The Emperor\'s New Expert System

June 15, 2017 | Autor: Philip Leith | Categoria: Law, expert System, Modern Law
Share Embed


Descrição do Produto

128

THE MODERN LAW REVIEW

[Vol. 50

right of journalists to deceive or intrude in pursuit of “news” is no less indefensible, if its regulation is more controversial. Bok’s discussion of this subject (Chap. XVI) is not entirely satisfactory. (I do not, for instance, accept that there is an “undoubted dialectical relationship” between the claims to “privacy” and to “knowledge” (p.256). Bok appears to have relaxed her rigour here. Nor do I think she has adequately distinguished between the means of obtaining information and its publication.) But this is to carp. As an exercise in practical ethics, Secrets is a distinguished successor to Lying. It is an impressive work which unravels the moral complexity of social political behaviour at their most unyielding, and which exposes and challenges some of the assumptions upon which our kind of society is built. RAYMOND WACKS. *

CORRESPONDENCE THEEMPEROR’S NEWEXPERT SYSTEM Introduction

IN his recent article in this Journal (Susskind, 1986), Richard Susskind swathes himself in an Emperor’s new clothes, intent upon bringing “jurisprudential rigour to the process of building expert systems in law.” Briefly, his argument is that there has been much misguided research in the field of legal expert systems due to the lack of clear understanding and application of the philosophy of law. There are, he seems to suggest, no researchers who have given such philosophical conceptions sufficient study (“the projects constitute marginal contributions to, rather than exploitations of, the wealth of jurisprudential resources”). He provides us with a definition of what an expert system in law actually is (i.e. transparent, heuristic and flexible) and then proposes that there are none existing which live up to this definition: “However, although there are several claims of existing expert systems in law, close examination of the documentation reveals these pronouncements to be exaggerated.” After substantiation of this position, Susskind then moves on to suggest that he is in a position to correct the malaise. He will bring the power of jurisprudence towards sorting out just what it is that we need to put into legal expert systems: “It has been naive to suppose, as we shall see in the next section, that computer scientists could talk unobjectionably and unassailably of issues such as representing legal knowledge and legal inference procedures. These are highly complex matters of jurisprudence that require the attention of workers of that field.”

Leaving aside my confusion over how philosophers of law can themselves talk “unobjectionably and unassailably” when none of them ever seem to agree on issues such as “representing legal knowledge and legal inference,” Susskind then moves on to the “Grand Plan”-the research which: * Faculty of Law, Univcrsity of Hong Kong.

JAN.

19871

129

CORRESPONDENCE

“will require decades of attention from the skilful exponents in the field. It is the object of our interdisciplinary research project, currently being conducted

at the Programming Research Group of the University of Oxford

. . .”

I shall usurp the role of the child who points to the Emperor’s nakedness and argue that Susskind is doing nothing novel by looking at legal expert systems from his prescribed jurisprudential angle. I shall further suggest that his grand plan offers nothing worthwhile nor new; the Emperor’s old clothes (left at home but ready for use) are simply those of the legal formalist which jurisprudence has, time and time again, discarded. The Novelty of Susskind’s Claim Before I argue against the novelty of Susskind’s claim to be bringing a fresh start to the research area, I shall point to the technique he uses to justify his claim. It is one which Dolby (Dolby, 1982) has called the strategy of constructing intellectual distance. At its most basic, this strategy is one of suggesting that there is a marked discrepancy in the intellectual understanding which one side of an argument has about the area of expertise being argued over (thus in this case Susskind suggests that no one yet has had the relevant expertise to use jurisprudence properly within the area of legal expert systems). Dolby’s illustration of the strategy is the debate between Wilberforce and Huxley over Darwin’s evolutionary theory: “Huxley sought to distinguish expert discussion of evolution as a biological theory from popular discussion of evolution as a challenge to accepted religious views. He sought to show that although Darwin’s theory might have religious implications, it must first be evaluated as a contribution to biology, and in this decision, those, like Bishop Wilberforce, who could not show the relevant expertise, revealed themselves as incompetent judges.”’ Susskind seeks to distinguish expert discussion of legal expert systems “properly so called,” from the computer scientist’s discussion of legal expert systems. H e seeks to show that although computer scientists might be looking to the similar problem of legal expert systems, through their lack of legal expertise and training they cannot show themselves as having relevant expertise. We thus (for I am a computer scientist) reveal ourselves as incompetent judges of just what a legal expert system is or should do. But this strategy can be countered; and it can be done so quite simply. For I simply have to show that, as a computer scientist, my concerns have been of a jurisprudential nature and that I do have a modicum of relevant expertise which has been applied to my own research in the area. In effect, I wish to suggest that I am not an incompetent judge of just what a legal expert system is, nor of what it should do. The practical examples of Susskind’s strategy, with regards to my own work at any rate, are not too difficult to find. Every mention of my name signals such a move. Thus: “However, although there are several claims of existing expert systems in law, close examination of the documentation of the systems invariably reveals these pronouncements to be exaggerated. Phillip [sic] Leith, for instance, . . .” Collated, the main points which Susskind makes are: (1) My program was not an expert system according to the definition provided by Susskind; therefore, presumably, my program can be disregarded as providing any insights into legal expert systems.

