I don’t think EA funds will outperform the current situation.
In prediction, what generally has the most predictive value is an ensemble of models. What EA funds does is take this ensemble of models on reality that people have, throw away all the information, and use only one model of reality, Elie, Lewis or Nick, to make a prediction about where the best use of money is.
I suppose the general argument is that Elie, Lewis or Nick have an information advantage. Which is plausibly true, however, why not just make this information public and let people decide with their own models what the best use of their money is? Given that we are in a cooperative environment here where we all want to maximize to total amount of impact, in contrast with general asset management, and one only wants to maximize the returns of the investors of the fund.
In conclusion, we lose out on the benefits of model averaging for seemingly no reason.
However, what I believe to be useful, would be a set up where EA funds makes all the information that they have public. Then people can decide for if they want to donate to EA funds, and save themselves valuable time, or make their own judgement on where the best place to donate is.
I don’t think EA funds will outperform the current situation. In prediction, what generally has the most predictive value is an ensemble of models. What EA funds does is take this ensemble of models on reality that people have, throw away all the information, and use only one model of reality, Elie, Lewis or Nick, to make a prediction about where the best use of money is.
I suppose the general argument is that Elie, Lewis or Nick have an information advantage. Which is plausibly true, however, why not just make this information public and let people decide with their own models what the best use of their money is? Given that we are in a cooperative environment here where we all want to maximize to total amount of impact, in contrast with general asset management, and one only wants to maximize the returns of the investors of the fund.
In conclusion, we lose out on the benefits of model averaging for seemingly no reason.
However, what I believe to be useful, would be a set up where EA funds makes all the information that they have public. Then people can decide for if they want to donate to EA funds, and save themselves valuable time, or make their own judgement on where the best place to donate is.