The Pleadings Game: An Artificial Intelligence Model of Procedural Justice

254 Pages·1995·6.88 MB·English

by  Thomas F. Gordon (auth.)

The Pleadings Game
The Pleadings Game
An Artificial Intelligence Model of
Procedural Justice
by
Thomas F. Gordon
GMD-German National Research Center for
Information Technology
SPRINGER-SCIENCE+BUSINESS MEDIA. B.V.
Library of Congress Cataloging-in-Publication Data
Gordon, Thomas F.
The pleadings game: an artificial intelligence model of
procedural justice / by Thomas F. Gordon.
p. cm.
Includes bibliographical references and index.
ISBN 978-90-481-4591-1 ISBN 978-94-015-8447-0 (eBook)
DOI 10.1007/978-94-015-8447-0
1. Law--Methodology. 2. Logic. 3. Reasoning. 4. Artificial
intelligence. I. Title.
K213.G68 1995
340' .11--dc20 95-30098
ISBN 978-90-481-4591-1
Printed on acid-free paper
All Rights Reserved
© 1995 by Springer Science+Business Media Dordrecht
Originally published by Kluwer Academic Publishers in 1995
Softcover reprint ofthe hardcover 1st edition 1995
No part of the material protected by this copyright notice may be reproduced or
utilized in any form or by any means, electronic or mechanical,
including photocopying, recording or by any information storage and
retrieval system, without written permission from the copyright owner.
To my mother; for the self-confidence needed to begin
and to my wife, Ines, and children, Dustin and Caroline,
for the patience needed to finish
Contents
Preface ix
Chapter 1 Introduction 1
Chapter 2 The Legal Domain: Article Nine 11
2.1 Introduction to Article Nine 13
2.2 Article Nine Priority Rules 14
2.3 Higher-Order Conflicts Between Rules 16
Chapter 3 Philosophy of Legal Reasoning 21
3.1 Hart's Analytical Jurisprudence 22
3.2 Rodig's Legal Logic 31
3.2.1 Axiomatization ........ 35
3.2.2 Formalization 44
3.2.3 Evaluation ofROdig's Theory 51
3.3 Alexy's Theory of Legal Argumentation 53
3.3.1 Normative Theories of Practical Discourse 58
3.3.2 Habermas' Logic of Discourse .......... 63
3.3.3 The Erlangen School .................. 67
3.3.4 Perelman's New Rhetoric 69
3.3.5 Evaluation of Alexy's Theory 71
3.4 Dimensions of Discourse Games 73
Chapter 4 Formal Models of Argumentation 76
4.1 Lorenzen's Dialogue Logic .. .. .... 76
4.2 Pollock's OSCAR Model of Defeasible Reasoning ........ 83
4.3 Simari and Loui's System .. .. .. .. .. .. .. .. .. .... 88
4.4 Conditional Entailment .................. 94
4.5 Related Work ................ 106
Chapter 5 The Pleadings Game 109
5.1 Civil Pleading 110
5.2 A Language for Explicit Exceptions ........ 114
5.3 The Pleadings Game ................ 126
5.4 A Detailed Example 142
5.5 The Trial Game 152
vii
viii
5.6 Dialectical Graphs 155
5.7 The Concept of an Issue 162
Chapter 6 An Implementation in Standard ML 170
6.1 Standard ML's Module System 170
6.2 Modules Overview.. .. .. .. .. .. .. .. .. 173
6.3 ci 1 - A Theorem Prover for Clausal Intuitionistic Logic 175
6.4 Rules - A Rule Language Translator 180
6.5 Mrms - A Minimal Reason Maintenance System 182
6.6 Record - The Pleadings Game Playing Board 186
6.7 Ce - A Theorem Prover for Conditional Entailment .. 189
6.8 Clerk - The Pleadings Game Mediator 196
Chapter 7 Conclusion 201
Appendix A The Article Nine World .............. 206
A.l A9W Article 1; General Provisions ........ 207
A.2 A9W Article 2; Sales .. .. .. .. .. .. .. 207
A.3 A9W Article 9; Secured Transactions 208
A.4 Legal Principles .............. 218
A.5 Common Sense Knowledge ...... 219
A.6 Dictionary of Predicate Symbols 221
Appendix B Glossary of Legal Terms 226
References 229
Index 239
Preface
The British philosopher Stephan Toulmin, in his The Uses of Argument, made the
provocative claim that "logic is generalized jurisprudence". For Toulmin, logic is the study
of nonns for practical argumentation and decision making. In his view, mathematical
logicians were preoccupied with fonnalizing the concepts of logical necessity,
consequence and contradiction, at the expense of other equally important issues, such as
how to allocate the burden of proof and make rational decisions given limited resources.