’ Dolby, p.270, 1982.

130

THE MODERN L A W REVIEW

[Vol. 50

(2) I did not reason with an expert’s heuristic knowledge when inputting the “legal knowledge” into my program (i.e. I do not have a legal qualification). (3) The legal area which I used as a test bed for my program was only a subset of the law (and “we might justifiably query whether that chosen area is indeed a suitable domain of application”). (4) I produced my program in a second-rate research environment (“This is not to belittle Leith’s achievements, for he was clearly working with limited resources”).

Briefly, to reduce the intellectual distance between Susskind and myself, I shall counter these points; for convenience I shall do so in reverse order. 4. I consider the research environment open to me at the Open University where my work was done to have been first rate with regards to hardware (though old, the DEC-20 was an easy to use mainframe), software (the massive programming environment I used was mounted on the mainframe at substantial ongoing cost for my use alone), research time (i.e. limited teaching was expected from me, and very minimal administration) and financial support (I was never refused funding for anything I wished to do). It should be pointed out that at that time, computing academics at OxfordSusskind’s research environment-were working with very much more limited computing facilities than I was, and probably with very many more teaching commitments. I do not know where Susskind gleaned the information which, he felt, allowed him to make this comment. 3. The legal area I was interested in was, of course, a subset of the law. But it is a subset which is of practical import to a substantial number of this country’s poor. If welfare legislation and supplementary benefit law are not suitable areas for lawyers or expert systems to deal with, then why are so many Law Schools presenting courses in same? Of course, another problem for Susskind is that when he criticises others for using legal sub-domains, he forgets that his own chosen area (divorce law) must too be a sub-domain. 2. It might well be that I have no formal law training, but I most decidedly d o have practical knowledge of giving advice on supplementary benefit and welfare law. I worked at the Citizen’s Rights Office, Edinburgh, before entering the academic world and had experience of both the theory of the legislation and the representation of clients in National Insurance and Supplementary Benefit Tribunals. Does that experience allow me to claim “an expert’s heuristic knowledge”? Of course, one of the reasons I chose my particular expert system domain was that I had such a practical viewpoint. Susskind tells us that he will be using Scottish Divorce Law; has he the practical expertise he calls for (“normally acquired over many years”), or has he simply read the textbook? 1. I care not one whit whether my program is described as an expert system or not. I would simply note that Susskind causes himself some considerable problem when he states that, according to his own definition, my program is not an expert system; for, later in this same paper he points out that, “There is little agreement over suitable terminology in the field, to the extent that researchers disagree over what constitutes an expert system in law properly socalled.” If there is no agreement, why should computer scientists accept the definition proposed by a lawyer? Will Susskind accept the definition of “divorce” proposed by a computer scientist?

