trust, one emphasizing the instrumental connections between trust and assurance, the other emphasizing the uncertainty, risk, and freedom inherent in trust, and b) the interdependence between technical decisions about system design and the merging dominance, online, of one or other of these conceptions. My thesis is that instrumental conceptions of, which appear dominant in the concerns over trust online, will tend to squeeze out trust-as-freedom when embodied in our networked information systems. Should this occur, something valuable will be lost as result. Part One: Where Trust is a Means One way to appreciate how trust emerges as a concern for participants in the development and critical assessment of the online environment is through clusters of scholarly, trade, and popular writings each focusing on particular aspects of online experience and respective dimensions of trust. Clusters suggested by my own informal survey of these literatures include: Computer and network security; E-commerce; Content; Interface Design. In each case, trust had been identified as an explicit end, or goal, which could and should shape their development. Brief descriptions follow. 1. Computers and network security Draft-2 3/8/16 2 The field of computer security, which is now quite vast, grew in size and scope alongside the developing fields of computing and information and communications technologies. At some point along the way, perhaps when networked information systems overtook stand alone computers as the focus of public attention, the purposes or ends of the field found expression in terms of trust. That it, it became meaningful, even entrenched convention, to refer to higher order goals of the community of engineers and scientists within the branch (practice and research) of computer security to express their mission in terms of trust. They sought to build “trusted” systems as well as “trusted” components of systems and networks. I should note that for the most part it would be more accurate to describe their purposes in terms of trustworthiness as this characterizes the state of the artifact (worthy of trust, or not), while trust or trusted, according to standard usage, is more fittingly applied to the attitude observers or users hold toward these artifacts. A definitive report, Trust in Cyberspace (1999) of the Computer Science and Telecommunication Board of the United States National Research Council undertakes to revise usage within the technical security community to match the standard. According to this same report, networked information systems (like the Internet) are trustworthy in the extent to which they feature a constellation of properties, including correctness, reliability, security, privacy, safety and survivability. In other words, trustworthiness is conceived as an aggregate, or higher-order property, of systems that manifest a constellation of other properties, presumably in some suitable degree and in some satisfactory distribution. Each of the properties listed, though labeled in ordinary terms and grounded in ordinary meanings have, over time, developed a quite specific, Draft-2 3/8/16 3 technical character in the context of of computer security. (Privacy may be an exception.) Security, for example, here conceived as a component of trustworthiness, itself refers to a cluster of properties, usually including secrecy, confidentiality, integrity and availability of data, integrity and availability of computer systems, and integrity and availability of networks. Safety refers to a system’s or network’s capacity to buffer users from harm or injury even when systems fail in some way; survivability (or robustness) to the capacity of a system or network to function even when parts of it have been attacked or disabled; and so on. Survivability has been of particular concern to experts who have keenly observed the effects on the whole of a recent spate of attacks on key components, such as Domain Name Servers, and routers. (References) Trustworthiness, and its aggregate dimensions, applies not only to whole networks but to component parts. Thus, we may speak, for example, of trustworthiness (or reliability, robustness, etc.) of networks themselves, of network components, such as routers, servers, etc., of elements of the physical infrastructure, or even of individual computers, or hosts, within a network. To attempt more than a superficial survey of the components, or dimensions, of trustworthiness in networked information systems would take us well beyond the scope of this paper. Each component is itself the subject of large bodies of work, annual conferences, government sponsored projects, and the sole focus of innumerable engineers and technical experts working in industry. Furthermore, the interplay among the dimensions, is itself complex and of significant interest; some dimensions of trustworthiness are correlated, but not all are. In promoting some, we might need to trade Draft-2 3/8/16 4 off others because of the possibility of interference, prohibitive cost, or technical infeasibility. Nevertheless, even setting aside the task of substantive elaboration, it is worth noting that research and development focused on trustworthiness devolves to the task of developing security mechanisms, each dedicated to some small part, or some small goal, developed out of a need defined by the overarching goal of trustworthiness. Later in the paper, I will discuss some of these mechanisms. ((Our distance from trustworthiness is sometimes expressed in terms of degree of risk, that