Legal Theory lecture
Ronald Stamper and Norm based systems

Author: Dr Robert N Moles

Stamper criticises the lack of semantics in the logic programming approach to legal expert systems. [1]

He suggests that many people working in this area work with paradigms which suggest an objective reality, which can be understood by human intelligence and which in turn can be replaced by machinery. He has in mind the sort of work which we have examined in the logic programming section. The alternative which he proposes is that of a subjective product of human beings, who are trying to collaborate by shared norms and experiences. He believes that whilst this may be computer assisted it cannot and should not be replaced by computers.

Stamper takes the view that there are metaphysical metaphors, which are hindering our progress and give rise to a series of errors in our thinking.

The conduit metaphor of language . This presumes that words and sentences are vehicles or containers for knowledge, through which meanings and concepts can be carried from one place to another. So we can bottle knowledge with an independent existence and sell it contained in software to the user. The problems with this are that:

1. It obscures responsibility.

2. It pretends that knowledge can be detached from its social context.

3. It assumes that signs carry this expert substance and the human process of interpreting signs is not essential to the nature of the knowledge represented.

The solutions which Stamper says we must direct our thinking to are:

1. To ask who provides the knowledge, and what agendas that person or persons have.

2. To be suspicious if we cannot examine the knowledge base as a whole.

3. To apply normal lawyer's skills to the possibility of conflicting interpretations.

The chemical engineering metaphor of data-processing assumes the objective reality of information.

input--------process-----------output,

whereas it should be

signs-------encode-------process------decode--------signs.

[interpreter..................................................interpreter]

For law, if we are to have expert systems, we need a system which will handle the situations that do not support the illusion of “data flow”. Since information is seen as an inherently subjective entity which will have different interpretations according to the circumstances.

Stamper is pointing to the fact that the computer programmers are often using a formal language, under the assumption that it can capture the richness of a natural language. They assume a set metaphor of reality, where everything can be related by pairs and into higher order structures, and that predicate logic adequately accounts for reality.

As Stamper says:

"The role of the law is to establish boundaries and maintain them, even to allow them to move gradually in a controlled manner. To assume...that all kinds of boundaries are fixed, and fixed in an objective way, independently of any human agency, is to evade the central issues with which the law is concerned."

In the early days, we had high technical skill but poor results - we did not understand what "information" really was. A theory of signs can transform our understanding of information systems. [2] The classical methodology shares the view that the world is an objective reality.

Professionals in information processing think that they can process the messages, leaving the concern for the purpose and meaning to the user. Semiotics (which is the theory of signs) leads us to a very different view. Stamper says that the starting point is to begin by admitting that “objective reality” is only a convenient fiction for a limited range of problems, and that data does not “flow” into databases and onwards to the users through application software.

Instead, we regard reality as subjective and constructed by users within their informal, culture-based information systems. The technical can only supplement the real social information system by supporting those who display behaviour we want the organisation to display. We see how users can create their own systems as though they were writing the laws or norms that should govern their organisation. This is why he describes his system as "norm oriented". People do not mechanically follow programs or conform to social norms. The idea of “responsibility” is the key idea here – and this has nothing to do with computers. However, it is essential to an organisation. We cannot expect to turn a set of norms into a computer based system, but we can create a system to support the responsible agents in society. The normbase, says Stamper, is a by-product of semiotics and a big step beyond the database.

The theory of signs means that we must have a semiotic framework. In semiotics, the theorist Peirce focussed on the more logical aspects – which we call "semiology". The theorist Saussure focussed on the more social and linguistic aspects. The ideas of a sign, says Stamper is the "primitive" (or basic) notion. The other aspects which we need to understand are:
Syntactics - which deals with structure
Semantics - which deals with meaning
Pragmatics - with deals with use
There is also physics and empirics as well.

The most recent advances in semiotics make it clear that we cannot account for the properties of signs without recognising the social dimension in which they find their purposes.

Physical - every sign has a physical aspect to it – it also has something to which it refers – it also has a person or group whose sign it is to interpret it. This is what Stamper calls “the representational triangle”. The physical part of the sign might be a signal which is dynamic, or it might just be a static mark. The model use is that of “tokens” (such as money). But signs will have sources, destinations, routes for transmission or circulation. They may also be affected by their rate of deterioration or speed of transmission.

Empirics – this has more to do with the use of the sign. Here we look less to the physical aspects, and concern ourselves only with their probability of occurring in a stream of messages. An ergodic source is one which behaves in a statistically uniform manner over a period of time. This is where one considers the matching of the signal to an appropriate media for its transmission.

