Running a charity that you think generates utility less efficiently than some existing charity would be madness.

Many charities could have close marginal worth, and rational allocation of resources would keep them that way. A charity that is less efficient could still perform a useful function, merely needing a decrease in funding, and not disbanding.

And you can't have statically super-efficient charities either, because marginal worth decreases with more funding. For example, a baseline of hundred million dollars SIAI yearly budget might drive marginal efficiency of a dollar donation lower than of other causes.

In a sane world where everyone had the same altruistic component of their values, the marginal EU of all utilities would roughly balance up to the cost of discriminating them more closely. I'd have to think about what would happen if everyone had different altruistic components of their values; but if large groups of people had the same values, then there would exist some class of charities that was marginally balanced with respect to those values, and people from that group would expend the cost to pick out a member of that class but then not look too much harder. If everyone who works for a charity is optimistic and claims that their charity alone is the most marginally efficient in the group, that raises the cost of discriminating among them and they will become more marginally unbalanced.