is, as a function of how far away we are in meeting goals implied by these various dimensions of system trustworthiness. Note: an ambiguity can creep in here, as in other contexts as trustworthiness need not imply perfection, it might simply mean that it makes sense to trust, it is rational to trust, even though there’s a non-negligible risk of harm.)) A different way to characterize the quest for trustworthiness is in terms of threats. A networked information system (like the Internet) is trustworthy to the extent it can withstand, deflect, of minimize the impacts of breaches of the system, which can take on a variety of forms from technical failure (or breakdown), incompetence of designers and operators, and a surprisingly large array of malicious attacks. In the face of these multifarious threats, trustworthy systems are those that remain available and true to function; data both in storage and in transit is protected against undesired exposure and corruption. As with clean air, a condition that mostly went unremarked until threatened Draft-2 3/8/16 5 by growing pollution, so trustworthiness of our networked information systems is appreciated in the face of increasing threats of failure and attack. 2. Electronic Commerce Another domain where trust features prominently is electronic commerce (or, ecommerce). Ecommerce, here, is taken to cover a variety of commercial exchanges transacted via the Internet, including those where buyers place orders online for delivery of physical goods, (Peets.com, Amazon.com, Ebay.com, etc.), where the goods delivered are intangible (e.g. information, subscriptions to online publications), where what is purchased is a membership of one kind or another (e.g. a pornographic site, a dating service, etc.), and where a website provides services to customers (e.g. online bill payment, online gambling, participation in game playing). These transactions could occur between individuals and businesses or between businesses. Proponents of Internet commerce had seen promise of great advantage to both consumers and businesses. Despite the more muted hopes and expectations following failure of many dot.com initiatives, many remain sanguine. Even the most optimistic, however, stress the importance of trust to success, to the realization of promise. (For example, Resnick, Zeckhauser, Friedman, and Kuwabara; Jones, Wilikens, Morris, Masera; Hoffman, Novak, and Peralta) Consider a few of the junctures where trust is crucial. Take a typical example of an individual ordering physical goods from a merchant, who, for argument’s sake, does Draft-2 3/8/16 6 business only online. The consumer must trust the merchant to charge the correct price, provide goods of satisfactory quality, and deliver as promised. Merchants must trust customers to pay and not repudiate their commitments. Consumers and merchants must trust banks and credit card companies to convey funds and they, in turn, must trust consumers and merchants. All parties must trust the competence and integrity of vendors of ecommerce systems -- that the systems perform their functions reliably and safely, and, on the minds of concerned consumers, have the capacity to securely transmit and store critical sensitive information like credit card numbers, personal information about consumers, including the content of their orders. Finally, trust in ecommerce, presumes a degree of trust on the Internet, generally, as discussed earlier. One could go through a similar analysis of all the modes of ecommerce transactions. Many issues raised in the context of ecommerce face parties in traditional commerce as well, such as trusting a merchant to sell high quality goods, or not to overcharge credit cards, though we are likely to be more familiar and comfortable with policies and strategies to deal with risks and failures. Some, however, like those involving trustworthiness of the underlying protocols and systems are unique. And others are novel because they involve entities or transactions that deviates somewhat from the familiar. Although auctions are not new, for example, online auctions such as those conducted through eBay, are different in detail in ways that might make a difference in degree of trust. Purchasing travel reservations online from companies we may not previously have known, raises similar concerns of sheer novelty. For example, downloading information Draft-2 3/8/16 7 or software from a dot.com company might (or at least should) raise concerns over unwanted code that might be hidden within it. Ecommerce is undergoing rapid evolution as individuals and businesses learn from their own experiences and those of others, at the same time that legal and financial institutions adapt old and craft new policies and practices for them. One important example of adaptation was legal recognition given (through an act of the U.S. Congress) to digital signatures. We have also seen evolution in societal responses to “fringe” commerce in such areas as online gambling, pornography and other “adult” offerings, the latter constituting a large proportion of online activity. I should qualify that I use “fringe” to contrast these activities with culturally more central activity, not with size as adult content offerings have proven enormously popular. These fringe websites are of particular interest because they are facing changes attributed to trust, or rather, lack of trust. For a start, Internet gambling has been all but squelched by credit card companies refusing to honor gambling debts incurred online. These