There's an important difference between the substitutability of mere instruments and the equal weighting of intrinsic goods [e.g. persons]. In the former case, the swap makes no difference. In the latter, there is a normatively relevant difference (Bob no longer exists!), it is just that the result is equally desirable [Sally exists!]. There are distinct reasons offered up by the competing intrinsic values; it merely happens that they are in balance. I expect that any plausible view of value will need to accommodate this possibility.
Jack responds that he finds the following bit of reasoning just as compelling:
I am building a fence. Hence, I have *a* reason, countervailed though it is, to prefer that God not annihilate my hammer, EVEN IF God will immediately replace that hammer with another equally useful hammer. After all, there is a normatively relevant difference: the ever so useful, and for that reason instrumentally valuable hammer no longer exists!
One who recognizes the sort of distinction I'm after here will insist that the latter claim is mistaken: the non-existence of a useful instrument is not in itself normatively relevant. What would be normatively relevant would be the destruction and non-replacement of an instrument. (Of course, destruction is typically followed by non-replacement, so it isn't surprising that one might confuse the two.)
But is there anything helpful we can say to the person who doesn't immediately "get" the distinction? Perhaps the best we can do is to, er, hammer on the intuition that replacing persons is obviously significant in a way that replacing tools is not. Once one is on board with this datum, we can inquire into the most sensible way to formally capture this intuitive idea: e.g., perhaps swapping items of (equal) intrinsic value involves conflicting but balanced reasons, whereas substituting mere instruments involves no (different) reasons at all.
To spell it out in more detail: (i) the value of Sally's life gives us a different (albeit equally weighty) reason from that given by the value of Bob's life, whereas (ii) the reason the builder has to want a hammer is satisfied no matter which of the two equally useful hammers he receives. It's not as though he has two different reasons, one to want each particular hammer. Something of value is lost when Sally exists without Bob; nothing is lost by the builder having one hammer but not another that would be useful in exactly the same way. In sum, then, we might say: we have reasons to want any adequate instrument, and reasons to want each intrinsic good.
Even on a purely theoretical level, this sort of distinction makes a lot of sense. It seems extremely intuitive to me to think that there's a conceptual difference to be had between (i) a pair of options serving distinct-but-equal intrinsic values; and (ii) a pair of options that serve literally one and the same values. I think something would be missing from our conceptual repertoire if we couldn't make sense of such a distinction.
The "big-picture" relevance of this concerns the 'value receptacle' objection to utilitarianism, according to which utilitarians are thought to treat people as mere instruments to the ultimate end of promoting aggregate utility.
Indeed, I do think of (e.g.) chickens as mere instruments to the intrinsic value of aggregate animal welfare. So one might worry that utilitarians must think of people in the same way. But (I argue) this is not so. Utilitarians may well impute normative significance to the separateness of persons, for they merely claim that a harm to one may be normatively outweighed by a benefit to another. To outweigh a harm is not to (normatively) cancel it, or to deny that there is any real trade-off here at all. It is just to say that the trade-off is worth it. We should be able to make such distinctions.