Dworkin's purposes were not those of the administrative lawyer, but I think the example he sets out is helpful in addressing an important question. It is sometimes said (and with increasing frequency by Canadian courts) that whether a decision is reasonable depends on whether it falls within a "range". The narrower the range, the more likely the decision will be struck down; the broader the range, the more likely the decision will survive. So, borrowing Dworkin's example, sergeant A has a much broader range than her counterpart, sergeant B, who is constrained by the need to pick soldiers based on their experience.
An example, hot off the press, of judicial approval of the "range" concept comes from the Supreme Court of Canada's decision in McLean v. British Columbia (Securities Commission), 2013 SCC 67:
 It will not always be the case that a particular provision permits multiple reasonable interpretations. Where the ordinary tools of statutory interpretation lead to a single reasonable interpretation and the administrative decision maker adopts a different interpretation, its interpretation will necessarily be unreasonable — no degree of deference can justify its acceptance; see, e.g., Dunsmuir, at para. 75; Mowat, at para. 34. In those cases, the “range of reasonable outcomes” (Canada (Citizenship and Immigration) v. Khosa, 2009 SCC 12 (CanLII), 2009 SCC 12,  1 S.C.R. 339, at para. 4) will necessarily be limited to a single reasonable interpretation — and the administrative decision maker must adopt it. (See also Zalaski v Law Enforcement Review Board, 2013 ABCA 347 at para. 38)This approach reduces the reasonableness analysis to a single step. All the reviewing court has to do is to construe the statute, thereby determining the limits of permissible outcomes, and ask itself whether the decision under review falls outside those limits (see e.g. Stephenson and Vermeule, "Chevron Has Only One Step"). So with Dworkin's sergeants, if sergeant B picked the five most junior members of the corp, her decision would be outside the range and therefore unreasonable. But if sergeant A did so, her decision would be perfectly acceptable.
One difficulty I have with this approach is that it conflates two distinct questions. The range or "scope" of the decision-making authority granted to the sergeant is an important consideration. But when the reasonableness of her decision is called into question, a court must ask itself a distinct question: "Was the decision unreasonable?" The answer to this question does not necessarily turn on the interpretation of the statute, for there may be other considerations at issue such as the rationality of the resulting decision.
In my view, the question of unreasonableness is separate from the question of the "scope" or "range" or "extent" of the authority given by statute. Sergeant A's decision may be irrational if she picks five incompetent men for the mission. If Sergeant B picks five men who are the most experienced in a particular type of operation (e.g. night operations) rather than the five who are the most experienced in terms of years of service, the reasonableness of her decision will depend on weighing the qualifications of her men in light of the purpose of the mission. In both cases, the question of reasonableness is distinct from the question of range.
McLean further illustrates this difficulty. The question was whether a limitations period in a securities statute was triggered by original misconduct or a settlement agreement in respect of the misconduct. In part, determining whether the decision-maker's interpretation of the limitation period provision was reasonable required the reviewing court to have regard to the underlying values of limitations periods, values not unique to the statutory provision in issue for they are found throughout public and private law:
In other words, the reasonableness of the decision-maker's interpretation had to be judged not just in terms of the statute, but in terms of its effect on the overall fabric of the legal system. Determining the "range" is only one aspect of the judicial review inquiry. It does not tell us why the decision is unreasonable. A separate inquiry into the substantive reasonableness of the decision-maker's interpretation is also necessary. In McLean, the Court acknowledged that in some circumstances, decisions to initiate regulatory proceedings could run afoul of the underlying values of limitation period provisions (see paras. 64-69). Limitations periods exist for good reasons, two of which deserve mention here. First, “[t]here comes a time . . . when a potential defendant should be secure in his reasonable expectation that he will not be held to account for ancient obligations” (M. (K.) v. M. (H.), 1992 CanLII 31 (SCC),  3 S.C.R. 6, at p. 29). Second, at some point “[i]t is better that the negligent [plaintiff], who has omitted to assert his right within the prescribed period, should lose his right, than that an opening should be given to interminable litigation” (Cholmondeley v. Clinton (1820), 2 Jac. & W. 1, 37 E.R. 527, at p. 577; see also M. (K.), at p. 30).
This inquiry may sometimes seem to shade into statutory interpretation, especially where arguments are made based on the supposed purposes of the relevant statutory provisions (this phenomenon is evident in the paragraphs just cited from McLean). But it must remain distinct even in those situations. The question of reasonableness should not turn on the interpretation of the purposes of the statute, for the court ought to be deferential to the decision-maker's view. If deference protects decisions only when the reviewing court agrees with the decision-maker, then it serves no purpose.
My difficulty with conflating the two questions is similar to the difficulty Mark Elliott has with Lord Carnwath's recent speech on Wednesbury unreasonableness:
The former is concerned with the burden of justification the decision-maker is to be subjected to—a consideration that ought properly to be informed by the normative fundamentality of the value upon which the public decision is alleged to impact. The latter concept—adjudicative deference—is concerned with the rigour with which the court will review whether the burden of justification has been discharged. It is in relation adjudicative deference that considerations such as institutional competence can properly enter into play. But the fact that such considerations may sometimes operate does nothing to detract from the fact that the initial allocation of the justificatory burden can (and should) be informed by what is normatively placed at stake by the impugned administrative decision.For Mark, what is "normatively placed at stake" is the value of the decision to the individual: the higher the value, the more constrained the decision-maker's range. Whatever the pertinent considerations, there are always two questions: first, what is the burden of justification (a question which I think should be answered in part by reference to, amongst other things, the scope of the authority granted to the decision-maker) and second, has the decision-maker discharged its burden. These two questions should not be conflated.
The other difficulty I have with this approach is that it essentially dresses up jurisdictional questions in the guise of "clear" statutory provisions or "range setting". All of the difficulties that are present with "jurisdictional" questions are equally present here. Simply put, the concept of "range setting" is a means of identifying some decisions as intra vires (and beyond review) and some questions as ultra vires (and thus liable to be struck down).
This is an argument that I have been developing over a series of posts -- see here, here and here -- with a view to a paper I will publish next spring. Comments are welcome!