Reading about the SCOTUS arguments on Section 2 of the voting rights act got me to thinking about a comment someone made years ago, namely, that one could quickly draw fair congressional districts blindfolded. Now, I know that presently the rule is that political gerrymandering is just fine: only racial gerrymandering is bad (and, it seems, racial gerrymandering is maybe OK too as long as it is politically effective as well). However, while the holidays are getting close, it seems best to wish for as much as possible, in hopes of a miracle.
There is an obvious problem with voting districts. If they are deliberately constructed to create groups of like-minded people, you get gerrymandering that deprives minorities of all kinds of any meaningful voice. On the other hand, if you set out to make districts deliberately bland and homogenized, you also run the risk of electing representatives responsive to a majority or plurality of citizens. In either case, you risk majoritarian rule. The latter scenario will likely produce better results, because it is less likely to elect the sort of extremists that gerrymandering favors.
I am curious, though, what would happen if someone were to design a mathematical model that looked at nothing more than precise population data. Ignoring things like cities, towns, counties and streets, a computer could look at nothing more than dots on a map. I am not sure what system would work best. For example, depending on shape, a state with 6 districts might be divided once horizontally with a straight line, with equal populations north and south. Then each of those divisions could be divided into three with vertical lines (which would not line up with each other north and south). I am sure there is probably a better system than that example.
The question for demographers is whether this sort of blind segmentation would more often or not result in a reasonably fair distribution of votes? In Louisiana, for example, about ⅓ of the population is Black, and leans Democratic. The paler ⅔ population leans strongly Republican. Existing districting, now challenged, results in only one majority-Black/Democratic district, against five whiter/Republican districts. There is a threat that the state might go even further, and gerrymander away all Black votes, leaving a 6-0 distribution.
Since the distribution of voters is irregular across the state, I am wondering whether the geometrical districting suggested above might in fact have a decent chance of producing some majority-minority districts, without having any need to look at the underlying demographics during the districting process.
A possible side benefit from this approach might be to produce a closer mix of voters concerning a wider range of issues. Gerrymandering favors extreme positions, and as a result also limits the issues that come into play in elections. A better mix of voters could lead to politicians paying greater attention to a wider range of concerns.
On a realistic note, for any given state the manner of “blindly” dividing it geometrically could be chosen to favor a particular population, so the actual choice of a method of population-weighted geometric division would have to be dictated in some agnostic manner as well.
This is just a random thought that came to mind reading about the recent SCOTUS arguments — any suggestions?
