Say a rule R establishes some outcome O if it explicitly calls for the realization of O. Say R engenders O if the realization of O is merely a forseeable consequence of R. (Cf. Pogge 1989, p. 38) It seems obvious to judge a rule unjust if it establishes an unjust outcome. It’s perhaps more controversial in some cases but not uncommon to judge a rule unjust if it engenders an unjust outcome.
Suppose someone claims that a rule R (and our enforcement of it) is unjust because the rule incentivizes predatory conduct on the part of state leaders under certain domestic conditions C. (Thomas Pogge and Leif Wenar make such a claim with respect to the “international resource privilege”; see Pogge 2002 and Wenar 2008.) Suppose that R induces the predatory conduct only in the presence of C; in the absence of C, no predatory conduct ensues. In some cases, I want to claim that R does not engender the predatory conduct even if it does generate incentives to engage in predation. Here’s the argument.
Whether a particular rule or class of rules R engenders a particular outcome Oi depends on whether the set of possible outcomes O is sufficiently responsive to changes in R. If this were not the case, it would be difficult to sustain the judgement that (enforcement of) R causally contributes to the realization of Oi in any salient way. If Oi would be realized under any or most admissible rules, then R is, intuitively, not a salient part of the causal process generating Oi. For example, if interstate wars would occur at roughly the same rate under any admissible set of rules for governing international conflict, it seems misguided to declare that the interstate wars are engendered by any particular set of rules governing international conflict, even if each set of rules induces or fails to take adequate precaution against interstate wars. If this presupposes a controversial view of the engendering relation, we could instead say that R would be an inappropriate target for our moral assessment of Oi (or the process by which Oi is realized), even if we want to say that R does engender Oi in the end.
The basic idea is that a rule isn’t causally important in the production of an outcome if the outcome is a foreseeably consequence of any admissible rule. Does this seem right?
 “Admissible” is here meant to exclude alternate rules that would be plainly unjust, or too complicated to follow and apply, or not minimally feasible to implement.