companies have also begun levying surcharges on adult-content sites, defending their actions against charges of private censorship, by pointing to higher incidence of untrustworthiness in both the operators of adult websites (who regularly engage in dishonest billing practices, like charging credit cards even after subscriptions are cancelled) and in consumers, who (perhaps out of remorse or fear) regularly deny prior associations with these sites and repudiate charges (resulting in costly “chargebacks”). (See “Credit Cards Seek New Fees on Web’s Demimonde,” by Matt Richtel and John Schwartz, The New York Times, November 18, 2003) Draft-2 3/8/16 8 3. Interface Design and Content Two other topics that have generated clusters of work on trust are interface design and content. I provide a brief overview of each. Content. The intention behind creation of the World Wide Web was to build a system of ready access to information stored in computers around the globe (Berners-Lee). The Web would take advantage of Internet connectivity but would create standards for sharing information to overcome existing barriers to free flow. Our experience of the Web and Internet reflects the enormous success of this effort, which provided access not only to established institutions, like the ones with which Berners-Lee began, but to an enormous array of individuals and collectives from multinational corporations to governments to renegade groups and individuals. The results are breathtaking in size and scope. At the same time that this freedom to share and find information, the absence of global gatekeepers, the unconstrained access to publishing leads to a vast and varied trove, it also results in content of varied quality. In referring here to quality, I do not mean to imply a single, linear, standard by which content may be measured. I am suggesting, however, that commitment to any standard or standards will lead to a set of question or worries about what content is credible and can be trusted and what is not and cannot. One can imagine this question ranging over a broad array of possibilities from announcements Draft-2 3/8/16 9 of events to claims of miraculous medical cures to world news to scientific reports to financial offers and more. Individuals struggle with these questions (though perhaps not as much as they ought to), information scientists pose them as the fundamental challenges of their profession. Interface Design. Research on website design, particularly ease of use and appeal of interface, has also considered systematic relationships between design and trust. Goals of this line of research are somewhat different from the others as they are not necessarily focused on how trustworthy systems, commercial offerings, and content are but whether users will in fact extend trust to them, whether users will perceive them as trustworthy. In other words, the questions asked by those interested in usability and interface design include how to frame information, what sort of information to provide, what aesthetic or visual factors have bearing on users’ responses to an online presence, offering, or invitation. Some of the efforts to shape user experience and reaction through website and interface design draws on findings from other social and commercial experiences. In one case, for example, on-screen, animated characters representing real-estate agents were programmed to follow interactional rituals like pleasant greetings including small-talk that referred to stories of past benevolence. Studies indicate that such techniques increase users’ trust in these systems. (Cassell and Bickmore) Or, in other cases, experts advise designers via guidelines and principles such as disclosing patterns of past performance (e.g. an airline reservation system) through testimonials from other satisfied users, Draft-2 3/8/16 10 making policies clear and salient, avoiding misleading or ambiguous language, and so forth. (Friedman, et. al., Schneiderman, B.J. Fogg, et.al.) Findings on the factors that induce trust, or distrust, in computer and Web users though related to trust online generally, in the end, is probably best handled as a separate inquiry. While the latter focuses on trustworthiness and trust, the former (study of interface design) focuses primarily on how to seem trustworthy no matter what the underlying case may be. Those who study user related design factors generally acknowledge the moral ambiguity of their craft and encourage readers to take the moral high road along with them. Achieving Trust Online Before considering responses to these challenges, it is worth noting, as should be evident from the cases above, that trust online is not a wholly new species of trust; challenges are quite similar in form to those we encounter more generally within both specialized relationships and in society at large. There is value, however, in dealing with trust online as a distinct topic of study because the online context which, in some sense, constitutes a simplified model of life in general, highlights particular aspects of trust that can be lost in the “noise.” There are, furthermore, historical and cultural dimensions of the online context that are relevant to our understanding of trust and distrust. Finally, because the online context is layered upon an artificial or constructed technological infrastructure, it Draft-2 3/8/16 11 allows for, at least in principle, trial and experimentation in ways not as readily available in the thick world of physical being. Many (though admittedly not all) of the