CAPITULO II: CONTEXTO GEOLÓGICO
3.7 Análisis estructural de las fallas en el Complejo Volcánico Chucapaca-SW
Diego Gambetta’s collection of works under the titleTrust(Gambetta, 1990c) gathers together the thoughts of many diverse areas, from biology to economics. Of particular interest in this section is the final chapter of that collection, by Gambetta himself,
Can we Trust Trust? In it, he presents a view of trust similar to that in the formalism presented in the following chapters. (The paper was found some way into the work documented here, but the similarities are often striking.) It is reviewed here for two main reasons:
• It presents a view of trust which, although from a different viewpoint than that presented here, echoes much of the views presented in this work as regards the workings of trust. It is thus of major interest both due to these similarities and because of its different viewpoint.
• It summarises a volume, edited by Gambetta (Gambetta, 1990) whose theme was trust. It is thus of great applicability to this work, and is a major contribution to the trust literature.
Gambetta’s view of trust, seen from the viewpoint of an amalgamation of differ- ing viewpoints, much as in the present work, is similar to the present work in many important ways. The most important similarity concerns the use of values. As dis- cussed in chapter 2, the use of explicit values for trust can be seen as something of a problem, particularly since trust is subjective enough that the same valuemay mean different levelsof trust for different agents. However, the use of values does allow us to talk succinctly and precisely about specific circumstances in trusting behaviour. In addition, it allows the straightforward implementation of a formalism. Gambetta uses values in the range 0 to 1. In other words, trust is a probability, which he defines as follows: “trust (or, symmetrically, distrust) 13 is a particular level of the subjective
probability with which an agent assesses that another agent or group of agents will perform a particular action, bothbeforehe can monitor such action (or independently of his capacity ever to be able to monitor it 14)and in a context in which it affectshis
own action.” (Gambetta, 1990a, page 217).
As the final part of this definition suggests, trust is a means of coping with the freedom of others, and how this affects us (Luhmann, 1979; Dunn, 1990). In other words, trusting a person means that the truster takes a chance that the trustee will not behave in a way that is damaging to the truster, given that choice.
Gambetta’s definition excludes certain aspects which are of importance to trust, however, in that clearly it refers only to trusting relationships between agents, not, for example, between agent and environment. It also excludes those agents whose actions have no affect on the decision of the truster, despite trust being present. The first circumstance is more problematic since we do have some form of trust in the natural order of things (Barber, 1983; Luhmann, 1979). However, from the point of view of the present work at least, this trust is simply implicitly assumed — that walls don’t move, or more practically that if an inanimate object is moved, it stays where it is put, all other things being equal. The second exclusion, of those whose actions do not influence the decision, is less of a problem, since, as trust is based in the possible future actions of those the truster perceives will be able to affect him, having trust in someone who is not perceived to be able to affect the truster does not affect the decision that is made. In other words, an agent may or may not trust his bank manager to look after his money, but this has little effect on the decision to trust his office mate to post a letter for him. Different situations, such as the decision to invest money, for example, would naturally concentrate on the agent’s trust in his bank manager.
Of course, there is always a risk involved in trusting (Luhmann, 1979; Luhmann, 1990), and thus people tend to try to remove the need to trust by establishing pre- commitments, or constraints, on the truster and the trustee. In a discussion closely related to those presented later in this work on legal aspects of trust, Gambetta states that pre-commitments establish that we as trusters can also be trusted, thus we make promises, sign contracts, and so forth. Ulysses, when he tied himself to the mast of his ship, showed a lack of self-trust, but combatted this by committing himself. Because he could not trust his sailors, he made sure they could not hear the Sirens’ songs or his
13As Luhmann notes, trust and distrust are not merely opposites, but are functional equivalents of
each other (Luhmann, 1979; Luhmann, 1990).
orders to take him closer to the Sirens (Elster, 1979). Such a form of pre-commitment is usually applied towards others, so that they may rely on, or trust, us. On a weaker note, contracts and promises are made, not physically binding, but costly to renege upon.
