Abstract
In the course of legal reasoning—whether for purposes of deciding an issue, justifying a decision, predicting how an issue will be decided, or arguing for how it should be decided—one often is required to reach (and assert) conclusions based on a balance of reasons that is not straightforwardly reducible to the application of rules. Recent AI and Law work has modeled reason-balancing, both within and across cases, with set-theoretic and rule- or value-ordering approaches. This article explores a way to model balancing in quantitative terms that may yield new questions, insights, and tools.
Similar content being viewed by others
Notes
Note that the weight being accorded a factor is a kind of co-efficient, not to be confused with the metaphorical weight of options in the imaginary balancing process. These two different senses of ‘weight’ are often conflated and a source of confusion.
Examples of these models can be found in the references in Sect. 3.6 below.
In addition to readily handling option sets with more than two members, the approach outlined here has the advantage over the two-pan scale model of showing all assessments on a given factor in the same row, for easier comparison.
In conceptual spaces greater than two dimensions, of course, a line does not suffice to partition one area from another, and hyperplanes (of one dimension less than the ambient space) are needed. By expressing an arbitrary number of comparative ‘dimensions’ in the single mode of ‘factor,’ interestingly, the choicebox model can largely remain within an intuitive three-dimensional space for interface purposes at least.
Other work by some of these authors does introduce quantitative aspects. E.g., in Chorley and Bench-Capon (2005a) factors are held to promote values to different extents, and values are then assigned weights to give an overall score. In Bench-Capon and Prakken (2010) values are also promoted to differing degrees to accommodate the language in the decisions considered that speak of “reduced expectations” of privacy. No numbers are attached, but case facts are held to put particular situations above or below a required threshold. Bench-Capon et al. (2013) introduces argument schemes that encompass varying degrees to which values are promoted, although degree is limited to two states, strong and weak. The ‘triadic’ model proposed by Alexy (2003) incorporates ‘light,’ ‘moderate,’ and ‘serious’ degrees to which principles are satisfied.
Some weighting schemes require weights to sum to a fixed quantity, but it can be easier for users to express them in whatever relative numbers they choose, and let the system handle the associated mathematics behind the scenes. In the example here, for instance, it is easier to say that the two factors stand in an importance ratio of 5 to 2 (totaling 7), rather than having to figure out what that ratio would be against a fixed total of 10 (approximately 0.7142857–0.2857142).
Whether numbers, shapes, or other devices are used to express ratings and weightings quantitatively, in many cases results can be indifferent within ranges of quantities, and hence sensitivity analysis can be used to detect how much any particular judgments would need to change in order for the result to be different.
131 S. Ct. 1143, at 1160.
The literature on argument accrual of course reminds us that things are sometimes not so simple. Prakken (2005) for instance shows that accruals can be weaker than their accruing elements when those elements are not independent, and presents a logical formalization of argument accrual as a kind of inference. Lucero et al. (2009) turn to ‘possibilistic defeasible logic programming’ to model conflict and defeat in accrual structures.
References
Alexy R (2003) On balancing and subsumption. A structural comparison. Ratio Juris 16(4):433–449
Arrow K (1963) Social choice and individual values, 2nd edn. Wiley, New York
Ashley K, Brüninghaus S (2005-2006) Computer Models for legal prediction. Jurimetrics 46:309
Ashley K, Brüninghaus S (2009) Automatically classifying case texts and predicting outcomes. Artif Intell Law 17:125–165
Bench-Capon T, Prakken H (2010) Using argument schemes for hypothetical reasoning in law. Artif Intell Law 18(2):153–174
Bench-Capon T, Prakken H, Wyner W, Atkinson K (2013) Argument schemes for reasoning with legal cases using values. In: Proceedings of the Fourteenth International Conference on Artificial Intelligence and Law, pp 13–22
Brams S (2008) Mathematics and democracy: designing better voting and fair-division procedures. Princeton University Press, Princeton
Chorley A, Bench-Capon T (2005a) AGATHA: using heuristic search to automate the construction of case law theories. Artif Intell Law 13(1):9–51
Chorley A, Bench-Capon T (2005b) An empirical investigation of reasoning with legal cases through theory construction and application. Artif Intell Law 13(3–4):323–371
Dawes R (1979) The robust beauty of improper linear models in decision making. Am Psychol 34:571–582
Emerson P (2010) Consensus voting and party funding: a web-based experiment. Eur Polit Sci 9:83–101
Grabmair M, Ashley K (2012) A survey of uncertainties and their consequences in probabilistic legal argumentation. In: Zenker F (ed) Bayesian argumentation. Springer, Netherlands
Hage J (1993) Monological reason-based logic: a low level integration of rule-based reasoning and case-based reasoning. In: Proceedings of the Fourth International Conference on Artificial Intelligence and Law, p 30–39
Hoeflich MH (1986) Law and geometry: legal science from Leibniz to Langdell. Am J Legal Hist 30:95
Horty J, Bench-Capon T (2012) A factor-based definition of precedential constraint. Artif Intell Law 20:181–214
Jarke M, Jelassi M, Shakun M (1987) Mediator: towards a negotiation support system. Eur J Oper Res 31:314–334
Kennedy D (1986) Freedom and constraint in adjudication: a critical phenomenology. J Legal Educ 36:518
Lauritsen M (2010) Lawyer’s guide to working smarter with knowledge tools. American Bar Association, Chicago
Lauritsen M (2011) Intelligent tools for managing legal choices. In: Proceedings of the Thirteenth International Conference on Artificial Intelligence and Law, p 106–110
Lauritsen M (2014) ‘Boxing’ choices for better dispute resolution. Int J Online Disput Resolut 1(1):70–92
Lucero MJG, Chesñevar CI, Simari GR (2009) Modelling argument accrual in possibilistic defeasible logic programming. In: Symbolic and quantitative approaches to reasoning with uncertainty. Springer, Berlin Heidelberg, p 131–143
McFadden P (1988) The balancing test. BCL Rev 29:585
Morge M (2006) Collective decision-making process to compose divergent interests and perspectives. Artif Intell Law 13:79–92
Prakken H (2005) A study of accrual of arguments, with applications to evidential reasoning. In: Proceedings of the Tenth International Conference on Artificial Intelligence and Law (2005)
Saari D (1995) Basic geometry of voting. Springer, New York
Sartor G (2010) Doing justice to rights and values: teleological reasoning and proportionality. Artif Intell Law 18(2):175–215
Sartor G (2013) The logic of proportionality: reasoning with non-numerical magnitudes. German LJ 14:1419
Sieckmann J (2003) Why non-monotonic logic is inadequate to represent balancing arguments. Artif Intell Law 11:211–219
Sweet AS, Mathews J (2008) Proportionality balancing and global constitutionalism. Columbia J Transl Law 47:72
Acknowledgments
I am grateful to several anonymous reviewers for insightful suggestions and encouragement.
Author information
Authors and Affiliations
Corresponding author
Additional information
This work is based on an earlier work: “On Balance,” in Proceedings of the Fourteenth International Conference on Artificial Intelligence and Law, © ACM (2013). http://dx.doi.org/10.1145/2514601.2514611
Rights and permissions
About this article
Cite this article
Lauritsen, M. On balance. Artif Intell Law 23, 23–42 (2015). https://doi.org/10.1007/s10506-015-9163-0
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10506-015-9163-0