If legal decision-making were merely embodied in the exercise of legislative power, for example, it might appear that judgments, once embodied in statute, had been made once and for all. If, on the other hand, legal decision-making is conceived of as in pursuit of the goal of what is fair, just, and reasonable, as well as to determine whether there is any particular rule or legislation that governs in the area, it is arguable that judgment may, from time to time or from situation to situation, change with regard to what is called for in respect to rules and legislation. The problem remains whether we allow our reasoning powers to become “fettered”21 by prior judgments, or by judgments along the way, that we later come to question.

The judicial “doctrine” (cited at the head of this paper) that an agency that has discretion must also exercise discretion (presumably considering anything else in the body of the law that is favorable to the individual, as well as the literal reading of the particular controlling statute) has to have been revolutionary in Britain. De Smith, Woolf and Jowell appears to be the leading source in the literature, in a revision under the authority of now Lord Chief Justice Woolf. It is invoked even down to the level of the law examiners. It appears to be widely cited in the Commonwealth. It is also cited by the courts in Hong Kong as authoritative—although agencies there disregard it with impunity. But, such a doctrine is completely unheard of in the United States, where the Supreme Court has effectively adopted the position that it will not interfere where the agency has discretion. And, agency “discretion” there only means agency policy pure and simple.


Of course, if an agency should exercise its unfettered discretion, considering everything relevant also favorable to the individual, this could also be done by a computer programmer. But, there is the pitfall. The programmer, or whoever determines what gets programmed, practically has to have the qualities of a disinterested senior lawyer in order to put into the balance of what gets programmed, all that must be considered. Otherwise, the role of “law programmer” (or agency official) will go the way of the educational technologist (in section 2 below), whose only understanding of education technology is the memory aid.

Professor Craig W. Kirkwood has summarized the results of statistical assessment of comparisons of applying system dynamics models to decision-making with those of experts in the field, and finds, similarly, that the system model results to have been as good or better than those of the experts—except where expert opinion is needed to set up the models:

The research discussed … carries an optimistic message for those of us who work on quantitative models. Even simple quantitative models can outperform experts in prediction tasks. However, the research also points out that experts play a key role in developing such models: They are needed to identify the key variables to incorporate into a model. #p#分页标题#e#


The research also carries a cautionary message for those working on developing computer-based expert systems. The generally stated criterion for judging the effectiveness of such systems is how well they replicate the performance of an expert. However, the research indicates that at least in prediction tasks it is possible with even simple models to outperform experts once the key predictor variables have been identified. Thus the performance of experts may not be a good benchmark for judging the performance of a computer-based expert system. It is probably possible to do better.22

Probably none of us might be willing to have life or reputation left totally in the hands of a computer model when it comes to legal decision-making. Yet, if that process were still subject to review, and/or appeal, there would seem to be much to be said for this innovation. The danger is clear, of course, legal decision-making is also subject to equitable and other tests of reasonableness that we may not always agree can be so easily programmed in. However, the saying, “garbage in, garbage out”, also applies to decision-making by human experts.

In public administration, a computer programmed to account for “feedback” favorable to the individual as well as the often one-sided position of the agency might be easier to deal with than an intransigent clerk. Therefore, a well-programmed computer, that accounted for the “feedback” of the system, might be relied on for all that is programmed into it—something that the human mind, which may be better at making individual qualitative decisions, cannot always handle. Great judges do not sit in every court. A good computer model, subject to review, and, I would add, subject to observable inputs, might be able to overcome the worst plagues of the law: judges who do not read and do not listen, and lawyers who do not do the research, and are, therefore, incapable of persuading a court.


Taking the position that judgments should be reviewable is not to say that the power of judgment itself is so unreliable that we cannot trust anything. It is unlikely that we will ever decide, for example, that there is NO rule or legislation that a deliberate and willful taking of a human life for private motives is murder. But there is no question that in nearly a thousand years of the common law, there have been many concessions to recognizing that there are also less heinous crimes that involve the taking of human life, and that there is occasionally, either a measure of necessity or justification for a violation of rules, or evidence of diminished responsibility, which lead us to limit punishment accordingly.

The point is that neither the speed and efficiency of technology, nor the policy priorities of administrative agencies, must be allowed to cloud our better judgment. Similarly, the attractiveness of absolute rules and decisions must not be allowed to make the decision-making process so rigid that we cannot recognize when they produce a result that is repugnant to reasonableness and good judgment. Do the terms “better judgment” and “good judgment” present an insuperable obstacle—as if we were saying that they represented only somebody’s good or better judgment, and that we do not know the criteria by which such judgment is to be distinguished? I do not think so. Rather, these turns of phrase merely reflect that there are times when we question “judgment” of anyone, including ourselves: #p#分页标题#e#