After an earlier post on gerrymandering we were challenged by reader Greg to provide a more substantial piece on how one might fix gerrymandering.
We'll first note that while gerrymandering is of course problematic for fair representation ("The GOP scored 33 more seats in the House this election even though Democrats earned a million more votes in House races. Professor Jeremy Mayer says gerrymandering distorts democracy."), there's not evidence that it increases political polarization (which is ReConsider's core issue).
Gerrymandering happens because state legislatures can draw boundaries arbitrarily within some limitations about population size for each district. The original idea is that districts represent geographical groups with shared interests, and state legislators are somehow supposed to figure out who those groups are and then put them together. This is a tough task for even the most noble legislators.
In practice, of course, we sometimes get gerrymandering, where districts are tortured into bizarre shapes in order to increase the chances of a state sending more representatives from one party to Congress. North Carolina a the canonical example (Maryland is another pretty silly one).
Step one, then, of ending gerrymandering, is to remove the arbitrary power of legislators to draw lines however they please. Therefore, we need some objective principle or system that draws the lines, rather than people. We can bemoan gerrymandering all we want, but to stop it, we need an alternative.
What is such a system that we can agree on?
WaPo's Wonkblog put forth a map that seems tempting at first-glance. The lines of districts are automatically drawn by a computer algorithm (made by Brian Olsen) to "optimize compactness" (or generally minimize the perimeter of the districts), which would certainly eliminate the power to draw the ridiculous shapes that make up parts of NC and other states.
One question comes to mind: is compactness actually a good representation of these "areas of interest" by which one is supposed to draw these? Certainly it's better than the most absurdly gerrymandered districts, but that doesn't mean it's great.
The other big problem is minority representation: for the purposes of increasing representation of minorities, some states have concentrated minority groups into single districts. In fact, this is also mandated in some states by the Voting Rights Act.
But this also decreases their influence in surrounding districts, so a local minority group may end up with one pretty dedicated representative rather than many representatives. In a "maximum compactness" system, it may be the case that minorities are a very small group of almost all the districts, and therefore could become ultimately irrelevant in almost all races (like the "compact but unfair" case in the example below: imagine red = minority and blue = white).
This "compact but unfair" drawing may also represent a washing-out of political minorities. In Massachusetts, some one-quarter of voters are Republicans, but the state almost never sends a Republican as part of its 13-member delegation to Congress. (Blacks make up only 13% of the US population, and we can imagine that in a compact system they could also be largely washed out of representation).
Like most things in government, one probably requires some judgment here. Gerrymandering is bad ultimately because it's unfair. What does a "fair" districting look like? One might require some "independent" body to rule in cases of egregious perversion... but how does one create a truly politically-independent body?
We can't think of (or find out in the wild) a great solution. The takeaway here is that "someone should do something about X" is a fairly weak rallying call. We can complain about gerrymandering or some other practice, but without presenting an alternative, we're just having an enjoyable whine with our friends. In the case of gerrymandering, that solution is likely yet to come, but in other cases, the burden falls on the reformer to propose the new solution if they really want to see change.