I’m glad to see other people following this line of argument :) I agree that many true payoff functions are likely to have high-hanging fruits, e.g. any cause area aiming at social change that requires reaching a breaking point would have such a payoff function. However, it’s expected payoff functions we’re interested in, i.e. how we imagine the true payoff function to look like given our current knowledge. I’ve also thought a bit about whether there would “high-hanging fruits” in this sense and haven’t been able to come up with clear examples. So, like Harrison, I would take issue with your second claim that “high-hanging fruits may be prevalent”. I cannot think of any cause area/interventions which could plausibly be modelled, given our knowledge, as having high-hanging fruits on a sufficiently large scale (unlike the examples in your footnote 3, which have increasing marginal returns at a very small scale). This is because when we take into account our uncertainty, the expected payoff function we end up with is usually devoid of high-hanging fruits even if we think that the true payoff function does have high-hanging fruits. This happens, for example, when we don’t know where the threshold for successful change (or the stairstep in the payoff function) lies. And the less information we have, the less increasing marginal returns our expected payoff function will have. I think this applies very much to the two possible scenarios you give. The most promising examples of expected payoff functions with high-hanging fruits I can think of are cause areas where the threshold for change is known in advance. For example, in elections, we know the required number of votes that will lead to successful change (e.g. passing some law). If we know enough about how the resources put into the cause area convert into votes, our expected payoff function might indeed have high-hanging fruits. (However, in general we might think it would be increasingly harder to “buy” votes, which might imply diminishing marginal returns.) In any case, I would also be very interested in any convincing real-life example of expected payoff functions with high-hanging fruits.
I’m glad to see other people following this line of argument :)
I agree that many true payoff functions are likely to have high-hanging fruits, e.g. any cause area aiming at social change that requires reaching a breaking point would have such a payoff function.
However, it’s expected payoff functions we’re interested in, i.e. how we imagine the true payoff function to look like given our current knowledge. I’ve also thought a bit about whether there would “high-hanging fruits” in this sense and haven’t been able to come up with clear examples. So, like Harrison, I would take issue with your second claim that “high-hanging fruits may be prevalent”. I cannot think of any cause area/interventions which could plausibly be modelled, given our knowledge, as having high-hanging fruits on a sufficiently large scale (unlike the examples in your footnote 3, which have increasing marginal returns at a very small scale).
This is because when we take into account our uncertainty, the expected payoff function we end up with is usually devoid of high-hanging fruits even if we think that the true payoff function does have high-hanging fruits. This happens, for example, when we don’t know where the threshold for successful change (or the stairstep in the payoff function) lies. And the less information we have, the less increasing marginal returns our expected payoff function will have. I think this applies very much to the two possible scenarios you give.
The most promising examples of expected payoff functions with high-hanging fruits I can think of are cause areas where the threshold for change is known in advance. For example, in elections, we know the required number of votes that will lead to successful change (e.g. passing some law). If we know enough about how the resources put into the cause area convert into votes, our expected payoff function might indeed have high-hanging fruits. (However, in general we might think it would be increasingly harder to “buy” votes, which might imply diminishing marginal returns.)
In any case, I would also be very interested in any convincing real-life example of expected payoff functions with high-hanging fruits.