Abstract: Ever since the creation of the first human society, people have understood that the only way of sustaining and improving their societies is to rely on each other for exchanging services. This reliance have traditionally built on developing, among them, {\em trust}, a vague, intuitive to a large extend and hard to define concept that brought together people who worked towards the progress we all witness around us today. % Today's society is, however, becoming increasingly massive, collective, and complex and includes not only people, but huge numbers of machines as well. It is no overstatement to say that machines interconnected together into a complex communication fabric as well as communicating directly with people will very soon outnumber people by several orders of magnitude. Thus, trust, being already a difficult concept to define and measure when applied to a few people that form a cooperating group or a set of acquaintances, it is far more difficult to pinpoint when applied to large communities whose members may hardly know each other in person or to interconnected machines employed by these communities. In this paper we attempt to take a pragmatic position with regard to trust definition and measurement. We employ several formalisms, into each of which we define a reasonable notion of trust, and show that inherent weaknesses of these formalisms result in an inability to have a concrete and fully measurable trust concept. % We then argue that trust in the modern intertwined WWW society must, necessarily, incorporate to some degree non-formalizable elements, such as common sense and intuition. Although this may sound pessimistic, our view is that it is not, since by understanding these limitations of formalism with regard to trust, increases self-awareness and caution on people's part and avoid problems that result if one relies only on automation in order to deduce trust.