responses to trust-related challenges posed earlier can be understood as ways of coping with several of the distinctive properties of online experience. In an earlier paper, I described obstacles to trust online, in particular, properties and conditions, absent or attenuated online, that typically form the basis for the formation and maintenance of trust attitudes and judgments. One is that we may not know the true identities of those (people and organizations) with whom we interact, a condition that interferes with the possibility of learning from experience (one’s own and others), establishing reputations, and building cooperative and reciprocal relationships. Another is that the disembodied, mediated nature of relationships online lack sensory information about personal characteristics that cue us to trustworthiness of others. Finally, contexts and opportunities online remain somewhat ill-defined, ambiguous, inscrutable; we remain unsure of meanings, responsibilities, and norms. (For more detail, see Nissenbaum.) With these features of online experience as a backdrop, it is easier to understand the rationale behind various approaches that have been taken with the purpose of achieving trust, or increasing the level of trust, online within the areas mentioned above. Approaches, which include technical, institutional, and legal, attempt in one way or another to address, or compensate for, potentially limiting dimensions of the relevant features of online experience. I don’t mean to imply the existence of a unified, organized Draft-2 3/8/16 12 attempt to address the problem of trust online. Rather, what we see are numerous efforts - some independent others building upon each other – to produce a set of mechanisms which each tackle some small element of aspect of the larger issue of trust online. In the arena of trust in computer systems and networks, for example, it has been predominantly scientists, engineers, research sponsors, and government agencies, who have focused on approaches to creating more robust systems by such means as redundancy for critical elements, mechanisms for detecting unusual patterns of activity (that might signal attacks), capacity in the system as a whole to overcome untrustworthy components, and so on. They fortify systems against attacks with layers of security: passwords, firewalls, restrictions designed into computer languages (like JAVA) to limit what programmers can and cannot accomplish (not something generally grasped or even noticed by most non-technical users), limiting access to authorized people and code. Another part of the picture has been security of network connections, or pipelines to protect information in transit against theft or alteration, to ensure greater confidence in the integrity of information flowing throughout our networks. Secure transmission is considered an essential element for trust in ecommerce as consumers will want to avoid theft or tampering of sensitive information like credit card numbers, bank transactions, and passwords. Mechanisms to restore identity have become increasingly sophisticated. So-called “trust management systems”, for example, seek to establish reliable links between parties -– individuals and organizations – the commitments they make and actions they undertake Draft-2 3/8/16 13 online. These systems typically involve unique identifiers (for example, digital signatures) for participants backed by trusted intermediaries (“signature authorities”) who vouch for the validity of the linkage. As biometric schema become cheaper and more effective, we are likely to see their use in establishing even tighter and sounder links between online presence and identity. It is worth remarking that given massive databases of personal information, which have been accumulated and aggregated over years of data surveillance, identification links parties not only with particular actions online but with rich profiles as well. Thus, identification mechanisms promise to increase our capacity not only to hold parties accountable for wrongdoing but in the first place, to predict, more reliably, what they are likely to do. Mechanisms that enable sound (accurate and reliable) identification are fundamental to many other mechanisms that build upon them to address a series of specialized problems related to trust in the various arenas discussed above. One controversial application has emerged in the context of the content industries (movies, music, fiction, art, etc.). As has been amply documented in legal disputes, scholarly literature, as well as popular news media, a highly distrustful relationship has developed between the industry and its mass audiences. (refs.) From the perspective of the industry, viable systems of Digital Rights Management (DRM), which necessarily build upon the capacity to identify users as well as content, are necessary to rebuild trust. With DRM systems they will be able to keep track of sales and usage and put a stop to peer-based unauthorized sharing. Draft-2 3/8/16 14 Another family of mechanisms that have shown promise for increasing trust online are those that support reputation systems. Although these systems do require extended, unique, identities to which reputations attach, these identities might not necessarily link to anything outside of a given context. Clearly, reputation is not distinctive of the online context, but computational and network capabilities enables a variety of options including some that would be virtually impossible to implement offline. Some reputational systems