He also considered it a mistake to look primarily to psychology, linguistics or the cognitive
sciences for answers to these fundamentally nonnative questions.
Toulmin's concerns about logic, writing in the 1950's, are equally applicable to the
field of Artificial Intelligence today. The mainstream of Artificial Intelligence has focused
on the analytical and empirical aspects of intelligence, without giving adequate attention
to the nonnative, regulative functions of knowledge representation, problem solving and
decision-making. Nonnative issues should now be of even greater interest, with the shift
in perspective of AI from individual to collective intelligence, in areas such as multi-agent
systems, cooperative design, distributed artificial intelligence, and computer-supported
cooperative work. Networked "virtual societies" of humans and software agents would
also require "virtual legal systems" to fairly balance interests, resolve conflicts, and
promote security.
This book draws upon jurisprudence and moral philosophy to develop a fonnal model
of argumentation, called the Pleadings Game. From a technical perspective, the Pleadings
Game can be viewed as an extension of recent argumentation-based approaches to
nonmonotonic logic: 1) The game is dialogical instead of monological;.2) The validity
and priority of defeasible rules is subject to debate; and 3) Resource limitations are
acknowledged by rules for fairly dividing the burdens of representation and proof among
the players.
Rather than starting with abstract preconceptions about argumentation, the approach
taken here is to first evaluate some important jurisprudential theories of argumentation
and reasoning in the context of a particular legal domain, the U.S. commercial law on
secured transactions. This leads to a better appreciation of the variety and types of priority
relationships between rules than one would have expected from a survey of the literature
on argumentation or nonmonotonic logic alone. More importantly, it also leads to the
identification of the proper division of power between legislative and judicial branches
of government as being the main issue in legal philosophy driving the search for adequate
models of reasoning and argumentation. (No special legal knowledge is needed to
understand this book. All of the law and legal theory needed to understand the issues and
examples is contained here.)
The division of power issue is usually phrased in tenns of the limits of judicial
ix
x
discretion. Where is the border between a judge's power to decide cases by applying the
law and his power to create law? How can it be decided whether or not ajudge has exceeded
the limits of his discretion? What methods can be applied to construct decisions which
are correct or justified, given these limits? The Pleadings Game may be of interest to legal
theorists to the extend it helps to shed light on issues such as these.
Most prior work in AI and Law has been founded on the prominent legal philosophy
of H.L.A. Hart. In the basic version of Hart's theory, judicial discretion begins where
the literal meaning of authoritative legal texts, such as legislation and case books, ends.
To use a standard example, if a law prohibits vehicles from a public park, then a judge
would not have discretion to allow a tank, even if it is to be used as a war memorial. AI
and Law models typically try to represent the meaning of legislation and cases using some
formal language, often a variant of first-order predicate logic. In this approach, the task
of modeling discretion is reduced to finding an inference relation such that precisely the
decisions which the judge may reach are derivable from the representation of the law and
the facts of the present case.
It is argued here that relational models of this kind cannot capture the essence ofj udicial
discretion, as they fail to take seriously the problem of representing the knowledge of such
inherently political and dynamic domains as the law, where opinions and values differ so
widely and consensus is unlikely to ever be definitely established. These models also fail
to acknowledge the intractability of inference relations for languages expressive enough
to represent legal knowledge. How can such a relational model serve as a practical check
on discretion when no efficient inference procedure for the model can exist?
The Pleadings Game takes another approach, founded in the discourse theory of legal
argumentation developed by Robert Alexy, a contemporary German legal philosopher.
