Paul B. de Laat


It has become generally accepted, that trust is an important `lubricant’ of social relations, both in `real life’ and in cyberspace. However, what are the prospects for trust developing between `pure virtuals’? Recently, using some very sophisticated arguments, Philip Pettit (forthcoming) answered this question in the negative: trust on the Internet can only be fantasy. My article will be devoted to an analysis and critique of his arguments.

In Pettit (1995), the author argued for the existence of two mechanisms of trust. First, people can be seen as trustworthy because they are considered to be loyal (say a friend or colleague), virtuous (say with religious convictions) or prudent (meaning that they appreciate the benefits of establishing a long-term relationship). Next to this primary trust, a more subtle form is to be distinguished (`secondary trust’). I may trust someone, because I assume that (s)he desires my good opinion. My act of trust gives the trustee an incentive to behave well; alter is supposed to be `trust-responsive’. Here we witness, as the author phrases it, the `cunning of trust’. This secondary kind of trust might explain the existence of trust in very unlikely situations (e.g., between perfect strangers).

Then, in Pettit (forthcoming), he `applied’ this analysis to the Internet. Pure virtuals, he argued, cannot make use of the usual mechanisms of knowledge about the other: face (`bodily presence’), frame (watching the other interact with others), and file (my record of alter’s past behaviour). As a result, primary trust is impossible. Can we trust, instead, upon the mechanism of secondary trust? No, Pettit argues, we cannot. Alter will only possibly be motivated by my future display of esteem, if (s)he believes that my act of trusting reliance is motivated (as well) by my belief in his/her trustworthiness. Secondary trust has to hide, so to speak, behind primary trust. If ego is seen to play exclusively on alter’s trust-responsiveness, the mechanism will backfire, while the move is seen as straightforward opportunism. So in the case of Internet, both mechanisms of eliciting trust from the other are blocked.

My article aims to undermine this argument. I intend to show that Pettit’s analysis can be turned around and used to argue for precisely the opposite proposition: trust on the Internet is theoretically possible. My counterarguments are the following:

(1) `Pure’ virtuals do not exist. Each cybercommunication carries clues and meanings about oneself. These may look tiny but are significant nevertheless. Even more so, while the usual interpersonal clues of face-to-face communication are lacking. A decade of research into CMC has established that these clues are being blown up in significance, in a process of over-attribution; we construct so called `hyperpersonal’ cyberpartners. In addition, participants seem to manage these clues. Details of one’s social life are being circulated, even when these do not seem directly relevant to the group’s purposes. As a result, contrary to Pettit’s argument, virtuals do have clues upon which primary trust may be imputed.

(2) Empirically, there is evidence that sometimes internet interaction between unknown virtuals starts off with trust being assumed ex ante. In order to anticipate upon the anxieties of cyberspace, participants pre-invest attitudes and actions that signal high amounts of trust. This process has been explored in Jarvenpaa & Leidner (1999), and dubbed `cybertrust’ in a 2004 article of mine. Such cybertrust, it will be argued, is the analogon of `swift trust’ in real life situations, when temporary groups engage in tasks characterized by high time pressure, mutual interdependencies, and high risk (a term coined by Meyerson et alii, 1996). How is this anticipatory action to be interpreted? I would suggest that it is nothing else but a kind of secondary trust seeking as described by Pettit: cybertrust is an ex ante appeal upon one’s fellow virtuals to respond and be nice.

Why is the display of cybertrust not interpreted as ludicrous? A possible answer is, that it is not directed to a specified individual, but to a group, with fuzzy boundaries. One’s act does not say: `I trust you’, but `there is surely someone out there whom I can trust’. The virtual that takes up the invitation, by so doing identifies himself as trustworthy; trustworthiness is self-declared, not other-attributed. A cycle of trust may ensue.

(3) Once virtual communities have formed, these may construct their own mechanisms of (primary) trust building, whether by themselves, or aided by `cybermediaries’. For one thing, monetary trust in one’s partner can be created by cybermediaries that take on the escrow function. Moreover, systems of reputation tracking may be constructed. Their most famous pioneer, of course, is eBay. For each transaction, buyer and seller may rate each other; based on these scores, one’s reputation may rise or fall. Finally, groups may build collective frames of mind that guide expectations towards one another. An example coming to mind is the hacker culture, into which open source software projects are embedded.

(4) Finally, it would seem that trust building is much easier than usually assumed if competence trust is involved. Technicians exchange technicalities that gradually confirm or refute the talkers’ competencies. So the issue of competence trust sorts itself out without any social cues needed. In Pettit’s terms: from the outset, professionals assume each other to be competent (primary trust). While such an assumption is quite normal, the way is also paved for the motivating effect of secondary trust: alter feels flattered and challenged to respond. Note, that I do not (like Pettit does) collapse competence trust into confidence or reliance, in which case a trust analysis would not apply. If my analysis is correct, the Internet is just as good a setting for professionals interacting as `real life’.