have familiar form – individuals and organizations that undertake the role of trusted authority promulgating their appraisals of others – individuals, services, businesses, information, etc. These systems may range in focus, from the most specialized, like TRUSTe, which offers seals on the basis of their privacy policies, to the most general, like search engines. Among the most novel and interesting to proponents of online experience are peer-based reputation systems not based on authority or expert appraisals but upon a computed aggregate of peer appraisals. (Zagat’s restaurant guides may be an offline equivalent.) EBay’s Feedback Forum, which allows buyers and sellers on this massively popular online auction site to rate each other following transactions. Anyone considering whether to engage in a transaction with another via eBay can check the other’s Feedback Forum reputation, though, as shown by Resnick, et. al. reputation systems are far from perfect, even ones as mature as Feedback Forum. Because reputation systems can apply to people, as well as organizations, information, political opinions, and more, they offer promise for various of the contexts (above) in which the issue of trust arises. Draft-2 3/8/16 15 Trust and Assurance Readers who have tracked developments in new media and are familiar with the broad spectrum of social, technological, and creative possibilities online will probably complain that I have omitted a set of cases that get closer to the heart of trust than any that I have mentioned so far. They would be right. But before I turn to those cases, and to the second conception of trust, let us review features common to those discussed already and how they contribute to a conception of trust as a means. In the cases described earlier and in many others bearing no connection to the online context, trust functions to close a gap of uncertainty. In these cases a person is challenged to take action or make a decision without having all the relevant information to be able to know with certainty (or a high degree of certainty) what the outcome of her or his action will be. We place orders online not entirely certain that the businesses are “real”, whether we have recourse if they take payment and fail to deliver, whether they will respond to complaints if the items we buy are damaged. We avidly absorb information about cancer treatments, availability of drugs not yet approved by the FDA, the weather in Bangladesh, a new failsafe diet. We visit pornographic and other websites, place bets, download free software and songs, enter competitions, make airplane reservations, get legal advice, open up email messages from people known and unknown to us. Most of us are aware that there is risk involved in all these activities – in some more than in others. Yet we undertake these actions (or decisions) with an attitude of trust (or, if you Draft-2 3/8/16 16 prefer, a decision to trust), deliberately placing our fates in the hands of others who have the capacity to harm us in some way. As such, trust acts as a bridge between uncertainty and action. What I would like to suggest, is that for cases of the kind that I discussed above, trust is a compromise. In an ideal world, or an ideal state-of-affairs, we would be assured that networks will not fail, that we will not be attacked by criminals and viruses online, that financial transactions will go through as intended. We would be certain that our messages go where we send them and retain confidentiality and integrity. We would be assured of reliability of information we access, of the honesty of online merchants. But, in reality, we cannot even hope for these assurances and so in order to function, we need to make calculated judgments, trusting some, distrusting others. We do not necessarily trust blindly, but will invoke the experiences and strategies cited by many philosophers and social scientists who have studied trust. (These claims skirt on deep conceptual issues which will require further development and defense.) Proponents of online resources, too, will build on insights about trust in order to engender a climate of trust and encourage trust, and therefore, participation online. This conception of trust is somewhat humbling in that, as with other instrumental goods, it grounds the value of trust other ends, leaving open the possibility that to the extent these ends can be achieved other ways, trust conceivable loses its value. There are already cases where we reject trust in favor of other modes of assurance. Such is the case at airports, for example, where (I would argue) we have no interest in trust but prefer Draft-2 3/8/16 17 perfect assurance that fellow passengers will not be able to carry out lethal plans. The same is true, in general, of so-called “high security” zones. (This point is discussed also in my earlier paper.) Under what I have called an instrumental conception of trust, the distance between high security zones and online cases described above is a measure not of moral but of economic difference. Richard Posner’s rebuttal of Charles Fried’s defense of a strong but limited right to privacy is perhaps emblematic of this instrumental view of trust. Posner rejects Fried’s claim that privacy is the necessary context for both intimacy and trust not by denying the connection between privacy and, in this instance, trust, but by denying the importance of the loss: “trust, rather than being something valued for itself and therefore missed where full information makes it unnecessary, is, I should