Sunday, July 20, 2008

Fuzzy Logic -- Once Again

I keep returning to fuzzy logic. I do so plainly in part because I think fuzzy logic is an important key to human understanding of the world. But I also continue to revisit fuzzy logic because I firmly sense that the proper use of fuzzy logic in law remains to be worked out. (Indeed, I think this work has barely begun.)

When thinking about the implications of fuzzy logic for legal theory -- and for law in action --, I find it useful to consider, from time to time, the reception that has been given to fuzzy logic and how it has been used thus far. The following summary by Phil Serchuk in his 2005 honors thesis is quite useful in this regard:

Fuzzy logic is unique because it is a source of controversy not only in philosophy but in computer science as well, where it has been used in many successful applications. Yet despite its successes there are many compelling criticisms that fuzzy logic has yet to overcome. Both proponents and opponents of fuzzy logic tend to take extremist positions, a situation that makes fair accounts and criticisms of fuzzy logic difficult to come by. Engineers and computer scientists who use fuzzy logic vigorously defend their work and charge that their critics are conservatives who cannot see the fuzziness inherent to a given domain, a strong charge given the longstanding relationships many of the sciences have had with classical logic. Critics of fuzzy logic tend to take extreme positions as well: fuzzy logic has been described by Berkeley mathematician William Kahan as ‘the cocaine of science’ and as ‘pornography’ by Carnegie Mellon logician Dana Scott [Haa96, p. 230].

While the merits of infinitely-valued logics were being debated by philosophers, computer scientists had began building actual systems that used fuzzy logic. Early fuzzy expert systems and controllers were developed in the 1970’s and fuzzy logic’s ability to use linguistic rules were being explored and put into practice: steam generators and cement kilns were among the earliest applications of fuzzy control. Decision support systems were also being developed and one of the first commercial decision support systems to use fuzzy logic was developed by the INFORM Corporation in 1986 [vA95, p. 279]. Over the 1970’s and 1980’s Japanese researchers had begun to embrace fuzzy logic and by 1989 the Japanese government partnered with 49 companies to found the Laboratory for International Fuzzy Engineering Research (LIFE) [Ter95, p. 1]. LIFE’s six-year mandate was to develop new fuzzy technologies and applications in ‘human-friendly’ fields like robotics and expert systems. By 1990 fuzzy rice cookers, vacuum cleaners, refrigerators, and other electronics were yielding large profits for Japanese companies [LY99, p. 7]. The last part of the 1990’s saw many more concrete applications of fuzzy logic, particularly in computer systems designed to help humans make difficult and complex decisions. In addition to traditional applications in business and engineering, these applications were developed for many disciplines across the natural and social sciences as researchers in these fields slowly began to see the phenomenon of vagueness as being inherent to certain aspects of their work.

Like many other observers who are sympathetic to fuzzy logic, Serchuk stresses that it is important to distinguish uncertainty and vagueness.

the dynamic evidence page

Post a Comment