I've pondered on link for a while, and have come to a curious conclusion:
If the Zipf distribution holds true, in no sense can a social justice predicated on inequity function.
Attempts to decrease inequity necessarily decrease total wealth, and decrease the well-being of the worst-off.
However, this may be predicated on the Zipf distribution of wealth being a derivative of a Zipf distribution of wealth-producing capacity. (Which is itself derivative of several related variables, each following Zipf distributions; loosely speaking, intelligence, strength, charisma, etc., although those characteristics are also overbroad; it also requires negative characteristics as much as positive ones, such as inability to show up on time) It also presumes a feedback cycle; that wealth-earning capacity is a multiplier on existing wealth. (This might seem to suggest that spreading the wealth around will increase the rate at which high-capacity individuals ramp up. It does, in fact, but the benefits of this are exceeded by the costs of removing wealth from known high capacity producers.)
The Zipf distribution -doesn't-, strictly, hold true, in particular at the lower tail, which is why I choose wealth production, rather than wealth, for my Zipf distribution; given a societal multiplier for wealth production, this produces the lower tails seen. There are a few possibilities to explain this outside the wealth production distribution - first world nations exporting the lower tails, minimum wage guaranteeing those below a certain threshold don't show up in the lower tails, socially acceptable standards of living skewing the averages up, etc. These alternative explanations have particular explanatory power if the Zipf distribution does, above minimum wage, hold true in compensation -per hour-, with the number of hours worked producing the skew in wealth. But I don't have evidence for that.