think, merely an imperfect substitute for information.”1 Part Two: Where Trust is the Point In Part One, I reviewed a number of arenas in which trust has been explicitly acknowledged for its potential to undergird investment of one kind or another in the online promise. They shared an instrumental conception of trust. Here, I turn attention to a more difficult arena, where trust is clearly as important but tends to feature implicitly rather than as an explicit subject of systematic study. To learn more about this arena, we need to move away from the realms of Ecommerce, broadcasters, and mainstream information providers to the helter-skelter of some of the novel life-forms that the system 1 Richard Posner, The Right of Privacy, 12 GA. L. REV. 393, 408 (1978). Draft-2 3/8/16 18 of networked computers has borne. Although there is no simple way to characterize the entire range I mean to cover, certain features stand out. The activities tend to be outside the for-profit sector, tend not to be associated with corporate or established political presences, tend to involve peer-based interaction rather than classic hierarchy or authority. Though they extend across a wide variety of substantive interests from political to recreational to technical, they tend not to involve relationships of buying and selling. Examples of such activities are the myriad online communities that consolidate around grassroots cultural, gender, environmental issues and political interests facilitated by software systems that allow for the formation of such groups online. (Slashdot, ethePeople, Institute for Applied Autonomy, etc.) It includes also communities drawn together in recreational activities, MUDs (Multi-User Dungeons), MOOs (Multi-User Object-Oriented domains), collaborative art sites, peerto-peer music exchanges, instant-messaging, and more. Some revolve around the planning of civic action, the building of civic, geographically-bounded, communities, the promotion and protest of civic causes and more. Their mode of interaction (or performance) can be varied, too, including discussion, organization, game-playing, information production, protest, voting, and linking. (Note: This list needs a great deal more research and refinement.) Although it is important to understand whether and in what ways these online efforts and communities are unique, distinctive, unprecedented, it is also useful to consider ways, relevant to trust, in which they are continuous with analogous efforts and communities Draft-2 3/8/16 19 not mediated or enabled through the Internet. In particular, it is useful to highlight their connection with traditional ideals of human relations embodied in friendship, citizenship, community, camaraderie, family, and more. Trust within these relationships is not predominantly instrumental as it is in the contexts earlier discussed, but rather, in significant ways, trust is the point. Trust is an essential part of the ideals toward which relationships of these kinds strive. There are, no doubt, many heroic and legendary stories that illustrate the essential connection between trust and these relationships but for purposes of this paper, let us consider a somewhat obscure case from the world online. The case, Telegarden, described in a study of online collaborative art sites2, is typical of the fluid borders between online activity and its physical manifestations. The Telegarden website mediates interactions among a community of participants who are dedicated, primarily, to the cooperative design and maintenance of a garden. The garden itself is “real”, that is to say, it features six to seven square feet of cultivated soil, currently on exhibition in the Austrian museum, The Ars Electronic Center. The garden is cultivated by robot (with occasional help from museum staff), which is carries out commands of the Telegarden site participants. The robot can be commanded to plant seeds in specific locations, water the garden, and provide visual feedback to participants. The Telegarden website and robot, created in 1995 by a team of computer 2 I am grateful to Gaia Bernstein for sharing this case with me. It is part of a larger collaborative book project of the Information Law Institute at NYU. (Details forthcoming.) Draft-2 3/8/16 20 science and engineering faculty of the University of Southern California, is now fully controlled by participants who “visit” the garden frequently to perform maintenance, see the garden, and mingle with friends they have made on the site. Every few months, when the garden overfills with plants, the museum staff replace it with fresh soil and a new growing season is declared. The Telegarden project can be studied from a variety of perspectives, each focusing on particular dimensions: its technical features, its mode of governance, the nature of the community that has grown around it, why individuals derive a sense of enjoyment and satisfaction from it, and the aesthetic character of each completed garden. For purposes of this paper, let us focus on an aspect of the site’s governance (as embodied partly in its technical design and partly in explicit norms.) Though participants can plant only after they have served as members (which means registering by full name and email address) for a sufficient length of time, measured by the frequency with which they water the garden responsibly and post messages in the chat rooms, there are few restrictions thereafter. This freedom means that the garden is vulnerable to sabotage such as over-watering and planting seeds over