Open Source Software: A Case of Swift Trust?

Paul B. de Laat


The so-called `open source software movement’ enjoyed a huge increase in popularity over the last decade. The movement has multiplied in scale. This holds for both `output’ – software in source code form to be used freely by anyone, and `process’ – networks of thousands of volunteer hackers working together. Famous examples include hackers developing Linux, Apache, and OpenOffice. This development is, of course, mainly the result of the creation of the Internet; participants now have immediate access to each other, and downloading code is a matter of minutes only.

What motivates hackers to contribute? This question has been the subject of fierce debates. Both the joy of writing code and building a reputation are the best answers that have come up. However, still puzzling is that such virtual teams assemble overnight and exchange comments and/or source code for some period of time, on a massive scale at that. How is it possible for people that are complete strangers to each other, as well as separated in space, to work together so quickly without much hesitation? What processes underlie this instantaneous creation of mutual trust?

Until recently, the development of trust in (normal) teams was believed to be well understood. Trust takes time to develop, must grow step by step, requires meeting face to face regularly, and is furthered by common language and common culture. All of this seems hardly of any use regarding virtual networks of open source hackers. Therefore some authors have taken to the concept of swift trust in order to come to grips with these teams. This article will explore the concept of swift trust and the theories behind it, and try to answer the question whether the comparison brings us any further. It is a theoretical exercise, which tries to connect two separate strands of theory.

The concept of swift trust was first developed in the context of film crews, theatre groups and the like, which come together for a short time and then disband. Their work mode necessarily is fast in, fast out. There is simply no time available for the usual processes of getting to know and trust each other. In these circumstances it is a norm of professional conduct to act along the lines of swift trust. Subsequently the concept has been adapted to virtual circumstances by authors like Walther, Meyerson and Jarvenpaa.

In their theories, swift trust refers to global virtual teams that are never to meet in reality. In contrast to the classical type of trust, swift trust develops in the first moments of electronic interaction (or not at all). It is task-oriented behaviour that keeps interactions going. Showing initiative as well as responding quickly to initiatives from others are crucial, given the inherent anxieties that cyberspace creates. At the same time, virtual communication tends to hide rather than reveal cultural characteristics. As a result, culture as a binding force recedes to the background, and efforts at building a common culture become irrelevant. Virtual trust hinges upon contributing creatively and swiftly, not about showing compassion and contributing to the social cohesion of the group. As I interpret these theories, again, like in the classical case of film crews, trusting swiftly is a mode of profesional conduct. Swift trust is the trade mark of the virtual worker that takes his profession seriously.

Questions to be answered are the following. Is it warranted to characterize the process of contributing to open source groups as swift trust? Does this analogy with temporary film crews yield new insights? More generally, are the two types of trust possibly a product of varying circumstances only? Related to this, might the impossibility of face to face contact by any chance be considered as a blessing instead of a curse, i.e. virtuality as freedom from the terror of social conventions that allows focusing upon the tasks at hand? Finally, are the two types of trust, classical versus swift trust, really opposites of each other? In particular, is not the vindication c.q. refutation of expectations of trust a driving mechanism common to both?