FOSS increases the general economic output. If you expect more people to donate to appropriate causes, it would generate a greater impact over time. However, if you feel that the public chooses their donation targets poorly, then writing proprietary software and donating that money yourself will produce larger benefits. This is ignoring the potential for "changing things" through the software itself, which is too difficult to calculate.

That's a fairly common and very interesting question. Carl's got some thoughts on it, which we'll hopefully get written up. It's closely linked to two big and controversial issues: how good is economic growth and how good is technological progress? It's a case of weighing your contribution to that against the extra donations you can make.