seeds already planted. There is also no censorship of materials posted to the community chat room. On rare occasions, when members over-watered or posted pornographic pictures to the chat room, organizers chose not to alter the sites mechanisms to make over-watering impossible or cancel the picture uploading function. This is crucial. Organizers could have prevented damage to the garden by an alteration to system Draft-2 3/8/16 21 software but they chose not to. In so doing, and in this small way, they declared trust as a defining character of the community. One could imagine a version of Telegarden whose system restricted member-actions, making it impossible for them to harming the garden but it would be significantly different, not as interesting, not as compelling, not as exhilarating. The reason is that in the existing version of Telegarden, trust is not simply endured as an imperfect substitute for information, trust is the point. Conclusion: Significance for Design Based on a review of formal and informal studies of the online milieu, I have posited the presence of (at least) two distinct conceptions of trust. In one, trust stimulates participation and investment online by emboldening people and organizations to act, instead of freezing in the face of risk and uncertainty. In another other, trust is a defining characteristic of relationships and communities and is a part of what induces participation and contribution to them. (Neither of the conceptions is necessarily tied to the online context.) One way to tell these two conceptions apart is to consider how a typical situation in which each operates, respectively, would evolve in the ideal. In the first, the ideal of a typical situation, say a commercial transaction, is when participants are fully apprised of all necessary details and are certain that they face no danger of harm. In the second, the ideal is a firmly entrenched resolve to accord participants the freedom to act, even to harm. Victory is not certainty of a good outcome, it is unbending commitment to freedom of action. Draft-2 3/8/16 22 Under the first conception, designers and developers of the networked information infrastructures strive to build in tight restraints, identifiability, complete surveillance, and accountability -- the analog to the perfect airport security system. The important question is whether choosing this alternative will bring too harsh a glare to the range of activities where trust is essential and whether, if it does, the tradeoff is acceptable. Partial Reference List Berners-Lee, T. (with Mark Fischetti) (1999) Weaving the Web: the Original Design and Ultimate Destiny of the World Wide Web by Its Inventor. San Francisco: HarperCollins Publishers Camp, J. (2000) Trust and Risk Cambridge, Mass.: The MIT Press. Cassell, J. and Bickmore, T. (2000) “External Manifestations of Trustworthiness in the Interface” Communications of the ACM, Vol. 43, No. 12 (December 2000) 50-56 COMMISSION ON INFO. SYS. TRUSTWORTHINESS, NATIONAL RESEARCH COUNCIL, TRUST IN CYBERSPACE 1 (Fred B. Schneider, ed. 1999) Corritore, C.L., Kracher, B., Wiedenbeck S. (2001). Trust in the online environment. In M.J. Smith, G. Salvendy, D. Harris, and R.J. Koubek (Eds.), Usability Evaluation and Draft-2 3/8/16 23 Interface Design Cognitive Engineering, Intelligent Agents and Virtual Reality (pp.15481552). Mahway, NJ Erlbaum. Fogg, B.J., Soohoo, C., Danielsen, D., Marable, L., Stanford, J., & Tauber, E. (2002). How Do People Evaluate a Web Site’s Credibility? Results from a Large Study. Stanford Persuasive Technology Lab, Stanford University. Available at http://www.consumerwebwatch.org/news/report3_credibilityresearch/stanfordPTL_abstra ct.htm or http://credibility.stanford.edu/mostcredible.html. Friedman, B., Kahn, Jr. P. and Howe, D. (2000) “Trust Online,” Communications of the ACM, Vol. 43, No. 12 (December 2000) 34-40 Russell Hardin, Trustworthiness, 107 ETHICS 26, 33 (1996) Hoffman, D., Novak, P. and Peralta, M. “Building Consumer Trust Online,” Communications of the ACM 42, 4 (Apr. 1999), 80-85 Jones, S., Wilikens, P.M., Masera, M., (2000) “Trust Requirements in E-Business,” Communications of the ACM 43, 12 (Dec. 2000), 80-87 Kim, J., & Moon, J. Y. (1998). Designing towards emotional usability in customer interfaces--trustworthiness of cyber-banking system interfaces. Interacting with Computers, 10(1), 1-29. Draft-2 3/8/16 24 NIKLAS LUHMANN, Trust: A Mechanism for the Reduction of Social Complexity, in TRUST AND POWER: TWO WORKS BY NIKLAS LUHMANN 8 (photo. reprint 1988) (1979) Manz, S. (2001). The Measurement of Trust in Web Site Evaluation. http://www.swt.iao.fhg.de/pdfs/Manz2002_Trust.pdf Nissenbaum, H. (2001) “Securing Trust Online: Wisdom or Oxymoron?” Boston University Law Review, Volume 81 No. 3, 635-664 Philip Pettit, The Cunning of Trust, 24 PHIL. & PUB. AFF. 202, 204-05 (1995) Richard Posner, The Right of Privacy, 12 GA. L. REV. 393, 408 (1978). Resnick, P., Kuwabara, K., Zeckhauser, R., Friedman, E., “Reputation Systems.” Communications of the ACM 42, 12 (Dec. 2000) 45-48 Schneiderman, B. “Designing Trust into Online Experiences,” Communications of the ACM, Vol. 43, No. 12 (December 2000) 57-59 ADAM B. SELIGMAN, THE PROBLEM OF TRUST 19 (1997) (arguing that trust in systems entails confidence in a set of institutions). Draft-2 3/8/16 25 Shelat, B. & Egger, F.N. (2002). What makes people trust online gambling sites? Proceedings of Conference on Human Factors in Computing Systems CHI 2002, Extended Abstracts, pp. 852-853. New York Draft-2 3/8/16 26