My approach to that is not to try to donate a few bucks to every project I use, but to randomly choose a few about quarterly, especially among the ones I use the most, and make larger donations to those. Otherwise it is too hard to remember to donate, and just plain too much work.

If a lot of people do that, it'll even out.

I'd like to have an automated way to make my choices more random, and to weight the probabilities better toward the projects that give me the most value, or are the most generally underfunded compared to the global value they produce. At the moment, I do that more or less by guessing. Unfortunately it sounds Hard(TM) to do it right.

It'd also be nice if you didn't have to go searching for how to donate to every project.