Should you not compare the mean benefit to some community for the two cases of releasing the code either as BSD or as AGPL? Of course you will have to include c

Should you not compare the mean benefit to some community for the two cases of releasing the code either as BSD or as AGPL? Of course you will have to include counterfactual scenarios, rather than what you did: summing up the total benefits of only observed scenarios. Obviously, averaging over counterfactuals is difficult, if not impossible in practice, so people go with intuition. It remains up for debate which way is the better basis for a ethical decision theory.