Yunsieg Kim and Jowei Chen have posted this draft on SSRN. Here is the abstract:

What are “traditional” districting criteria? The meaning of that term is critical to curbing abusive districting practices, because states must draw electoral districts according to those criteria to defeat claims of racial gerrymandering. Yet, the Supreme Court has never explicitly stated the qualities that make a districting criterion “traditional,” or given an exhaustive list of such criteria. Exploiting the Court’s failure to intelligibly define traditional redistricting criteria, conflicted interests are attempting to define that term in service of their private interests at the expense of the public’s. For example, legislatures pushing redistricting plans that would advantage a particular party or protect incumbents from competition claim that those districting objectives are “traditional”—and therefore must be judicially protected—by relying on anecdotal examples of a state having used them.

This Article proposes a definition of “traditional” districting criteria that would both curb such abuse and stay faithful to the commonly understood meaning of that word: widely accepted as standard practice. Under this alternative, which we call the empirical definition, a criterion is “traditional” only if a majority of the states require or allow it and fewer than a quarter prohibit it in state constitutions, statutes, or legislative guidelines. According to the empirical definition and our database of the redistricting laws of the fifty states, compactness, contiguity, equal population, and preserving county and city boundaries are traditional criteria. Among others, partisan advantage, incumbent protection, and preserving communities of interest are nontraditional criteria that should not be judicially protected.

In addition to the quantitative indicator, we argue that nontraditional districting criteria tend to be defined so as to be applied in a geographically inconsistent manner. Advantaging a certain party or protecting incumbents, by definition, requires disparate application, whereas preserving county boundaries applies regardless of whether a county votes Republican or Democratic. We also argue that preserving communities of interest is not traditional because it imposes a procedure instead of a substantively unambiguous districting principle, and because the term is so open-ended that it can be abused to justify nontraditional criteria such as partisan advantage or incumbency protection.

The empirical definition presents both legal and political advantages over the status quo. By presenting an objectively discernible definition of traditional criteria, the empirical definition would reduce the influence of undesirable judicial activism over election litigation. By excluding objectives such as partisan advantage or incumbency protection, it would also provide more constitutional protection from abusive districting.