Syntactics – this is where one has regard to the form and structure of the message, rather than the form of what Stamper calls the “terminal tokens”. Certain symbolic forms can be transformed - essentially in a mechanical manner by computers as we saw earlier. However, communication and meaning depend on the means, by which these transformations take place. In the discussion on logic programming, we saw the many forms of transformations which were taking place, and which were assumed by those responsible for them to be insignificant to the end users.

Communication means that the output of one process becomes the input of another.

The Technical Platform includes the physics, empirics and syntactics which supply the technical platform for information systems. Stamper makes it clear that in this area of information systems, all problems can be solved without regard to what the tokens represent in the real world. He says that we are in these three domains indifferent to any human values or intentions, associated with responsible agents. [As were the logic programmers]

However, it is the next three areas which are vitally concerned with this human aspect of information processing.

Semantics - are essentially concerned with meaning. In Stamper’s view, many writers prefer to reduce the semantics to a problem within the technical platform, just as we have illustrated in logic programming debate.  These people see the world as made up of discrete and individual items with mappings from formulae. Stamper says that this may not be unreasonable, where the boundaries reflect a well established consensus. However, they clearly fail when that consensus breaks down. When opinions differ and negotiations are needed to draw boundaries, as is essentially the case with law, then we have to treat the information system as much as a means of creating the reality as a means of describing it.

Stamper, like Searle, is concerned with the objectivist principle, which assumes that meanings are mapped from the syntactic structures to an objective reality which is the same for everyone

The constructivist principle, which he like Searle supports, is that meanings are constructed and continually tested and repaired. When speaking of mentalistic – or conceptual models, he says that the objectivist principle suits only certain forms of routine administration. However, only the constructivist principle is of use where conflict and negotiation arise, as is the case with law. He says that we may have an utterance – a significatum – and an interpreter. He says that one significatum, becomes the utterance for the next stage, and “is relative to the interpreter”. His reason for emphasising this is that in information systems, the organisation does not want to have the interpretation of the computer scientist, but that of the responsible person within the organisation, who is charged with, or has delegated to them, that particular responsibility.

At the level of pragmatics, if a sign is to be of use, it must always have an intention imputed to it by its creator and also by its interpreter. This area concerns the relationships between signs and the responsible agents in a social context. Context is essential, says Stamper, for signs are used for action, and often have little syntax when taken out of context. Illocutionary verbs reveal the intention of the speaker and require some intentional response.

As part of the context, we need to know the potential changes which communication of the sign might provoke. Whether we are designing signs around road works, or for modern missile systems, if the message is inadvertently changed through the process of communication, the consequences could be very serious.

At the social level, the social world consists of “norms” of many kinds (like those of law) which define the shape of social reality for us. Stamper says that in his view of things, “Information”, is seen by him more as "giving form" to something, as a potter informs the clay. He says rather interestingly that:

"Information at the social level is perhaps best understood as a process of imparting form to a social situation."

The social level focuses our attention more on the perlocutionary effects rather than on the illocutionary or intended effects. “Without the token fitting into a resonant social context, it could not function fully as a sign.” A sign is meaningful if it actually alters those otherwise existing norm structures. Communication is a reflection of the generation of community or shared norms - a communion or sharing. The concept of a norm is not vague, but one that can be operationalised. Information, communication and meaning are not simple concepts, but clusters of concepts. The information systems discipline (just like the legal discipline) needs to tidy up its thinking about such basic concepts.

Just as we have seen earlier with Searle and the others we have discussed, when one has regard to the “paradigm” or “frame of reference” which we employ, we can see that the frame of reference determines what you see and what remains invisible. [4] Stamper says that it is a common view that an information system takes raw data and converts it into "information". What he calls the “information plumbing” metaphor, has no room for the people who give meaning and intention to the signs, nor about the relationships between people which are created sustained and exploited through signs. However, he says that technical questions are secondary to organisational needs.

Information systems are social systems and the new power which we have to create and handle different kinds of signs, may be as important as the development of writing itself. However, in and of itself, it does no more to create intelligence that did the creation of writing. The formal system must be correctly located within the necessarily more complex social system if it is to be of use.

He recognises that we must look at the importance of the informal and tacit knowledge which operates within any system. In fact, he says that it is this informal system that matters most. [5] The formalisation of tasks creates a bureaucratic system whereby people can work without a precise understanding of meaning and purpose. It is a more mechanical way of transmitting and transforming sign tokens. De-skilling is an essential step towards automation. If machines can read signs without understanding them, and re-arrange them according to rules, then we can automate the formal part of a system. The largest area of knowledge then is the INFORMAL which contains the FORMAL which in turn contains the aspect of knowledge which can be formalised by our information technology.

