"Ordinarily, I believe that the order of influence is from character to ideas. But surely, what we think can affect how we act. Especially when, as in Parfit’s case, an intellectual project persists for decades and grows from fascination into monomania."
Good observation.
It doesn't explain all his eccentricities and it certainly doesn't excuse the EA movement from canonizing the monomania.
The latter leans on complex reasoning, understanding of interconnected systems, and rational empathy for a faceless stranger. The former requires emotional intelligence, understanding the inner lives of yourself and others, and emotional empathy for a loved one standing in front of you.
Some people can make 'good' large-scale ethical conclusions when in a calm room by themselves, and can also be unkind to ones they love when in emotional situations.
I'm not sure I agree. "Society" is not something separate from the humans in it and their relationships. And societies that do well at ethics, from what I can see, don't impose "ethical conclusions" from the top down but evolve things like ethics and institutions from the bottom up, starting with the individual humans and their relationships.
> The latter leans on complex reasoning, understanding of interconnected systems, and rational empathy for a faceless stranger.
My question would be, when does this actually work? Looking at human history, it seems to me that this kind of thinking doesn't actually end up with good ethical conclusions. It ends up with outcomes like Dickens's Mrs. Jellyby, who not only sacrifices the needs of those closest to her in order to "help" people far away, but doesn't actually help even the people far away. She feels like she's helped, but she actually harms them.
See https://www.astralcodexten.com/p/in-continued-defense-of-eff...
The question is, why do we pretty much all agree that those obviously good things are obviously good things to do? Do we agree because of some abstract "rationalist" calculation? No, we agree because of obvious human ethical intuitions: wouldn't you want to be free from malaria, have your parasite infection treated, and have clean drinking water? The only "rationalist" calculation that comes in here is figuring out how to give such things to the most people for a given amount of money, and that also doesn't require any huge abstract calculation: it just requires looking at obvious metrics of efficiency, like minimizing administrative overhead and minimizing the impact of the corrupt governments in the countries where these charitable efforts take place.
In other words, we don't need Derek Parfit or any other ethical philosopher to tell us how to be "effective" with charity when the things we are doing are things that pretty much everyone agrees are obviously good things. It's nice if the ethical philosophers agree that those are obviously good things--but it's not clear that Derek Parfit, for example, would agree, since he can't even agree that it's a good thing to not treat people close to you callously.
Also, even if they agree on the obviously good things, the ethical philosophers go way beyond that. They argue for ethical claims that are not at all obvious--that most people would say are obviously wrong--and they do it based on abstract "rationalist" calculations even when those calculations fly in the face of the very ethical intuitions that lead most of us to agree that things like preventing malaria, treating parasite infections, and providing people with clean drinking water are obviously good things to do. But if you discard those ethical intuitions--and certainly someone like Parfit who is willing to treat those close to him callously, and do it intentionally, based on his ethical philosophy, is doing that--what do you have left? What basis is there for making any kind of ethical claim at all? At that point, whatever concept of "good" the philosopher is arguing for is just made up out of whole cloth, based on their personal ideology.
"Complex reasoning" without the guardrails of empirical tests and lived implementation is very error prone.
The number of variables is gigantic. In the armchair you can reduce them and tweak the remaining ones, and they will do whatever you like within the boundaries of your imagination.
That method seems just not very predictive of what will really happen.