These are all relatively minor points (to my view). What is more serious my view) is the damage which Susskind implicitly does to my own research credibility by omitting to mention my writings on jurisprudence and the way I have seen the relationship between computing and law. From relatively early on in my research career I saw the importance of a sound jurisprudential understanding of my computer work. Even before the programming on my system began, I was writing and researching into this area. One early paper (Leith, 1984) written in 1982/3 and cited (to

JAN. 19871

CORRESPONDENCE

131

(unread?) by Susskind arose from my discussions with Neil MacCormick into the role of logic, rules and computer program, and another (Leith, 1985) also cited by Susskind used some (quite complex to my view) legal arguments in support of my contention. Papers which are yet to appear ((Leith, 19861 for example) make similar use of jurisprudential argument. Even that paper which Susskind quotes from, hints at its less than purely technical content in the title: “Cautionary Notes On Legal Expert Systems.” Further, I found the jurisprudential approach so important that I left a well equipped Department of Computer Science in order to join a (less well endowed) Department of Jurisprudence. Is Susskind doing anything new by bringing the philosophy of law to legal expert systems research? I suggest not. Yet Another Legal Formalist It is one thing to understand jurisprudential debates. It is another actually to use them constructively in computer science. While Susskind claims that he has the ability to d o this, I suggest that he does not-and that the evidence for my view is contained within a companion paper (Susskind, 1985). For though he believes that he can come to some sort of compromise with the various theoretical positions taken by the renowned thinkers of the field and produce “a general theory” (an indication, I might suggest, of Susskind’s poor theoretical conceptions) it is obvious that his approach has already been mapped out for him. He states: “Our own work, accordingly, is concerned mainly with rules, or productions . . . The jurisprudential aspect of our research project, then, constitutes an inquiry into the presuppositions and implications of building and using expert systems in law together with an elaboration of the foundations of a theory of law compatible with their design, development and implementation. In the design of our own prototype, in the first instance, jurisprudence has been used to supply coherent but nevertheless informal models of law and legal reasoning that are being formally specified using . . . mathematical techniques . . .”*

Thus, Susskind sees jurisprudence as providing a variety of theoretical models which can be modelled mathematically and translated into computer programs. I can argue that his approach is mapped out for him, because of necessity formal specification requires a formalisation of law; there can be no “informal models” which are mysteriously formalised into a computer model. If he believes that such an “automatic formalisation” mechanism exists, Susskind should explain how it operates, for it would be of incalculable importance to the entire development of the field of computer science. There can be no such mechanism. Therefore, in order to use formal specifications, Susskind must provide a formal specification of law which can then be incorporated into the rule-format of a computer program (after Turing, Church, etc., we can state that any computer program, expert system included, can be restated as a series of rules). And, of course, formal specifications of law are renowned for their theoretical and practical inadequacy. Legal formalism can, surely, hardly be the compromise he wishes might arise from the conflicting positions of Kelsen, Hart, Dworkin et al.

*

Susskind, p.310, 1985.

132

THE MODERN LAW REVIEW

[Vol. 50

It is ironic that Susskind, who attempts to use an intellectual distancing strategy, should be caught by his own technique. For, if he really does believe that his informal theoretical models can be transformed into formal theoretical models without loss of their informal attributes, then I must suggest that he has really little understanding of the discipline of computer science. And, of course, these are theoretical models which he promises, not which he presents in this rather insubstantial paper. I look forward to inspecting more closely the Emperor’s legal expert system, in more rigorous and future writings. PHILIP LEITH* References Dolby R. G. A. (1982) “On the Autonomy of Pure Science: The Construction and Maintenance of Barriers between Scientific Establishments and Popular Culture”, in Scientific Establishments and Hierarchies, (Eds.) Elias, Martins, Whitley; D. Reidel Publishing Co. Leith P. (1984) “Logic, Formal Models and Legal Reasoning”, in Jurimetrics Journal, Vol. 24, No. 4. Leith P. (1985) “Clear Rules and Legal Expert Systems”, in Logica, Informatica, Diritto, Conference Proceedings, Florence, 1985. Leith P. (1986) “Fundamental Errors in Legal Logic Programming”, to appear in The Computer Journal. Susskind R. E. & Gold D. I. (1985) “Expert Systems in Law: A Jurisprudential and Formal Specification Approach”, in Logica, Informatica, Diritto, Conference Proceedings, Florence, 1985. Susskind R. E. (1986) “Expert Systems in Law: A Jurisprudential Approach to Artificial Intelligence and Legal Reasoning” (1986) 49 M.L.R. 168. * Faculty of Law, Queen’s University, Belfast.

Lihat lebih banyak...

Comentários

Copyright © 2017 DADOSPDF Inc.