Note: This is Part 3 of the “Justice as Conditionals” series. This post’s concept is actually a recent addition to the theory but is crucial enough to the theory’s rigor that I’m inserting it before we move on to more of the older ideas.     

As it stands, plain Conditionals cannot properly support things like governments without assuming some aspects of morality. The issue arises because not all moralities are compatible. In fact, when defined solely as strict preference sets, almost no moralities are compatible. We need to allow them to be compatible if the theory is to continue. Fortunately, simply looking at the world as it exists provides us a solution: that people demand not specific actions in Conditionals, but a range of actions, whether that range be thin or wide.

               To show what I mean, here’s an example. The Conditional “If you rat me out, I’ll beat the crap out of you” seems fairly specific. We’re obviously missing some context, like who the speaker is being ratted out to, but we can make sense of what is being said overall. The “acceptable range” of the Conditional consists, to the receiver, of the set of actions which do not involve ratting out the speaker. The Conditional has nothing to say about where the receiver can go to dinner, nor does it care the tone of voice in which the receiver is speaking when ratting them out. Any Conditional is going to have an acceptable range, if only because language itself doesn’t allow for the the stated restrictions to be absolutely precise; one can’t meaningfully include the provision that “that atom of iron in that one red blood cell can’t twitch at 12:37 PM tomorrow!”

               There is one more way to phrase this concept that will be even clearer to anyone of the right background. Basically, a Conditional said by A to B is a function from the set of B’s future behaviors to the set of A’s future behaviors. The set of future behaviors n-dimensional for a very large n and the function is discontinuous only on a finite number of (n-1)-dimensional surfaces. Just to be clear, this is still an imperfect analogy. I doubt that language or preferences can be properly modelled by Euclidean space. It’s a fairly good way to phrase it, though. Because of this, I except that rather than using “acceptable ranges” as the term for them, I’ll generally call them spaces within the Conditional’s domain and codomain (the receiver’s future behaviors and the speaker’s future behaviors, respectively).