I think the beatpath method to avoid intransitivity still results in a sadistic repugnant conclusion. Consider three situations. In situation 1, one person exist with high welfare 100. In situation 2, that person gets welfare 400, and 1000 additional people are added with welfare 0. In situation 3, those thousand people will have welfare 1, i.e. small but positive (lives barely worth living), and the first person now gets a negative welfare of −100. Total utilitarianism says that situation 3 is best, with total welfare 900. But comparing situations 1 and 3, I would strongly prefer situation 1, with one happy person. Choosing situation 3 is both sadistic (the one person gets a negative welfare) and repugnant (this welfare loss is compensated by a huge number of lives barely worth living). Looking at harms, in situation 1, the one person has 300 units of harm (400 welfare in situation 2 compared to 100 in situation 1). In situation 2, the 1000 additional people each have one unit of harm, which totals 1000 units. In situation 3, the first person has 200 units of harm (-100 in situation 3 compared to +100 in situation 1). According to person-affecting views, we have an intransitivity. But Schulze’s beatpath method, Tideman’s ranked pairs method, minimax Condorcet method, and other selection methods to avoid intransitivity, select situation 3 if situation 2 were an option (and would select situation 1 if situation 2 was not an available option, violating independence of irrelevant alternatives).
Perhaps we can solve this issue by considering complaints instead of harms. In each situation X, a person can complain against choosing that situation X over another situation Y. That complaint is a value between zero and the harm that the person has in situation X compared to situation Y. A person can choose how much to complain. For example, if the first person would fully complain in situation 1, then situation 3 will be selected, and in that situation the first person is worse-off. Hence, learning about this sadistic repugnant conclusion, the first person can decide not to complain in situation 1, as if that person is not harmed in that situation. Without the complaint, situation 1 will be selected. We have to let people freely choose how much they want to complain in the different situations.
Hmm, ya, this seems right. At least for beatpath and the way I imagine it’s used (I haven’t read the paper in a while, and I’m just checking Schulze method on Wikipedia), there is a path from 1 to 3, with strength equal to the minimum of the (net?) betterness of 2 over 1 (300=400-100) and the net betterness of 3 over 2 (500=(1000-0)+(-100-400), or maybe just 1000, counting only the positive votes here), so 300.
The direct path from 3 to 1 has only strength 200=100-(-100) (we ignore the contingent people here, since they have nonengative welfare). Since 200<300, 3 is beatpath better than 1.
For what it’s worth, an option like 2 would have to be practically available to result in 3 being required this way. We can imagine creating many humans, nonhuman animals or artificial sentiences to improve the welfare of existing beings by exploiting these extra moral patients, although their lives would still need to be at least net neutral.
I think the beatpath method to avoid intransitivity still results in a sadistic repugnant conclusion. Consider three situations. In situation 1, one person exist with high welfare 100. In situation 2, that person gets welfare 400, and 1000 additional people are added with welfare 0. In situation 3, those thousand people will have welfare 1, i.e. small but positive (lives barely worth living), and the first person now gets a negative welfare of −100. Total utilitarianism says that situation 3 is best, with total welfare 900. But comparing situations 1 and 3, I would strongly prefer situation 1, with one happy person. Choosing situation 3 is both sadistic (the one person gets a negative welfare) and repugnant (this welfare loss is compensated by a huge number of lives barely worth living). Looking at harms, in situation 1, the one person has 300 units of harm (400 welfare in situation 2 compared to 100 in situation 1). In situation 2, the 1000 additional people each have one unit of harm, which totals 1000 units. In situation 3, the first person has 200 units of harm (-100 in situation 3 compared to +100 in situation 1). According to person-affecting views, we have an intransitivity. But Schulze’s beatpath method, Tideman’s ranked pairs method, minimax Condorcet method, and other selection methods to avoid intransitivity, select situation 3 if situation 2 were an option (and would select situation 1 if situation 2 was not an available option, violating independence of irrelevant alternatives).
Perhaps we can solve this issue by considering complaints instead of harms. In each situation X, a person can complain against choosing that situation X over another situation Y. That complaint is a value between zero and the harm that the person has in situation X compared to situation Y. A person can choose how much to complain. For example, if the first person would fully complain in situation 1, then situation 3 will be selected, and in that situation the first person is worse-off. Hence, learning about this sadistic repugnant conclusion, the first person can decide not to complain in situation 1, as if that person is not harmed in that situation. Without the complaint, situation 1 will be selected. We have to let people freely choose how much they want to complain in the different situations.
Hmm, ya, this seems right. At least for beatpath and the way I imagine it’s used (I haven’t read the paper in a while, and I’m just checking Schulze method on Wikipedia), there is a path from 1 to 3, with strength equal to the minimum of the (net?) betterness of 2 over 1 (300=400-100) and the net betterness of 3 over 2 (500=(1000-0)+(-100-400), or maybe just 1000, counting only the positive votes here), so 300.
The direct path from 3 to 1 has only strength 200=100-(-100) (we ignore the contingent people here, since they have nonengative welfare). Since 200<300, 3 is beatpath better than 1.
For what it’s worth, an option like 2 would have to be practically available to result in 3 being required this way. We can imagine creating many humans, nonhuman animals or artificial sentiences to improve the welfare of existing beings by exploiting these extra moral patients, although their lives would still need to be at least net neutral.