An interesting point in Gambetta’s paper concerns competition. It is accepted (for example, in economics, concerning monopolies) that there are times at which we would say that cooperation was not a good thing, and would wish to discourage it. One example is monopoly control in the business sector, another is the control of criminals — a cooperating society of robbers is quite undesirable, where cooperating police forces are unquestionably useful in this context. It seems, then, that we should try to find “the optimal mixture of cooperation and competition rather than decid- ing at which extreme to converge.” (Gambetta, 1990a, page 215). With regard to competition, even then, cooperation is of great importance, since, “Even to compete, in a mutually non-destructive way, one needs at some level totrustone’s competitors to comply with certain rules.” (ibid., page 215). As Gambetta notes, “there is a difference between outdoing one’s rivals and doing them in” (ibid.), and inter-species rivalry is considerably more inclined towards the former.
Despite the important insight of using values for trust, Gambetta does not develop the idea in any concrete fashion. He does mention contracts and how the law and legal aspects will lessen the need to rely on trust (pages 221–222; see also chapter 8, section 8.5 in this work). In addition, the concept of a threshold is mentioned at several points, and closely resembles the cooperation threshold discussed in chapter 4, often with significant insight: “We may have to trust blindly, not because we do not or do not want to know how untrustworthy others are, but simply because the alternatives are worse.” (page 223). This is a direct similarity to Deutsch’s view of trust as despair (see earlier in this chapter). Deutsch ignores this aspect of trust, but Gambetta includes it not as a separate aspect, rather as a part of the whole which is necessary in order to get the ‘whole’ picture of trust. In the present work, trust is seen as somewhere between the two — clearly, trust as confidence (to use Deutsch’s definition) is a major aspect of proper functional trust, but the other, less important aspects allow an agent to behave in a reasonable manner even when rationality fails. It is this which trust as despair actually allows, and this is why Gambetta refers to it in such a way.
The final part of Gambetta’s paper concerns whether or not trust is a rational and sensible option, i.e., whether we can trust trust, and correspondingly distrust distrust — that “it can be rewarding to behaveas if we trusted even in unpromising situations” (page 228). So that, given trusting behaviour, others will learn to cooperate, much as in Axelrod’s Prisoners’ Dilemma tournaments, where Tit for Tat actually encouraged cooperation. And even Tit for Tat needed at first to behave as if it trusted, so that it cooperated on the first move. There are some conclusions to present here:
• “I cannot will myself to believe that X is my friend, I can only believe that he is.” (page 231). In other words, trust cannot be brought about at will. Indeed, the statement ‘trust me’ does not work unless trust is present in the first place (Boon & Holmes, 1991).
• “if X detects instrumentality behind my manifestations of friendship, he is more likely to reject me and, if anything, trust me even less” (page 231). This is of particular importance in the formalism to be presented, and is similar to Deutsch’s idea of thefocus of intention(see earlier, also Deutsch, 1973). If the workings of the formalism are known, deceit can become problematic. Bearing this in mind will help armour the trusting agent against such actions.
• Trust is of use in situations of ignorance. However, the seeking of evidence in situations often affects the evidence itself (Boon & Holmes, 1991). Thus, “While it is never that difficult to find evidence of untrustworthy behaviour, it is virtually impossible to prove its mirror image.” (Gambetta, 1990, page 233). In other words, once distrust has set in, it is particularly difficult to know if such distrust is justified since such experiments will not be carried out. Trust is capable of spiralling dramatically downwards (Golembiewski & McConkie, 1975). Conversely, however, it is capable of spiralling upwards, and being self- reinforcing (ibid.).
Perhaps most importantly from the point of view of this work, and multi-agent systems, Gambetta states that “sustained distrust can lead only to further distrust. Trust, even if always misplaced, can never do worse than that, and the expectation that it might do at least marginally better is therefore plausible.” In other words, a knowledge of the workings and usefulness of trust can help artificial agents get along better in the unpredictable world they exist within.