E.G.

I'm a fan of the view that

All moral obligations are directed: If S1 has a moral obligation to φ, then there is some individual, S2, to whom this obligation to φ is owed.

The view runs into problems when we consider cases where it seems that there is a moral obligation to φ but it is hard to say who the obligation might be owed to. Here is a case:

The charity case
You have to decide whether to donate to Charity A or Charity B. You have no other options, somehow.

If you donate to Charity A, you'll enable the charity to save the lives of three individuals in Population X. That is, there are three individuals in Population X who will live if you donate and who will die if you do not.

If you donate to Charity B, you'll enable the charity to save the lives of thirty individuals in Population Y. That is, there are thirty individuals in Population Y who will live if you donate and who will die if you do not.

You aren't directly involved with either charity. And you have no way to observe, control, or influence their work. So you know next to nothing about the individuals whose lives you'll save, apart from which population they're in. 

The case can be fleshed out in such a way that it is clear that there is a moral obligation to donate to Charity B. So suppose the case is fleshed out in some such way.

Here's what I'd quite like to say about this case:

The straightforward solution: You have thirty separate moral obligations: one for each of the individuals whose life will be saved if you donate. So, if we say that those individuals are S1, S2,…, S29, and S30, then: You have an obligation to S1 to donate to Charity B, and a separate obligation to S2 to donate to Charity B, and so on, through S30. If you choose to donate to Charity A instead of Charity B, then you violate thirty different obligations all at once.

I'd like to just endorse this straightforward solution and call it a day. But there is at least one problem with this solution that I think requires attention and makes me a bit nervous.

To show the problem, I'll add a few details to the case. Imagine that Charity B is in the business of delivering life-saving medicine to people who are suffering from a certain deadly illness. The medicine is in short supply, so Charity B uses a random number generator to determine who gets it. If you donate to Charity B, they'll use the money from your donation to purchase enough medicine to save thirty people, and then they'll pick thirty members of Population Y at random, and then they'll save those thirty individuals' lives.

In that variant of the case, it may be suggested that, before you've decided whether to donate to Charity A or Charity B, there is no fact of the matter about whose life will be saved if you donate to Charity B. The identities of the individuals whose lives will be saved are determinate only after the random selection process has been completed.

My preferred solution requires me to either deny that plausible-sounding suggestion, or allow that it is possible to have an obligation to an indeterminate individual. And I do not want to allow the possibility of an obligation to an indeterminate individual. So I think I need to be able to say the following two things:

(A) If you will donate to Charity B, then there is a fact of the matter about whose name will be selected in the random process that will be used to determine whose life gets saved.

(B) If you will not donate to Charity B, then there is a fact of the matter about whose name would be selected in the random process that would be used to determine whose life gets saved, were it the case that (contrary to fact) you will donate to Charity B.

I'd say that both of these claims are problematic. But (B) seems especially problematic.

The claim in (B) is that there is a fact of the matter about what the output of a random process would be, even if the process in question has not been and never will be initiated. It is unclear what would fix such a fact.

Suppose that I have a perfectly fair coin in my hand. I consider flipping it, but choose not to. It sounds weird to say that there is a fact of the matter about whether it would have been heads or tails. But it seems that this is precisely the sort of thing that my preferred solution to the present problem requires me to say.

I do not know of any knock-down arguments against (A) and (B). Nor do I know of any knock-down arguments against the view that there is a fact of the matter about what the outcome of a coin toss that hasn't happened, and never will happen, would have been. Until I stumble upon such arguments, I believe I will continue to side with the straightforward solution described above, though I will remain a bit uneasy about this.