Perhaps a better idea would be to spend money on education of women in poor areas, something that is known to reduce the fertility rate. By reducing the fertility rate we also reduce the number of poor, starving, dying in HIV etc children born into this world.
I think that simply measuring the number of dead children may be useful as a simplification, but it’s too simplistic. Really, to me it seems like it’s just something that people believing in axiomatic morals are having problems dealing with. “But, think of the children!”
If the answer to “is it better to spend this money on saving a kids life?” is always yes, I’d say you have a problem with your value system.
I’ve thought about this problem before, but in the context of peer peer-to-peer file sharing.
The problem is that everyone is acting independent and with limited knowledge. It’s hard to know what other people are choosing. There may also be long delays between you and others paying and the cost changing.
Say that the optimal outcome is that out of $1000M, $200M is spent on insect nets and $800M on wells, and that you can only donate to one charity (too bothersome or high transaction costs or something). Now, if everyone is rational they are going to donate to the wells, and no one to nets. This is a suboptimal outcome. It’d also be difficult to coordinate the millions of people donating, so that just the right amount choose nets instead of wells. A solution to such coordination is to roll a dice. If everyone makes a random selection and lets the probability of choosing nets be 2/10ths then the expected outcome is just what we want.
Now, you can adjust this to how many (you think) are playing like yourselves. E.g. if you know most people are going to give to wells, perhaps it’d be better if you put higher probability on nets (perhaps 100%).