I think you’d have to be making one of two assumptions for this argument to work:
We don’t have the ability to restrict how our grantees use the funds we give them, or
We can restrict our funding to projects aligned with our focus areas, but we can’t properly assess whether an organization’s other work or overall philosophy will spoil/reduce the impact of the work we want to fund.
You could also have:
2.5. Donating money to grantees which have very different goals is likely to incur in an impact penalty, and this impact penalty is likely to be larger the less the grantees understand and share your goals.
So for instance, donating to an organization not sharing your goals reduces their effectiveness through fungibility, as you point out. But it would also reduce their impact by doing a worse job than someone who was more aligned, making it less likely that they will continue pursuing their goals after you cease funding, employing worse strategies to pursue your goals in situations where its hard to model, and sometimes even trying to actively mislead you.
Taking a guess, this alone does seem to me that it could make a 2 to 100x difference.
In practice, this doesn’t seem so salient for something like, say, GiveDirectly, because once you decide that direct transfers are cost-effective, they seem very aligned in making the transfers happen. But the wiggle room for political grants seems much greater.
A large part of the job of a Program Officer at any funder is to conduct a comprehensive investigation into potential grantees to determine if they are capable of carrying out the funder’s goals.
Note that some of the grants in my discussion section, and discussed in the comment were “discretionary grants”: (Essie Justice group, Justice Strategies, Reframe mentorship). And in addition, many of the criminal justice reform grants also tend to have particularly short writeups.
You could also have:
2.5. Donating money to grantees which have very different goals is likely to incur in an impact penalty, and this impact penalty is likely to be larger the less the grantees understand and share your goals.
So for instance, donating to an organization not sharing your goals reduces their effectiveness through fungibility, as you point out. But it would also reduce their impact by doing a worse job than someone who was more aligned, making it less likely that they will continue pursuing their goals after you cease funding, employing worse strategies to pursue your goals in situations where its hard to model, and sometimes even trying to actively mislead you.
Taking a guess, this alone does seem to me that it could make a 2 to 100x difference.
In practice, this doesn’t seem so salient for something like, say, GiveDirectly, because once you decide that direct transfers are cost-effective, they seem very aligned in making the transfers happen. But the wiggle room for political grants seems much greater.
Note that some of the grants in my discussion section, and discussed in the comment were “discretionary grants”: (Essie Justice group, Justice Strategies, Reframe mentorship). And in addition, many of the criminal justice reform grants also tend to have particularly short writeups.