1. We cannot be worse off for having more information.
For the purposes of this premises, X is defined as a constant level of utility.
2. If we have a choice between helping one person to an extent X, or helping 100 people to the extent X, with both of these options incurring equal cost to ourselves, we should help the 100 people.
3. If we have a choice between helping an extremely utility-poor person to extent X, and helping a considerably better-off person to extent X, with both of these options incurring equal cost to ourselves, we should help the utility-poor person. This holds regardless of our own level of well-being.
4. According to a self-interest view of the basis of morality, if given a choice between helping two sets of people, we should help the one we would expect ourselves to fit into. (e.g. helping family/country before foreigners)
It is my contention that, for a well-off person, premises, 3 and 4 are inconsistent - a well-off person would expect themselves to be helped more by a strategy of helping well-off people than one of helping the downtrodden. The objection that our principles should be devised at a level where we are ignorant of our positions within society conflict with my first premise. This will need a lot of tightening up, and I'm far from convinced that premise 1 is indeed true; however, it's something perhaps worth thinking about.
(This occurred to me while reading Jan Narveson's paper "We Don't Owe Them A Thing!"