2007 UbiLexIbiPoenaDesigningNormEnfo

Jump to: navigation, search

Subject Headings:


Cited By



The viability of the application of the e-Institution paradigm for obtaining overall desired behavior in open multiagent systems (MAS) lies in the possibility of bringing the norms of the institution to have an actual impact on the MAS. Institutional norms have to be implemented in the society. The paper addresses two possible views on implementing norms, the so-called regimentation of norms, and the enforcement of norms, with particular attention to this last one. Aim of the paper is to provide a theory for the understanding of the notion of enforcement and for the design of enforcement mechanisms in e-Institutions.

1 Introduction

The purpose of electronic institutions (e-Institutions) is to guarantee the overall behavior of an open multi-agent system (MAS) to exhibit desired properties without compromising the agents’ autonomy. Aiming in particular at easing interactions and enhancing trust between agents [11]. This is accomplished through norms directed to the agents taking part in the society which specify the behavior that the institution expects from the agents. As such, institutions can be seen as normative systems [1], i.e., as sets of norms.

Institutions do not have access to the internal states of the agents and hence, they cannot modify them in order to avoid any incongruence between the goals of the agents and the norms of the institutions. Therefore, the problem arises of how to let those norms have an effective influence on the activities of the agents. This is the problem of norm implementation. This issue consists of two main aspects.

First, there is an interpretation issue concerning the concepts used in the formulation of the norms in terms of the ontology used at the society level. It is well-known feature of normative codification]]s (especially in legal systems) to be “open-textured” [6] or abstract, that is, to be in need of interpretation in order for them to be translated into norms which are meaningful for the regulated society. This is what we have called the “ontological” aspect of norm implementation [4] or, to use terminology common in legal and social theory, the “constitutive” aspect [9]. For instance, an institution might require personal data to be treated according to specific procedures. The notion of “personal data” is of an abstract nature and, in order for the norms concerning the treatment of personal data to be implemented, a clear specification of what counts as personal data in the given institution should be made precise. Much attention to this issue has been dedicated by the authors in previous work (see for instance [5, 4]). The present paper will leave the problem of the interpretation of norm codifications aside.

Second, there is the issue concerning the design of appropriate “enforcement mechanisms” required to push the society toward the compliance to the norms of the institution. For instance, if personal data is not treated in accordance to the institutional regulation, the institution should trigger some kind of reaction. This broad notion of “institutional reaction” corresponds to what we call here enforcement.

The present paper focuses on this last point, aiming at discussing a theory for understanding the implementation of norms in institutions and the design of enforcement mechanisms.

The core of the enforcement implementation strategy presented in this paper is summarized in the saying “Ubi lex ibi poena” (“where there is law, there is sanction”). In other words, if norms are to be enforced, then the institution should specify and handle sanctions for every possible violation of the norms. The paper is trying to give some answers around two concrete questions surrounding the enforcement: How do institutions handle violations and specify enforcement mechanisms? And how should sanctions be designed in order to be effective for the enforcement of norms in eInstitutions?

In Section 2 we discuss different enforcement strategies (regimentation vs. reaction). The effect of these different enforcement strategies on the society are discussed in Section 3. In Section 4 we discuss what are the possible sanctions that an institution can take in a society consisting of software agents. In Section 5 we give some conclusions and areas for future work.

2 Dealing with Violations

There exists an obvious way in which the compliance to the norms of an institution can be implemented, namely by making the violation of the norms impossible. When this is the case we talk about regimentation ([ 7]): norm compliance is unavoidable, and hence, with respect to what is stated by the norms of the institution, the space of the agents’ autonomy is strongly limited. This typically happens in e-commerce: when shopping on the web, you cannot get your goods delivered before giving consent for using your credit card number for paying those goods. Regimentation guarantees the compliance of the society to the norms of the institution. However, it has been argued, for instance in [2], that violations can be functional for the society as a whole. Even stronger, if no violation can occur, if nothing can go wrong, it does not make sense any more to talk about norms at all. From the agent point of view, a regimented norm, is just a fact.

With enforcement we mean the reaction that the institution specifies to respond to a violation of its norms. Enforcement presupposes, therefore, the possibility of violation. Institutions aim at regulating the behavior of agents through norms, but it is commonplace that norms are useless if the violation of those norms is ignored (to quote the Romans again: “ubi culpa est, ibi poena subesse debet”, that is, “where there is a violation, there must be a sanction”). In other words, the enforcement of a norm by an institution requires the institution to be in the condition of recognizing the occurrence of violations of that norm in the society and to react upon them. Not surprisingly, this check-react enforcement mechanism is specified by means of more norms. Enforcement is sought through further regulating the domain, i.e., adding norms imposing checks and norms specifying reactions to the occurrences of a given violation. Regulations on tax evasion are a typical example in this sense: tax payment is impossible to be regimented but checks, which could detect possible violations, are made obligatory. Once the detection takes place, precise reactions are also specified and made obligatory.

On the basis of these considerations, we can isolate three types of norms involved in the specification of institutions. In fact, the whole statute of an institution could be analyzed in terms of sets of norms of these types. There is a set of substantive norms which consists of those norms which describe the society’s behavior desired by the institution, and there is a set of enforcement norms consisting of norms regulating checks and reactions on violations of other norms.

The following is an example inspired by the domain concerning the policies for data protection followed by the Spanish National Transplant Organization in the organ allocation process [10].

Example 1. (Types of norms for the specification of institutions)

Substantive norm
The National Transplant Organization is not allowed to use racial data for allocating organs to patients”.
Check norm
“The inspecting authority should perform random checks of the compliance to the previous norm every two months...”.
Reaction norm
“If racial data are used in the allocation process, then the hospital has to be fined accordingly.”

The enforcement activity can thus be split in two sub-activities: check and reaction. Check norms deserve some further comments. They specify the way the institution is supposed to perceive the occurrence of violations. Needless to say, this can happen in many different ways. Either directly, via random checks, like in the above example; or via constant monitoring activity, like a referee in a sport match. Or indirectly, allowing agents to denounce the occurrence of a violation and then verifying their claim. This last checking activity is of an intrinsically more complex nature, calling for the establishment of tribunal-like sub-institutions within the main institution. It would be appropriate, in this case, to talk about check sub-institutions rather than check norms. For the present paper, we leave these complexities aside focusing rather on direct forms of checks.

Via such a normatively specified enforcement of the substantive norms, the enforcement issue is just lifted up to the set of enforcement norms because, if not regimented, those norms could be violated and be thus in need of enforcement. In principle, this pattern could be endlessly iterated unless there exists a final enforcement level, whose norms are all regimented, or whose violations are not punished (see Figure 1).




 AuthorvolumeDate ValuetitletypejournaltitleUrldoinoteyear
2007 UbiLexIbiPoenaDesigningNormEnfoDavide Grossi
Huib Aldewereld
Frank Dignum
Ubi Lex, Ibi Poena: Designing Norm Enforcement in E-institutions10.1007/978-3-540-74459-7_72007
AuthorDavide Grossi +, Huib Aldewereld + and Frank Dignum +
doi10.1007/978-3-540-74459-7_7 +
titleUbi Lex, Ibi Poena: Designing Norm Enforcement in E-institutions +
year2007 +