> What's always missing to me in these types of discussions is what this value should be at, and what the distribution should look like qualitatively (i.e. what should the ideal Lorenz curve look like).

Just looking "more uniform" would be a great starting point. The optimal distribution is left as an open question for future research.

Not having the perfect solution upfront should not block incremental improvements, which is more or less what is happening ATM. E.g. people argue we cannot have free healthcare because the proverbial "welfare queen" will also receive it.