From a welfare economics point of view there is of course considerable debate. No surprise there. Figuring out the winners and losers (and the net gain/loss) is tricky business in the best of circumstances. Nevertheless, here's my best shot.

First, let me assume that agri-subsidies have the effect stimulating supply in the first world thus lowering world prices. (Tyler Cowen is entirely correct however that many ag programs in fact raise world prices by reducing supply, but let's not think about this right now.) Second, let me assume we're only asking about the welfare implication in the third world. (If we are concerned about the welfare of the entire world then these subsidies are clearly net losses to the world.) So the question is do first world agri-subsidies, that have the effect of stimulating supply and lowering prices, help or hurt the third world?

I think this depends on whether the third world country is (or rather would be under free trade) a net importer or net exporter of the food in question. If a country is a net importer of the food being subsidized, then they are better off (in the usual Marshallian way of things). The gains to the consumers in the form of lower prices/more food consumed will exceed the losses to the domestic producers in the form of lower prices/less food sold. If it is a net exporter, then the country is worse off. The gains to the consumers will be less than the losses to the producers. Inasmuch as many third world nations are (or rather again would be under free trade) net exporters of food, then they are probably worse off as a result of the subsidies. (This is another assumption, but it seems reasonable to me.) It is conceivable that the first world could be subsidizing food that other countries would normally import. In this case, they gain by receiving the dumped products.