The technology of writing relieves the load on the memory and liberates the intellect to perform other tasks. In an oral culture, social memory depends on repetitive behaviour of the group. It is the concept of the “norm” which for Stamper provides the link between the formal and the informal areas of our knowledge. Telematic systems, he suggests, can embody norms. Expressed more formally it would look like this:

if  then is permitted / forbidden / obliged to do

This formalisation tells us which agents must be provided with the information specified in the condition. The consequences of the act have to be communicated to those that have to respond. It is this system of norms together with an understanding of its “responsible agents” which allows us to map the communication of an organisation.

Information field theory tells us that to write down a rule is only a sign to represent a norm which is a field of force that tends to make members of community behave in certain way. Stamper suggests that by calling it an “information field” it will make us more sensitive to the subtleties of meaning. It is better than the idea of “data flow” with its implicit assumption that messages have a fixed meaning and are derived from an objective reality. Each field is a system of norms. Stamper acknowledges that there may be tensions between norm fields. "Words and other signs will have different meanings in different fields." [6] Any single organisation will have overlapping fields. The problem he suggests is how can we represent norms and yet not ignore their informality? We can write an explicit rule - but the meaning will depend upon the interpretation by a person in the light of their tacit knowledge of how to use the language in which the rule is written. [7]

NORMA - A Logic of Norms and Affordances [8]

It is clear, says Stamper, that there can be no knowledge without an agent, and he refers to Wittgenstein and his discussion of "language games" to support this view. [9] Essentially this is similar to the discussion by Searle we looked at earlier.

There can be no knowledge without a knower - knowledge depends on what the knower does. 
There can be no reality without an agent - the agent constructs reality through action
Stamper is totally opposed to the objectivism of the natural sciences and the Platonism of mathematics. [10]

If we wanted to express this in a formalism the formula should be:
<knower-term> <behaviour-term>
<agent-term>  <action-term>

This would tie every item of knowledge to an agent who is responsible for it. Truth is something which agents and group agents have to decide upon and the consequences for which they have to accept responsibility. [11] Essentially this means that the institutional facts of a group can only be created, supported and maintained by that group. 

The idea of “responsibility” says Stamper, plays same role as truth in classical logic. TRUTH here, is not a primitive (or basic) concept but a derived one, and is explained in terms of AGREEMENT AMONG AGENTS.

Within a changing world, certain thoughts - feelings - actions are possibilities, which are allowed or not allowed by invariants. Gibson, in psychology, says that they afford or make possible some repertoire of behaviour. [12] Each aspect of behaviour “affords” or allows for some possibilities, but not others. Jogging affords motion, but not reading. Norms are for Stamper, the social equivalent of physical affordances. NORMA - is the logic of norms and affordances.

Words do not refer to an external and objective reality, but to sub-routines of behavioural or perceptual ability. When we use a word, the listener is disposed to act or perceive or otherwise behave in accordance with the repertoire of affordances that it labels. Whether the authority concerned is a judgment of an individual or of a group, mediated by a norm, it is a logically necessary component of any affordance. Even when judging a physical fact, a judgment must be made. Everyone depends on a socially constructed view of the world.

Ontological dependency is the dependency of one type of behaviour on another

The factors A X Y - might stand for John - Water – Swim. It follows that the invariant cannot be realised without its antecedent - therefore it has to work within its temporal constraint plus or minus the time at which it begins and ends. There may also be group agents and an agent with many parts or roles. A role is a “function set” or a repertoire of behaviour which may be performed by an agent. With these factors in mind, Stamper suggests that one could draw up an “ontology chart” of organisational behaviour.

Composite Realisations involve the appreciation that there can be restriction, conflation or negation of affordances. However, they are to be seen as a joining of behaviour patterns, not propositions. Stamper says that it is important to appreciate that one cannot have a unary negation of action. The role of time is important. However, there may be joint ontological antecedents:

A(x,y)z   Society (person, nation) citizenship –

Citizenship requires joint antecedents, just as the terms "on" or "marriage" would.

Stamper argues that in systems design, no more than 2 antecedents need be permitted, as more complex relations can be made up by a process of binary accumulation. [13]

However, because of this, we will need to develop what he calls “related affordances”. A way of indicating beginning and ending:

Ax<  John sleep beginning > = ending. “Birth” and “marriage”, he suggests, are both words for beginnings although of different types. One can, of course, relate behavioural invariants with the beginning and ending of others:

begin < end > sustain >> prevent < >

When we develop our understanding of “semiological affordances”, we can see that they are different ways of using signs - 

Metonymy – is where we use part of a situation (which is present) to stand for the whole which is past, future or out of sight. A footprint in the sand would be an example of this.

