There haven't been very good Robin Hood analogies made in the last several posts, but let's just assume that we stick to the point where we kill someone who sucks at using their money for societal good and give it to people save lives:

Now I'm a stone-cold utilitarian and really struggle understanding any other perspective on ethics, so I'll just say that buying a 1 million dollar condo instead of using that money to save 10+ lives in Africa seems obviously like a non-optimal decision on the societal scale. I don't know how to explain how it's non-optimal, it just seems obvious to me. Whether or not you can term this as "killing" 10+ people is just getting caught up in rhetoric, imo.

HOWEVER, where I don't go near as far as people like Peter Singer is that I don't think that all actions need to be optimal on the societal scale for you to be a decent person. Again, people get too caught up in rhetoric here and think that pointing it out to them that their use of money results in 10 less lives being saved than other uses of that money is calling them "murderers" and all that stuff. I can see an argument that says, "buying a condo is a non-optimal use of money on a societal scale, but it is possible to do things that are good for me and me only and for that to be okay, and this is okay." I feel that there's a line somewhere, but I have no idea where it is and it's not relevant to this convo anyway.

Now I admit that I was wrong when I said that the only difference between the pushing the button thing and the shooting a rich man in the face thing is the agency. This is wrong. The biggest difference for me is actually that pushing the button is an exception to the categorical imperative. On the other hand, killing anyone who uses their money non-optimally and using their money in a way that nets more than 1 life (after adjusting for life potential, etc.) gets VERY sticky for the categorical imperative.

This should be very easily understandable for poker players who understand meta-game and shania. Basically, pushing the button is an "in the vacuum" scenario because it's going to come up SO infrequently that deeming that it's okay for anyone who's ever in that scenario to take the money and use it to save lives is a fairly uncomplicated decision. On the other hand, when you're talking about more common actions, then you have to kind of form a range that optimizes overall outcome (simply killing anyone who in a vacuum saves lives leads to a society that's so brutish that you've actually made life shitty for everyone).

It's actually funny because that last part with the categorical imperative is like literally the introduction to my novel (which is about someone who uses eV calcs to justify being a vigilante who kills terrible lobbyists and shit).