A challenging focus on trust emerges, if we reserve the term for acts of dynamic, interactive, and trusting reliance upon the other (Pettit 1995). The risk-taking involved may be based upon two kinds of assumptions. On the one hand, A may suppose B to be trustworthy on the basis of some concrete evidence. On the other, A may suppose B to be trust-responsive: B is sensitive to acquiring A's esteem. In that case, A just acts from the presumption that B can be trusted, reasoning that B is forwarded an incentive to react favourably.

This remarkable mechanism of secondary trust formation deserves more attention. In De Laat (2005) I tried to show that it does play a role in forging relations in cyberspace. In this article I want to explore some wider implications. First, it will be shown that of late several authors have taken up the same theme. Secondly, it is argued that the mechanism of secondary trust opens up new and easy avenues for virtual trust to be generated. Cyberspace may function as a virtual breeding ground for trust. Thirdly, I will discuss whether software agents and human agents can be trusted in a similar sense. My answer is that the mechanism of secondary trust marks a decisive difference between the two.

The secondary trust mechanism has been touched upon of late by several other authors. Debra Meyerson and coworkers coined the notion of `swift trust', which bears a close resemblance to secondary trust (Meyerson et al. 1999). The authors argued that `real life' temporary groups facing high time pressure, tight interdependencies, and high risk tend to take trust for granted at the outset; trust is assumed, not inferred. Thereupon, in a similar vein, Sirkka Jarvenpaa and Dorothy Leidner (1999) researched the formation of trust within teams of students, this time cooperating exclusively in cyberspace. Also here, some signs of swift trust formation were observed. So the phenomenon of swift trust had been put on the agenda, of both `real life' and `virtual life' research. As far as cyberspace is concerned, these clues were explicitly taken up later by authors like Wallace (1999) and Weckert (2005a). Moreover, the same Pettit applied his argument about secondary trust (from 1995) to cyberspace, concluding that virtual trust is an illusion - whether of the secondary kind or not (2004). Thereupon, in De Laat (2005) that argument was turned around: when primary trust is lacking in `virtual life' (which is often the case), secondary trust may step in, precisely because of the peculiar characteristics of virtual communication.

This reconstruction omits one important forerunner. Also Niklas Luhmann, namely, has touched upon the phenomenon (1979). His analysis allows a more fundamental interpretation of secondary trust. Roughly stated, the swift trust literature presents the decision to assume trust as a gamble, without any apparent deeper foundation. Pettit's analysis is a step forward: the trustor plays upon the trustee's assumed sensibility for earning esteem. By taking up Luhmann's earlier work this interpretation may be deepened still: the secondary trust mechanism acquires its force not only from the trustee being esteem sensitive, but also from the trustor being so. The mechanism is essentially a symmetrical exchange of esteem between participants, esteem being the reward for an act of trusting reliance. These roots in the human psyche will turn out to be important below, when human and software agents are compared (issue 3).

In cyberspace, empirically, acting on the presumption of trust does seem to happen. In person-to-person relationships, in tasks groups and the like, at times participants do seem to take chances and decide to trust without any positive clues about trustworthiness (cf. De Laat 2005). I want to emphasize two points. On the one hand, cyberspace unfolds an additional space for trusting relations to develop, potentially connecting more people than ever possible in real space. On the other, virtuality may serve not as an obstacle, but as conducive to trust. Behind the screen, as disembodied presences, participants may more easily focus on essentials of the interaction. Thus, trust on the Internet may emerge between more people and in more circumstances than we ever considered likely. Pettit (1995) argued that trust-responsiveness may account for the ubiquity and creativity of trust in civil society. By analogy, my argument can be rephrased as follows: based on the mechanism of secondary trust, ubiquitous and creative trust also applies to cyberspace, and even more so than to real space. While in the final analysis virtual trust and real trust cannot be separated, it is trust tout court between humans that is furthered.

This argument is explored in an analysis of blogging networks like MySpaces.com, Hyves.com and Livejournal.com. Amidst endless pages of self-indulging prose, nevertheless instances can be found of participants exposing themselves honestly and intimately to fellow participants (who reside all across public space). Without much of a presentation of a profile, they jump in medias res (e.g., concerning mental or medical problems). In some cases that surely represents a quest for public fame, not for private comfort. Apart from that, however, my conjecture would be that for a considerable range of interactions a presumption of trustworthiness is the basic mode of operation. Without the secondary trust mechanism, virtual exchanges would hardly emerge, let alone continue.

Currently, many people (still) consider the Internet to be a civilized society, warranting the assumption of trust-responsiveness to a large degree. However, this is a delicate balance. As soon as harassment and intrusion over the Internet become commonplace, this perception will be shattered. Accordingly, the Internet will come to be viewed by most as an uncivilized society, where basic trust is a naivety to be avoided. Where we are now on this continuum, is a matter of empirical enquiry.

Can software agents be trusted in the same way as human agents? That is the question treated by Weckert (2005b). His answer, as rooted in the concept of autonomy, is a positive one. Based on the analysis above, my answer is a different one. Bots may be trusted indeed, if we have evidence to consider them worthy of our trust (primary trust). So far so good. However, if we lack any such evidence, no reasonable human would decide to trust a bot just like that. Whenever a `software doctor' intrudes upon our screen offering to repair our `defective software', we feel annoyed, and rightly so. The mechanism of secondary trust simply cannot come into play. This is so, while it is rooted upon the notion of mutual esteem being exchanged. It is what we are or will possibly be in the eyes of the other that counts. In that respect, software agents will never be able to deliver.

A possible counterargument might be, that no difference between humans and nonhumans may be noticeable on the Internet. My answer is: so much the worse. If this distinction is not explicitly noticeable and/or signalled, the option of trusting in the secondary sense is severely hampered. The new breeding ground for trust (cf. (2) above) is severely restricted. Seen from this angle, the current web practice of ascertaining whether a human or a bot is actually involved at the other side of an exchange (by, e.g., asking to read and reproduce a string of letters and numbers) may have deeper significance than only averting possible damage. The very foundations of virtual interaction in general are thereby protected.

REFERENCES

De Laat, P.B. (2005) Trusting virtual trust. Ethics and Information Technology, 7: 167-180.

Jarvenpaa, S.L. and Leidner, D.E. (1999) Communication and Trust in Global Virtual Teams. Organization Science, 10(6): 791-815.

Luhmann, N. (1979) Trust and power. John Wiley, Chicester, etc. Translation of Vertrauen, 1968, and Macht, 1975.

Pettit, Ph. (1995) The Cunning of Trust. Philosophy and Public Affairs, 24(3): 202-225.

Pettit, Ph. (2004) Trust, Reliance and the Internet. Analyse und Kritik, 26: 108-121.

Wallace, P. (1999) The Psychology of the Internet. Cambridge University Press, Cambridge.

Weckert, J. (2005a) Trust in Cyberspace, in The impact of the internet on our moral lives (ed. R.J. Cavalier), State University of New York Press, New York: 95-117.

Weckert, J. (2005b) Trusting agents, in Proceedings of the 6th International Conference of Computer Ethics: Philosophical Enquiry (eds. P. Brey, F. Grodzinsky, L. Introna), CEPTES, University of Twente, the Netherlands: 407-411.