Metaphor – is where something present stands for another entity, because of similarity in some respect. A portrait might stand for a person.

Signs are products of social norms and signs may be seen as tokens and types. A person utters signs and engages in communication acts -
illocutionary – refers to the communication
perlocutionary – refers to the act on communication
Any affordance with an utterance as its antecedent will be an illocution. Every illocution presupposes that an utterance is made.

Truth is a complex semiological notion concerned with the relationship between direct knowledge of the world and propositions spoken about the world. When a community reaches a consensus about the correspondence between a proposition and the direct experience of the aspects of the world which the proposition represents, then, as a shorthand expression we call the proposition true.

To know anything about a reality that is not immediately present to us, we need to use signs. Time is therefore, a key semiological construct and will require us to symbolise this is some way, such as the use of "before" and "after".

Norms, mechanisms and responsibility :
Norms are not true or false. They are not like the material implications of classical logic. The existence of a social norm is not related to the existence of its components. Modus ponens is only a weak counterpart. A norm may yield an expectation that the agent will comply. In the use of social (or legal) norms, one has to use judgment before applying them to particular cases. [14] This is a point which was developed in Definition and Rule in Legal Theory.  

"The last thing we want to do in the real world of practical affairs is to treat laws with the mechanical certainty of an unconstrained  deductive system assuming a closed world governed by predicate logic. We must have a place for the agents who take responsibility for formulating the facts about an infinitely open reality and deciding which norms are relevant to them."

When we are utilising written laws as signs representing norms, we need to represent them simply and clearly, taking the forms evolved in natural language as the guide. Mathematics is misleading as a source of inspiration for the understanding of social norms. There is nothing to link a mathematical model to the outside world.

NORMA, the system developed by Stamper, assumes some agent (or group) must determine all starts and end events. We can automate aspects of it, but we cannot run it autonomously. We need an agent therefore to supply the knowledge of the informal norms that cannot be expressed in the system. The formal system can only be an over-simplified version of the real social system.
Stamper is quite clear that we cannot provide mechanical surrogates for agents based on probabilities, as a number of systems attempt to do. NORMA - by requiring the agent to be specified, makes the embedding of the formal system within the informal system more explicit.

Determiners: Individuality - agent realising one or more instances of an affordance

Discreetness: several individuals at the same time. I and D are variable factors. But individuality may also incorporate uniqueness - done symbolically by use of name - hash indicates determinate affordance. A determiner is a universal - when weighted or valued, it is a determinant.

Generic and specific hierarchies -

Practical consequences: as team or community held together by norms, we can structure this -

MEASUR -[this is the name of Stamper's specific methodology]

The first stage is Problem Articulation : this involves a negotiated statement of problems - indexed hypertext to set out the group's understanding of the problem.

The second stage is Semantic Analysis : the clarification of concepts in the light of the problem to be solved.

The third stage is Norm Analysis : this involves mapping the dynamics of the organisation in the form of social norms.

Problem articulation

This is sometimes called 'soft systems analysis' - it involves

Valuation framing - understanding the purpose of innovation and interest of groups involved –

Functional morphology - understanding the mores of the groups and their respect for norms –

Collateral analysis - this covers the infra structure needed to begin maintain and end the task and to relate it to the environment. This stage ensures that the whole problem is considered and related to the organisation's informal context.

Semantic analysis

This involves the development of an ontology chart - it asks the “problem owners” what they mean by the terms they use. It might be that the "Learned society" is the ontological antecedent, and the "conference" an ontological dependent. Start and finish - the actual meeting is only part of the work to be done. Part to whole signified by dot

learned society----------- . ----------------conference -------------- . -------------meeting

The dynamics of the organisation is sometimes determined by rules and sometimes by an official exercising discretion. The ontological structure is the framework on which we can hang the norms and responsibilities. Ambiguity and differences of meaning are removed by analysis and negotiation. Each element of the schema can be related to particular instances which will have the same relationships as the antecedents - start finish etc.

One test which can be applied to semantic schemas is, what happens when the antecedent ceases to exist? Semantic temporal database (std) can go beyond conventional database management such as that manipulated by LEGOL.

The Normbase

This strips away the business-specific knowledge, and holds it in the normbase as a business resource. From this specification, a prototype supporting the information system, automated where necessary, can be generated by the normbase. It uses a graphical form of the ontology chart where possible. The normbase can be linked to established databases.

