Scott Sumner:

Unlike many on the left, I don’t envy the rich. I’m really happy that
Larry is really happy. If Larry was even a bit happier, and if that
boosted total global happiness, that would be fine with me. But I can’t
get away from the implication of (what I perceive as) diminishing
marginal utility. Some redistribution is justified.

Larry Ellison is extremely lucky to live in a universe where:

1. Government is very inefficient.
2. Incentives have a surprisingly strong effect on behavior.

If not for those practical limitations to redistribution, it would be
hard to argue against people who favor taking away almost all of
Ellison’s wealth, and reducing his consumption levels back to roughly
average.

I have two responses.

1. Sumner neglects important additional practical limitations to redistribution.  Most obviously:

a. Targets of coercion often resent the fact they’re being coerced.  So the relevant utilitarian calculation for forcible redistribution from A to B is not just “Will B enjoy A’s stuff more than A” but “Will B enjoy A’s stuff more than A plus the resentment cost.”

b. Due to the endowment effect, people who lose their stuff suffer more than equally wealthy people who gain the stuff.  So instead of looking at inequality alone, you need to put some weight on initial endowments!

2. All practical limitations aside, arguing against full redistribution is child’s play.  Just say: “Pure utilitarianism is not certainly true” and Sumner’s extreme (though hypothetical) conclusion collapses.  Suppose for example there’s a 99% chance utilitarianism is true, but a 1% chance that libertarian absolutism is true.  That 1% risk should at least slightly temper your redistribution.  And if, like Scott, you “know little about philosophy other than that Descartes said ‘I think therefore I am,'” isn’t substantial uncertainty about moral principles warranted?