You can only regard sub-agents if there is at least some overlap. Otherwise there exist an utility event horizon. Two agents with fundamentally different values can only ignore or fight each other. I did personally experience this a few times. If at a certain time I assign no utility to a certain activity and my future self which might value the activity retrospectively then there is absolutely nothing I could have done about it at that moment because I didn’t want to. That means that if you have a serious mental illness that might temporarily make you do something you don’t value on average then sometimes only other people can force you not to do what that transitory you wants to do.
There is some overlap. Among other things, I never stop caring for the people I love, although I’m a lot less selfish about it when Neutral. This is really quite important, since the fact that they care for me means that I always need to extend some modicum of care to myself on their behalf, even if I’m truly loathing myself at the time. Being forced to that contradiction sometimes makes me hate them, or my care for them, but it doesn’t diminish my care for them, and is still powerful.
if you have a serious mental illness that might temporarily make you do something you don’t value on average then sometimes only other people can force you not to do what that transitory you wants to do
I think I agree with the rest of your argument that agents need to have some common ground, but this seems to indicate that all their ground must be common (utilities with the same sign, if not the same weight). I don’t think this is the case, and I think I’ve done many things that I don’t value on average. Self-mutilation is something that even Depressed almost never assigns a positive value to, but I have done that.
You can only regard sub-agents if there is at least some overlap. Otherwise there exist an utility event horizon. Two agents with fundamentally different values can only ignore or fight each other. I did personally experience this a few times. If at a certain time I assign no utility to a certain activity and my future self which might value the activity retrospectively then there is absolutely nothing I could have done about it at that moment because I didn’t want to. That means that if you have a serious mental illness that might temporarily make you do something you don’t value on average then sometimes only other people can force you not to do what that transitory you wants to do.
There is some overlap. Among other things, I never stop caring for the people I love, although I’m a lot less selfish about it when Neutral. This is really quite important, since the fact that they care for me means that I always need to extend some modicum of care to myself on their behalf, even if I’m truly loathing myself at the time. Being forced to that contradiction sometimes makes me hate them, or my care for them, but it doesn’t diminish my care for them, and is still powerful.
I think I agree with the rest of your argument that agents need to have some common ground, but this seems to indicate that all their ground must be common (utilities with the same sign, if not the same weight). I don’t think this is the case, and I think I’ve done many things that I don’t value on average. Self-mutilation is something that even Depressed almost never assigns a positive value to, but I have done that.
Would you mind elaborating?