In chapter 9 of his forthcoming book The Limits of Kindness (cited with permission), Caspar Hare offers an ingenious argument for the view that transitivity + minimal decency (i.e. preferring pareto-superior alternatives, whereby some people are better off and nobody is worse off) actually commits us to anonymous benevolence, i.e. preferring world A over world B whenever the inhabitants can be put into a 1:1 correspondence such that the person in A is sometimes better off, and never worse off, than their paired member of B. So, for example, it commits us to preferring (contra Taurek) that we save the many rather than the few in rescue cases, and that we bring into existence better-off rather than worse-off people in non-identity cases. Very roughly, it's an argument that moves us from personal to impartial concerns.
Here's the gist of the argument: We can construct a 'morphing sequence' from any one person into another, whereby each step in the sequence involves just a tiny tweak to the person's qualitative characteristics, and hence (assuming that essence is not 'perfectly fragile', as Hare puts it) each step contains a counterpart of the person in the previous step. (The argument can be restated without relying on counterpart theory, but I won't get into that here.) It is an 'upslope' sequence if each step also features an improvement in welfare. Hare illustrates:
Since each world W_i+1 pareto dominates the previous world W_i, minimal decency requires us to have pairwise preferences for the latter of each such pair. Transitivity of rational preference then commits us to prefer the last world, W_Jill, over the first world, W_Jack, even though there's no particular individual who exists in both worlds that is better off in the latter.
It's a neat argument, but we can put pressure on it by observing that a similar argument would seem to imply that minimal decency commits us to preferring that more perfect people existed in place of our loved ones. Maybe we should prefer it, if impartial consequentialism is true, but it sure shouldn't be that easy to establish. So, what's gone wrong?
Suppose that Jack is your beloved disabled son. The natural place for the partialist to get off the boat is to draw our attention to the last world in the sequence that contains a counterpart of Jack -- call it W_j. The loving parent certainly prefers W_j to the actual world W_Jack, since Jack is better off there and all else is equal. But must they prefer W_j+1 to W_j, as Hare supposes? It's true that if W_j were actual, then we'd have to prefer W_j+1. But given that W_Jack is actual, our personal concern has latched onto this person, Jack, who has no counterpart in W_j+1. The loving parent of Jack could thus reasonably prefer the world W_j where Jack exists (albeit with significantly different characteristics from those he actually has) and is pretty well-off, over the world W_j+1 which instead contains an even better-off person who isn't their Jack (though they are a close counterpart of the person who would have "been Jack" had W_j been actual).
Similar objections can be raised in rescue cases: at some point in the morphing sequence, there will be an individual who is not a direct counterpart of any actual person, but merely a counterpart-of-a-counterpart of an actually existing person. The partialist might reasonably prefer that (counterparts of) actual people exist rather than better-off other people. They can thus break the chain of pairwise preferences that Hare relies on.
In short: if it's rationally permissible for us to be 'biased towards the actual', and prefer that the actually-existing people exist rather than slightly better-off different people, then Hare's morphing argument fails. Hare's "minimal decency" requirement is not so minimal as it sounds.