In Alexy's theory, it is not the meaning of the law alone which delimits judicial discretion,
but rather the discourse norms of rational argumentation. These norms regulate the process
by which arguments are presented and compared. The relational or descriptive model is
replaced by a procedural model: a judicial decision is assumed to be correct if there is no
reason to believe that the procedure by which it was reached was not fair.
The particular legal proceeding modeled by the Pleading Game is civil pleading,
where the parties exchange arguments and counterarguments to identify the issues to be
decided by the court. In the model, the arguments a judge may make when deciding a
case are constrained by the arguments made by the parties during pleading. The rules
of the game are designed to promote efficiency and assure that no player can prevent
termination. The concepts of issue and relevance are used to focus pleading and avoid
superfluous arguments. A tractable inference relation is used to commit players to some of
the consequences of their claims. The concept of an issue is formalized using the dialectical
structure of arguments which have been made pro and contra the main claim of the case.
The task of identifying issues is shown to be an abduction problem.
A prototype mediation system for the Pleadings Game has been implemented. A
mediator is a neutral third-party. Its job is to manage the record of the proceeding and help
assure that rules of procedure are not violated, by advising the parties about their rights and
obligations. This book contains a high-level description of the Pleadings Game mediation
system sufficient to reimplement it, in the programming language of your choice.
xi
The Pleadings Game has been tested using examples from Article Nine of the Uniform
Commercial Code of the United States, which covers secured transactions. A typical
transaction of this type is a loan by a bank to purchase a car, where repayment of the loan
is secured by the bank retaining a security interest in the car. Article Nine is full of various
kinds of exceptions and priority relations between conflicting rules. In an appendix, the
rule language of the Pleadings Game is shown to be suitable for modeling statutes of this
kind, by representing significant portions of Article Nine.
Let us end this preface by returning to Toulmin and his view of logic. To avoid
confusion with the now dominant conception of logic within AI, it might be preferable to
call Toulmin's broader view oflogic "dialectics". The study, within Artificial Intelligence,
of the theory, design and implementation of systems which mediate discussions and
arguments between agents, artificial and human, might then be called "Computational
Dialectics". This field would be closely related to what Carl Adam Petri, the German
computer scientist, proposed calling "formal pragmatics". The Pleadings Game is a
contribution to Computational Dialectics.
Acknowledgements
For providing me with the opportunity to pursue the dissertation which resulted in this
book, and for his encouragement, support and careful editing, I would first of all like to
thank heartily my principal thesis advisor, Wolfgang Bibel. Thanks also are due to other
members of my thesis committee, L. Thorne McCarty and Adalbert Podlech.
This book began in 1981 at the School of Law of the University of California, Davis.
For supporting my interest in legal philosophy, I would like to thank several of my former
law professors, John Ayer, Florian Bartosic, Gary Goodpaster, John Poulos and, last but
not least, Thomas Ulen. Special thanks go to John Ayer, for having actively sponsored my
artificial intelligence and law independent-study project, despite his own deep skepticism,
and for teaching me more about writing than I have ever learned, before or since.
Bernard Schlink introduced me to Herbert Fiedler shortly after my arrival in Germany
in 1982. Professor Fiedler gave me my start at the German National Research Center for
Computer Science (GMD), for which I am deeply grateful. I would also like to thank him
for the independence he allowed me during the several years I worked in his Information
Law and Legal Informatics group at GMD.
Several colleagues and friends have commented on various versions of this book:
Barbara Becker, Trevor Bench-Capon, Gerhard Brewka, Joachim Hertzberg, Christoph
Lischka, Ron Loui, Lothar Philipps and Henry Prakken. I am indepted to them all.
Special, warm thanks go to my friends Gerhard and Joachim for having taken the trouble
to carefully read the whole manuscript. Of course, I remain solely responsible for all errors
and omissions.
For giving me the time to write the thesis and then revise it for the book, I thank the
director of the Artificial Intelligence Division of GMD, Thomas Christaller.
Finally, I would like to thank the anonymous reviewers for their evaluations and
constructive criticism.
xiii