I think we're talking past each other. I'm saying that trying to measure overhead as a way to evaluate charities creates perverse incentives to try to reduce reported overhead, without truly reducing overhead. In order to be effective, charities of course still have to have low true overhead and actually spend money on their work.
Although in principle, if a charity were to spend 80% of its budget on fundraising, but per-dollar donated that charity gets the best results for saving children from dying of malaria or whatever, you really would refuse to donate to it?
Of course not, but the reason we look at these metrics in the first place is because we can't usually say with certainty that Charity A is head-and-shoulders the best charity for saving children from malaria. We're trying to sift through the pile to find the charity that will do the most with our dollar.
Sure, but I'm saying something more specific than just "head-and-shoulders the best". I'm saying we should entirely ignore overhead and in fact pretty much ignore what the charity itself does at all (how much of your donation goes to fundraising? How much goes to lavish hotels?), and just look at the actual malaria rate in the children they treat.
I highly recommend reading the New Republic article I linked to, it talks about all this in a lot more detail.
Although in principle, if a charity were to spend 80% of its budget on fundraising, but per-dollar donated that charity gets the best results for saving children from dying of malaria or whatever, you really would refuse to donate to it?