Applying NORMA to semiotics

Semantic analysis can be used as general tool to analyse complex conceptual structures - it is a knowledge elicitation method. Once we supply the normbase with a semantic schema in the form of a chart – it uses the normbase as a CASE tool generator. This will include universals, particulars with start and finish times, and authorities, and provide the essential record of the designer's work. We have to look at all of the six aspects of signs. The information technology practitioner’s focus has often been on the technical questions involved, and this has often been disastrous for the organisation. Similar terms may appear on more than one level with different meanings. The analyses in the charts should be considered to be a series of hypotheses. A different schema will manifest a genuine theoretical difference, which needs to be worked through and resolved.

Physics of signs

Some would say that there is a high level of inter-subjective agreement in dealing with the concepts of physics. In Stamper’s modelling, social concepts are inherent, because he does not allow any knowledge to be detached from “the responsible agent” and the responsibility for the commonsense concepts is shared by the culture that created and sustains them. If you want precise technical definitions, then you have to have the appropriate group of experts as the authority in such cases. We can go into levels of detail which may be appropriate for some purposes, but not for others. Such detail may be useful for detecting malfunction, but not for design analysis.

Empirics of signs

Channel and source abstractions, as we are not interested in their physical characteristics.

Semantics of signs

How the pattern types relate to the world.

Different linguistic communities

Interpretation depends upon the relationship which the language community establishes between the signs and reality which they represent. Word meaning is no more than a relationship between a sign and a disposition to act in a certain way. Information systems practice often has nothing to say about meaning. NORMA is more explicit about this, and where we cannot be explicit about it, we can be explicit about who is responsible for giving words their meanings. It is also an attempt to make the semantics of a complex expression depend on the semantics of its components and the way they are combined. It makes the syntax into a metaphor for the structure of behaviour patterns, so we build a NORMA expression in the same way as we build complex behaviour from simple behaviour. We should be able to interpret NORMA expressions, directly as a guide to action. 

Pragmatics of signs

The intentional use of signs as governed by conversational norms. Communication only gets anything done when received. This may be on the part of the addressee, but it may also be any other hearer. The functional grammars look to their intended effect. We can handle “speech acts” quite easily in NORMA – but it is not appropriate to go into this in detail here.

Social effect of signs

Each intentional speech act may lead to changes in the social world, which essentially involves the relationship of perlocutionary acts to illocutionary acts. A well-formed chain may lead to important changes, such as the making of a contract.

Conclusion

In some ways it seems that the Stamper approach recognises the social and linguistic complexity inherent in the idea of community. Something which the logic programmers appear to have ignored. At the same time, Stamper recognises the limitations of the tools he is working with and identifies a more limited, but potentially more valuable role in which they can be engaged in social activities.

[1] Ronald Stamper “Expert systems: Lawyers beware!” in Law, Decision-Making and Microcomputers, S Nagel (ed). Quorum Books, New York, 1991. The following section is taken from this article.

[2] The discussion in this section is taken from, Ronald Stamper, Signs Organisations, Norms and Information Systems, Keynote Address at the Australian National Information Systems Conference October 1992.

[3] By "primitive" Stamper means the most basic or fundamental aspect we can deal with. The smallest part or that which cannot in turn be subdivided.

[4] Remember the story about Geoffrey Bamber which we discussed earlier.

[5] This aspect is also emphasised by Michael Polanyi in his book Personal Knowledge.

[6] Judges and barristers are particularly good at changing the "norm field" or interpretive context so as to change the meaning of a piece of legislation, as we have already seen.

[7[ And I would add, of the context in which it is to be applied.

[8] NORMA is the name of the program which Stamper is developing.

[9] It is interesting to note that Hart, in his rule-based theory of law also appeals to Wittgenstein for support. It is my view that Hart failed to understand that Wittgenstein's views were in fact contrary to Hart's - so I think Stamper is correct. This view of Stamper's and the comments which follow have much in common with my own comments regarding the remarks of Collingwood which we referred to earlier. It is also very much in accordance with the views expressed by Aquinas.

[10] Although Morris Kline, Mathematics The Loss of Certainty, is a good account of how this Platonic view of mathematics has eventually come unstuck.

[11] It is this view which fundamentally distinguishes Stamper's work from that of the ICG which we discussed in the preceding section.

[12] Gibson is a major figure in psychology, and Stamper is here attempting to incorporate these psychological insights into his formalism.

[13] The attraction of binary approaches is that they are more easily modelled on computers which itself comes down to binary operations as we explained in the preceding chapter.

[14] This is precisely why I have argued that Hart's claims regarding the nature of law must be wrong. He claims that legal norms (or rules) may be applied, by and large, without the need to exercise any further judgment. See Definition and Rule in Legal Theory.

Top of Page